Conversations between the dead and the living (no candles)

My today’s editorial

I have been away from blogging for two days. I have been finishing that article about technological change seen from an evolutionary perspective, and I hope I have finished, at least as the raw manuscript. If you are interested, you can download it from  Research Gate or from my own website with Word Press. Now, as the paper is provisionally finished, I feel like having an intellectual stroll, possibly in the recent past. I am tempted to use those evolutionary patterns of thinking to something I had been quite busy with a few months ago, namely to the financial tools, including virtual currencies, as a means to develop new technologies. I had been particularly interested in the application of virtual currencies to the development of local power systems based on renewable energies, but in fact, I can apply the same frame of thinking to any technology, green energy or else. Besides, as I was testing various empirical models to represent evolutionary change in technologies, monetary variables frequently poked their head through some hole, usually as correlates to residuals.

So, I return to money. For those of my readers who would like to refresh their memory or simply get the drift of that past writing of mine, you can refer, for example, to ‘Exactly the money we assume’  or to  ‘Some insights into Ethereum whilst insulating against bullshit’, as well as to other posts I placed around that time. Now, I want to move on and meddle a bit with Bayesian statistics, and more exactly with the source method presented in the posthumous article by reverend Thomas Bayes (Bayes, Price 1763[1]), which, by the way, you can get from the JSTOR library via this link . I want to both wrap my mind around Thomas Bayes’s way of thinking, and refresh my own thinking about monetary systems. I have that strange preference to organize conversations between the dead and the living (no candles), so I feel like put reverend Bayes in conversation with Satoshi Nakamoto, the semi-mythical founding father of the Bitcoin movement, whose article, that you can download by this link, from my Word Press website, contains some mathematical analysis, based on the Poisson probability.

My initial question, the one I had been wrestling with this Spring, was the following: how can a local community develop a local system of green energy, and a local virtual currency, and how can these two help the development or the transformation of said local community? Why do I bother, posthumously, revered Thomas Bayes with this question? Well, because this is what he stated as the purpose of his article. In the general formulation of the problem, he wrote: ‘Given the number of times in which an unknown event has happened and failed: Required the chance that the probability of its happening in a single trial lies somewhere between any two degrees of probability than can be named’. The tricky part in this statement is the ‘unknown’ part. When we studied probabilities at high school (yes, some of us didn’t take a nap during those classes!), one of the first things we were taught to do was to define exactly the event that we want to assess the probability of happening. You remember? Read balls vs. black balls, in a closed box? Rings a bell? Well, Thomas Bayes stated a different problem: how to tackle the probability that something unknown happens? Kind of a red ball cross-bred with a black ball, with a hint of mésalliance with a white cube, in family records. In the last, concluding paragraph of his essay, Thomas Bayes wrote: ‘But what recommends the solution in this Essay is that it is complete in those cases where information is most wanted, and where Mr De Moivre’s solution of the inverse problem can give little or no direction, I mean, in all cases where either p or q are of no considerable magnitude. In other cases, or when both p and q are very considerable, it is not difficult to perceive the truth of what has been here demonstrated, or that there is reason to believe in general that the chances for the happening of an event are to the chances for its failure in the same ratio with that of p to q. But we shall be greatly deceived if we judge in this manner when either p or q are small. And though in such cases the Data are not sufficient to discover the exact probability of an event, yet it is very agreeable to be able to find the limits between which it is reasonable to think it must lie, and also to be able to determine the precise degree of assent which is due to any conclusions or assertions relating to them’.

Before I go further: in the original notation by Thomas Bayes, p and q are the respective numbers of successes and failures, and not probabilities. Especially if you are a native French speaker, you might have learnt, at school, p and q as probabilities, so be on your guard. You’d better always be on your guard, mind you. You never know where your feet can lead you. So, I am bothering late reverend Bayes because he was investigating the probability of scoring a relatively small number of successes in a relatively small number of trials. If you try to launch a new technology, locally, how many trials can you have? I mean, if your investors are patient, they can allow some trial and error, but in reasonable amounts. You also never know for sure what does the reasonable amount of trial and error mean for a given investor. You have the unknown event, see? Just as Thomas Bayes stated his problem. So I take my local community, I make a perfect plan, with a plan B possibly up our local sleeve, I take some risks, and then someone from the outside world wants to assess the odds that I succeed. The logic by Thomas Bayes can be a path to follow.

Satoshi Nakamoto, in that foundational article about the idea of the Bitcoin, treated mostly the issues of security. Still, he indirectly gives an interesting insight concerning the introduction of new inventions in an essentially hostile environment. When the simulates a cyberattack on a financial system, he uses the general framework of Poisson probability to assess the odds that an intruder from outside can take over a network of mutually interacting nodes. I am thinking about inverting his thinking, i.e. about treating the introduction of a new technology, especially in a local community, as an intrusion from outside. I could threat Nakamoto’s ‘honest nodes’ as the conservatives in the process, resisting novelty, and the blocks successfully attacked by the intruder would be the early adopters. Satoshi Nakamoto used the Poisson distribution to simulate that process and here he meets reverend Bayes, I mean, metaphorically. The Poisson distribution is frequently called as the ‘probability of rare events’, and uses the same general framework than the original Bayesian development: something takes place n times in total, in p cases that something is something we wish to happen (success), whilst in q cases it is utter s**t happening (failure), and we want to calculate the compound probability of having p successes and q failures in n trials. By the way, if you are interested in the original work by Simeon Denis Poisson, a creative French, who, technically being a mathematician, tried to be very nearly everything else, I am placing on my Word Press site two of his papers: the one published in 1827 and that of 1832 (presented for the first time in 1829).

And so I have that idea of developing a local power system, based on green energies, possibly backed with a local virtual currency, and I want to assess the odds of success.  Both the Bayesian thinking, and the Poisson’s one are sensitive to how we define, respectively, success and failure, and what amount of uncertainty we leave in this definition. In business, I can define my success in various metrics: size of the market covered with my sales, prices, capital accumulated, return on that capital etc. This is, precisely, the hurdle to jump when we pass from the practice of business to its theoretical appraisal: we need probabilities, and in order to have probabilities, we need some kind of event being defined, at least foggily. What’s a success, here? Let’s try the following: what I want is a local community entirely powered with locally generated, renewable energies, in a socially and financially sustainable manner.

‘Entirely powered’ means 100%. This one is simple. Then, I am entering the dark forest of assumptions. Let’s say that ‘socially sustainable’ means that every member of the local community should have that energy accessible within their purchasing power. ‘Financially sustainable’ is trickier: investors can be a lot fussier than ordinary folks, regarding what is a good deal and what isn’t. Still, I do not know, a priori, who those investors could possibly be, and so I take a metric, which leaves a lot of room for further interpretation, namely the rate of return on assets. I prefer the return on assets (ROA) to the rate of return on equity (ROE), because for the latter I would have to make some assumptions regarding the capital structure of the whole thing, and I want as weak a set of assumptions as possible. I assume that said rate of return on assets should be superior or equal to a benchmark value. By the way, weak assumptions in science are the exact opposite of weak assumptions in life. In life, weak assumptions mean I am probably wrong because I assumed too much. In science, weak assumptions are probably correct, because I assumed just a little, out of the whole expanse of what I could have assumed.

Right. Good. So what I have, are the following variables: local demand for energy D(E), local energy supply from renewable sources S(RE), price of renewable energy P(RE), purchasing power regarding energy PP(E), and rate of return on assets (ROA). With these, I form my conditions. Condition #1: the local use of energy is a local equilibrium between the total demand for energy and the supply of energy from renewable sources: Q(RE) = S(RE) = D(E). Condition #2: price of renewable energy is affordable, or: P(RE) ≤ PP(E). Condition #3: the rate of return on assets is greater than or equal to a benchmark value: ROA ≥ ROA*. That asterisk on the right side of that last condition is the usual symbol to show something we consider as peg value. Right, I use the asterisk in other types of elaborate expressions, like s*** or f***. The asterisk is the hell of a useful symbol, as you can see.

Now, I add that idea of local, virtual currency based on green energies. Back in the day, I used to call it ‘Wasun’, a play on words ‘water’ and ‘sun’. You can look up  ‘Smart grids and my personal variance’  or  ‘Les moulins de Wasun’ (in French) in order to catch a bit (again?) on my drift. I want a local, virtual currency being a significant part of the local monetary system. I define ‘significant part’ as an amount likely to alter the supply of credit, in established currency, in the local market. I use that old trick of the supply of credit being equal to the supply of money, and so being possible to symbolize with M. I assign the symbol ‘W’ to the local supply of the Wasun. I take two moments in time: the ‘before’, represented as T0, with T1 standing for the ‘after’. I make the condition #4: W/M(T1) > W/M(T0).

Wrapping it up, any particular event falling into:

Q(RE) = S(RE) = D(E)

P(RE) ≤ PP(E)

ROA ≥ ROA*

W/M(T1) > W/M(T0)

… is a success. Anything outside those triple brackets is a failure. Now, I can take three basic approaches in terms of probability. Thomas Bayes would assume a certain number n of trials, look for the probability of all the four conditions being met in one single trial, and then would ask me how many trials (p) I want to have successful, out of n. Simeon Denis Poisson would rather have taken an interval of time, and then would have tried to assess the probability of having all the four conditions met at least once in that interval of time. Satoshi Nakamoto would make up an even different strategy. He would assume that my project is just one of the many going on in parallel in that little universe, and would assume that other projects try to achieve their own conditions of success, similar to mine or different, as I try to do my thing. The next step would to be to define, whose success would be my failure, and then I would have to compute the probability of my success in the presence of those competing projects. Bloody complicated. I like it. I’m in.

[1] Mr. Bayes, and Mr Price. “An essay towards solving a problem in the doctrine of chances. by the late rev. mr. bayes, frs communicated by mr. price, in a letter to john canton, amfrs.” Philosophical Transactions (1683-1775) (1763): 370-418

Deux ans après la mort

Un petit éditorial de ma part

Je crois qu’hier, dans ma mise à jour en anglais ( “Lazy Sunday, watching the clouds” ), j’ai enfin cerné les hypothèses que je voudrais développer et prouver dans mon article sur l’application des modèles évolutionnistes à l’étude des changements technologiques. Les hypothèses, ce n’est pas facile à formuler, au moins si j’ai l’intention de les vérifier. Alors, je peux formuler trois hypothèses, que je trouve à la fois possibles à vérifier et liées à ce truc d’évolutionnisme. Hypothèse no. 1 : L’innovation contribue à réduire le déficit alimentaire. Hypothèse no. 2 : Le nombre des demandes de brevet est significativement déterminé par la quantité de facteurs de production – capital et travail – couramment engagés. Hypothèse no. 3 : des sociétés distinctes se caractérisent par une proportion distincte entre le nombre des demandes de brevet et la quantité des facteurs de production – capital et travail – couramment engagés.

En comparaison de mes réflexions passées, ces hypothèses peuvent sembler bien sèches, même simplistes. Eh bien, c’est le rasoir d’Ockham au boulot. J’ai revu les faits empiriques que je peux citer comme preuve de ce que j’avance, j’ai essayé de formuler une signification commune de ces faits et voilà ce que j’ai obtenu. Vous pouvez vous demander qu’est-ce que j’ai fait de toute cette réflexion évolutionniste. Alors, tout ce chemin de raisonnement était justement un chemin de raisonnement, une expression plus ou moins intelligible de mes intuitions. Dans le langage strict de Milton Friedman, que j’apprécie beaucoup par ailleurs, ce sont des hypothèses spéculatives. J’ai comme une petite intuition que le changement technologique observable dans l’économie mondiale est un processus évolutionniste. Si je décide de publier cette idée, à la fois les critiques et les enthousiastes de cette idée viendront tôt ou tard à ce moment du haussement des épaules : « Ouais, c’est chouette. Génial, même. Alors, qu’est-ce que ça prouve, exactement ? Si, par exemple, nous sommes au Kenya, quelle serait la différence dans leur politique d’investissement en technologies nouvelles, du point de vue de votre modèle ?». C’est précisément dans un désir de devancer une telle question que je me suis concentré sur les hypothèses énumérées ci-dessus. L’avantage réside dans l’ampleur du champ d’applications possibles. Si je présente une preuve convaincante de ces hypothèses, et si j’équipe ma preuve empirique avec une version élégante du raisonnement que vous avez pu suivre sur mon blog durant ces dernières semaines, je démontrerai que le raisonnement évolutionniste est apte à générer un outil de prédiction, qui peut aider à comprendre comment l’innovation peut aider à sortir de la pauvreté.

Comme je suis cette piste de distinction entre la spéculation intellectuelle et les preuves empiriques, je suis retourné à un classique : le révérend Thomas Bayes et son essai posthume sur le calcul de probabilité (Bayes, Price 1763[1]). Voilà une histoire intéressante, lourde en conséquences. En Décembre 1763, monsieur Richard Price adresse une lettre à John Canton qui était alors, selon toute vraisemblance, le rédacteur en chef ou le rédacteur adjoint d’une revue prestigieuse intitulée « Philosophical Transactions of the Royal Society ». Dans cette lettre, Richard Price communique qu’en classant les notes de son ami Thomas Bayes, après la mort de celui-ci, il eut trouvé un essai extrêmement intéressant. Richard Price en lui-même était un personnage extrêmement intéressant (lisez plus, par exemple, ici : https://www.york.ac.uk/depts/maths/histstat/price.pdf ) et il était aussi intéressant de constater qu’il a soumis cet essai de Thomas Bayes à la Royal Society deux ans après la mort de son ami. Qu’était-il de si important dans cette œuvre ? Essayons de reconstruire le chemin de raisonnement qui a donné naissance à ce qu’on appelle aujourd’hui la statistique Bayésienne.

Je procède donc de ma manière préférée et je saute jusqu’à la fin dudit essai. Voilà ce qu’écrit Thomas Bayes dans la conclusion : « Ce qui recommande le plus la solution contenue dans cet essai est qu’elle est complète dans ces cas où l’information est la plus voulue et où la solution de Mr de Moivre du problème inverse peut donner peu ou pas de direction du tout ; je veux dire dans tous les cas où p ainsi que q n’ont pas de magnitude considérable. Dans d’autres cas, lorsque p ainsi que q sont très considérables, il n’est pas difficile de percevoir la véracité de ce qui a été démontré, donc qu’il y a des raisons de croire en général que les chances de l’occurrence d’un évènement sont aux chances de sa défaillance dans le même ratio que celui de p au q. Néanmoins nous serons grandement dupes si nous jugeons de cette manière lorsque p ou q sont petits. Et ainsi dans de tels cas les Données ne sont pas suffisantes pour découvrir la probabilité exacte d’un évènement, quoi qu’il est plausiblement possible de découvrir les limites entre lesquelles il est raisonnable de penser qu’il se trouve, ainsi qu’il est possible de déterminer le degré d’assentiment dû à toute conclusion ou assertion relative à ces limites ».

 Voilà donc que le révérend Thomas Bayes expose sa manière d’explorer les évènements à occurrence peu fréquente. Voulait-il établir une preuve éclairée de l’existence de du Dieu ? Possible. Voyons donc comment il s’y prend. Je passe directement au contenu de l’essai lui-même et je n’entre pas, pour le moment, dans les détails de la préface écrite par Richard Price dans sa lettre. Juste une remarque en passant. A l’école, on a été habitués à voir p et q comme des probabilités. Ici, dans la notation originale de Thomas Bayes, ce sont des nombres d’essais, pas des probas. Nous parlons donc des cas, ou le nombre d’essais est tellement faible qu’il est dur de calculer les probabilités classiques, comme P/N.

Thomas Bayes pose un problème simple : « Etant donné le nombre des fois quand un évènement inconnu s’est passé ou a failli de se passer ; ayant comme requis que la probabilité de son occurrence dans un essai unique se trouve quelque part parmi deux degrés de probabilité qui peuvent être nommés ». Ce qui intrigue tout de suite dans ce problème est la notion d’évènement inconnu. Normalement, dans le calcul de probabilité, un pas prérequis est de définir exactement les évènements observés. Ici, Bayes pose l’hypothèse d’un évènement que nous ne pouvons pas définir. Après, ça commence mollo : presque toute la Section I sonne exactement comme le contenu standard des manuels de maths aujourd’hui. Ça commence à être vraiment intéressant avec Proposition no. 2 dans cette section : « Si une personne a une espérance qui dépend sur l’occurrence d’un évènement, la probabilité de l’évènement est à la probabilité de sa défaillance comme sa perte en cas de défaillance à son gain en cas de l’occurrence ». Voilà que Thomas Bayes annonce sa couleur pour la première fois : dans la vie réelle, nous n’avons pas l’occasion, d’habitude, de calculer les probabilités de succès ou de perte. Nous avons des informations générales et catégoriques du genre : « si votre ceinture de sécurité n’est pas bouclée, la chance que vous soyez atteint par une météorite est X ». On veut savoir comment utiliser ces règles générales dans la vie de tous les jours et Thomas Bayes conseille : établissez des scénarios du genre « si X alors Y » et puis calculez les probabilités conditionnelles.

Exemple : je choisis entre plusieurs investissements alternatifs en des technologies distinctes. Je sais que je n’aurai pas l’occasion de tester chaque technologie l’une après l’autre et que personne ne peut me garantir le succès avec un choix donné. Néanmoins je veux pondérer mes risques. Thomas Bayes dit : d’abord, définissez ce qu’est un succès pour vous dans cette situation. Pour chaque technologie alternative, faites une estimation raisonnable de remplir ces critères de succès. Après ce pas initial, vous aurez donc une probabilité de succès égale à « a » et une probabilité d’échec égale à « b ». Remarquez : échec ne veut pas nécessairement dire faillite complète. Là, j’ai un peu de flexibilité, suivant mes critères de succès. Maintenant, je construis in portefeuille de sept compagnies. Je suis préparé à en sacrifier deux sur l’autel de mon expérience en affaires, mais cinq d’entre eux doivent marcher, suivant ma définition préalable de ce que « marcher » veut dire pour moi. Je veux donc savoir quelle est la probabilité cumulative que 5 investissements sur 7 soient un succès et que 2 sur 7 soient des échecs.

Thomas Bayes postule que ma probabilité cumulative sera égale à E*ap*bq = E*a5*b2. Cela veut dire, dans le raisonnement original de Bayes, que mes 5 succès et 2 échecs peuvent survenir d’E façons différentes et mutuellement incohérentes. Il est utile de se souvenir, à ce point-ci, que les probabilités sont, par définition, des fractions plus petites que 1. Plus grande est l’exposante à laquelle j’élève une telle fraction, plus petite sera la puissance obtenue. Donc, même si mes chances de succès soient égales à celles de l’échec, donc si a = b = 0,5, 0,5 puissance 5 sera plus petit que 0,5 puissance 2. Plus je veux de succès sur un nombre total d’essais, plus petite sera la probabilité de les obtenir, et c’est alors que le facteur E prend de l’importance.  Plus de combinaisons différentes ai-je d’avoir 5 succès et 2 échecs sur un total de 7 investissements, plus grandes sont mes chances d’atteindre mes objectifs.

La théorie que Thomas Bayes exposa dans son essai est vraiment complexe. Là, je ne viens que survoler ses hypothèses initiales. La leçon à tirer est très proche de ce que je suis en train de développer dans mes modèles évolutionnistes : la flexibilité et le choix ont une importance primordiale pour l’issue finale d’un ensemble complexe d’actions entreprises par une société humaine. Nous pouvons dire que les probabilités simples, comme « a » et « b » dans l’exemple de Bayes, sont données par la Nature : elles sont largement exogènes. Largement, mais pas complètement. Beaucoup dépend de la façon de définir mon succès et mon échec. En revanche, le nombre requis de succès, le nombre tolérable d’échecs, ainsi que le nombre de façons possibles de les combiner sont principalement des décisions de ma part. Dans cet aspect particulier, la théorie de Bayes montre que la possibilité de multiplier les formes différentes de faire la même chose accroît mes chances de succès d’une façon capitale. C’est alors que le raisonnement évolutionniste revient en scène : l’évolution c’est une série d’expériences avec une série de mutations. Plus on invente de mutations pour une fonction donnée, plus on a de chances de survivre.

J’ai fait tout ce détour par l’essai fameux de Bayes pour montrer que l’approche évolutionniste n’est pas la même chose que le Darwinisme originel. L’évolutionnisme est une façon d’appréhender tout processus de changement qui implique une série de choix complexes.

[1] Mr. Bayes, and Mr Price. “An essay towards solving a problem in the doctrine of chances. by the late rev. mr. bayes, frs communicated by mr. price, in a letter to john canton, amfrs.” Philosophical Transactions (1683-1775) (1763): 370-418

Lazy Sunday, watching the clouds

Editorial

This is one of those days when I experience radically contradictory intuitions as for my research work. One voice in my head is telling me: ‘Stay focused. You have a nice research path here, with those evolutionary models of technological change’. Still, there is another voice, who is currently watching the clouds, as they rush through the late summer sky, and is wondering what the hell is it all about, you know, universe and stuff. By the way, can a voice watch clouds? Well, basically it can, if it has eyes, and some brain behind. What I need seems to be that special kind of broad picture. You know, that kind you can come by in some social relations. Somebody frames you into some lamentable deal, you say it was really bitchy from their part, and they say something like: ‘Yes, but we should see the broad picture’. Interestingly, said broad picture is focused on providing good excuses to that person. Still there is that rhetorical technique of focused broad picture, kind of precise and kind of overarching in the same time. This is what I need, to reconcile those two voices in my head.

In a picture, I’d better sketch before putting any thick paint in it. I start sketching with defining corners and frames first, and then I create a structure inside those frames. This is, at least, what I retained from my Art classes at school. So I sketch. Corner #1: Herbert W. Simon, reputed to be the patient zero of evolutionary approach in economics, and his article of 1955[1], treating of bound rationality in economic decisions. Corner #2: Arnold Toynbee, and his metaphor of civilisations seen as people struggling to get somewhere up from a mountain ridge, contained in his monumental work entitled ‘Study of History’, first published in 1939, then abridged, during World War II,  by an enthusiastic follower, David Churchill Somervell, and published abridged in 1946[2]. I mean more specifically the content to be found on pages 68 – 69 of this abridged version. Corner #3: a recent discovery in evolutionary biology – made and disseminated by professor Adam Hargreaves from the University of Oxford – that besides the known mechanisms of evolution, namely spontaneous mutation and natural selection, there is a third one, some kind of super-fast mutation in some genes, which works takes place so bloody quickly that those genes seem to disappear from the genome as we can sequence it . Corner #4: my own research, summing up, so far, to saying that our global achievement regarding technological change, is rather in ameliorating life conditions, for example in alleviating food deficit, rather than in maximizing Total Factor Productivity.

Fine, as I look critically at those four corners, I would add some more, but a frame with more than four corners becomes a bit awkward for sketching anything inside. I wanted a picture, I have a picture. Format is format, period. I draw a first edge, from corner #1 to corner #2, from Herbert Simon to Arnold Toynbee. The edge turns out to be somehow symbolic: professor Toynbee retired from scientific career in 1955, exactly the same year when Herbert W. Simon published that article I have in mind. Herbert Simon says: we can be biased, in our choices, as for very nearly everything. The range of options we can really choose between, their possible outcomes and payoffs, as well as exogenous conditions: we can subjectively distort all of that. Arnold Toynbee says: social change is a struggle with highly uncertain outcomes, and these outcomes are observable just sometimes, as pit stops reached in an endless race. Many a human civilisation failed in assuring continuous development. This edge, connecting Herbert Simon to Arnold Toynbee, is a question: how can we climb the cliff of history more efficiently, knowing that every hold is burdened with cognitive bias? Now, I connect corner #2 (professor Toynbee), with #3, the recent discovery of super-fast genetic mutation. Once again, a question arises on that edge. What happened if, in our civilisation, our cultural success depended on something that changes so fast we can’t even say how it is happening nor how is it subject to natural selection? Next edge, from that discovery by professor Hargreaves to my own research. This time, the edge question comes to my mind quite naturally. What if the technological change that we can observe, I mean invention, obsolescence in established technologies, production function, what if all that was a sort of blanket cover for some other process of change, taking place kind of underneath? What would be that process?

Finally, the fourth edge of my canvas, from my own research back to Herbert Simon and his theory of cognitive bias in economic decisions. We know that collective intelligence, understood as learning by experimentation and interaction, can reduce the overall impact of individual cognitive biases. Does the current proportion between the input of production factors (i.e. capital and labour), and the output of technological change (patentable invention, obsolescence of established technologies) reflect some kind of local equilibrium, a production function of technological change? How is that hypothetical function of technological change specific to precise social conditions, and how can it contribute to changing those conditions?

Thus, as I walk back in my footsteps, just to check if I haven’t trodden on something interesting, I reconstitute the edges of my canvas, and I try to define some kind of central point and the intersection of diagonals. What I am looking for is a model (theory?) of technological change, embedded in broader social change, which could help in discovering some possibly unexplained characteristics of our modern civilisation, and possibly assist future social change. Ambitious. Possibly impossible to achieve with my intellectual resources. Cool. I’m in, and now, I am restating the sparse hypotheses that my internal trio – the ape, the monk, and the bulldog – has hatched over the last weeks.  Hypothesis #1: innovation helps people out of hunger. Hypothesis #2: the number of resident patent applications per year, in a given country, significantly depends on the amount of production factors available. Hypothesis #3: evolutionary selection of new technologies works as an interaction between a set of female organisms disposing of production factors, and a set of male organisms generating intelligible blueprints of new technologies. Hypothesis #4: different social structures yield different selection functions, with different equilibriums between the number of patent applications and the input of production factors, capital and labour.

Good. Now, I check if all that intellectual diarrhoea makes a coherent logical structure. I start with bringing a little correction in the last hypothesis: I replace ‘equilibriums’ with ‘proportions’. In economics, an equilibrium is supposed to be something really cool, kind of serenissime; it is supposed to be a state of at least locally optimal use of resources. I cannot prove that any proportion between the number of patent applications and the input of production factors is such a state. I can observe that this proportion is somehow specific to distinct social structures, but I cannot see any way (nor any willingness, by the way) to prove that it is an optimal state for these structures. Now, provability. I can empirically check #1, #2, and #4, but not #3 (at least I cannot see how I could check it with the data I have). The #3 seems to be a speculative hypothesis, kind of a cherry I can put on the top of a cake, but I have to bake the cake first.

I have that logical construct, made of hypotheses #1, #2, and #4, which I place in the centre of my canvas. If I was painting a landscape, it would be that beautiful [lake, mountain, horse, river, sunset, or put here whatever you want] to be presented between those four corners and four edges. Now, following the logic by Milton Friedman, what I will be doing about those hypotheses would be not so much proving them true, as absolute truth does not exist in a probabilistic world, but finding conditions for not refuting those hypotheses as false. Supposing I have proven that (I kind of have, in my earlier posts), I can now try to connect the proof to edge questions, i.e. I can try to build a speculative reasoning. Tomorrow.

[1] Simon A.,H., A Behavioral Model of Rational Choice, The Quarterly Journal of Economics, Vol. 69, No. 1 (Feb., 1955), pp. 99-118

[2] Toynbee, J. Arnold. Study of history. University press, 1946.

Le patient zéro

Mon éditorial

J’ai trouvé quelques trucs intéressants dans la littérature sur les modèles évolutionnistes en sciences économiques. Andrew W. Lo, en 2005[1], a publié cet article que j’ai déjà commencé à discuter hier, dans ma mise à jour en anglais (regardez “Equilibrium in high esteem” ), où il développe la soi-disant hypothèse de marchés adaptatifs, ou Adaptive Markets Hypothesis (AMH) en anglais. Le truc intéressant en ce qui concerne cette hypothèse, et je ne m’en suis rendu compte que tard dans la soirée d’hier, est qu’elle ressemble beaucoup à ce que Thomas Malthus avait avancé, il y a plus de deux siècles, dans son « Essai sur le principe de la population » (1798 – 1998[2]). Thomas Malthus n’avait pas la moindre idée de la théorie de l’évolution – il était un homme d’église et il vivait presque un siècle avant Darwin. Néanmoins, il est arrivé à une vision très similaire à celle d’Andrew Lo : une population accroît son nombre en présence de ressources qui ne croissent pas à une cadence aussi rapide (ou dont la quantité d’accroît pas du tout) ; chaque génération fait donc face à une compétition de plus en plus intense et à une possibilité décroissante d’approprier les ressources en question ; à un certain moment, la situation devient critique, la population est décimée, son nombre décroit rapidement et la quantité de ressources accessible en moyenne par un membre de ladite population revient à un niveau plus ou moins confortable.

Ensuite, ça recommence en cycle. On peut tracer des scenarios de catastrophe où ce n’est pas seulement la population qui soit décimée, mais aussi son environnement de ressources qui croule. C’était la vision de Thomas Malthus et Andrew W. Lo l’a essentiellement reproduite, tout en la testant comme une hypothèse de changement dans un marché, avec les profits agrégés, possibles à gagner dans un marché donné, jouant le rôle de ressources sujets à la compétition. Remarquez : je n’ai pas la moindre idée si Andrew Lo savait qu’il reproduit le concept de « population check » formulé par Thomas Malthus. En tout cas, il ne le cite pas comme référence théorique.

En parcourant la littérature économique sous l’angle d’approche évolutionniste, j’ai trouvé un auteur, Herbert A. Simon, et son article de 1955, qui semble être le patient zéro de tout ce courant de recherche (Simon 1955[3]). J’ai retrouvé une copie de cet article. Herbert Simon semble avoir été orienté sur la théorie de choix économiques plus que sur la théorie évolutionniste strictement dite. Il pose son problème initial d’une façon intéressante : on sait que nos choix économiques sont loin d’être aussi parfaitement pondérés que l’assume l’école classique en sciences économiques. Simon démontre brièvement que dans la grande majorité de situations le décideur humain ne peut pleinement appliquer aucuns des modèles principaux de choix rationnel : il ne peut ni maximaliser, ni former une certitude, ni même formuler des probabilités plausibles. Dans nos décisions économiques on fait tout comme si on appliquait ces procédures, mais c’est juste tout comme, pas du vrai choix rationnel. Vous allez à Shanghai, vous visitez un marché aux puces très spécial placé dans un souterrain spacieux en-dessous du Musée de L’Art, et vous pouvez avoir une expérience directe de toutes les nuances subtiles entre, par exemple, un vrai sac Vuitton et un « tout comme ». Ces marchands-là, ils sont des fins théoriciens, croyez-moi.

Nous sommes imparfaitement rationnels dans nos décisions, ne serait-ce qu’en raison d’information incomplète. C’est un fait aisément vérifiable par expérience. Néanmoins, la psychologie sait très peu sur le mécanisme exact de ces choix imparfaitement rationnels. Nous faisons donc face à un paradoxe : nous savons qu’il faut incorporer la rationalité incomplète dans les modèles économiques mais nous savons très peu sur le « comment » de cette rationalité incomplète. L’article de Herbert Simon représente le type d’exercice intellectuel très proche à mon cœur : c’est le rasoir d’Ockham en action. Après avoir coupé et séparé la connerie, avec l’aide dudit rasoir, Herbert A. Simon vient à la conclusion que quoi qu’on fasse, on reproduit, dans notre tête, le mécanisme de choix rationnels dans la version « tout comme ». Si in individu a donc un répertoire A des comportements possibles, son choix véritable prend lieu dans un ensemble Ar, qui est une représentation imparfaite de A. Par analogie, en présence d’un ensemble réel S des situations futures possibles, le décideur produit dans son esprit un ensemble Sr qui, encore une fois, est une représentation imparfaite de S. De même, les récompenses futures associées avec les comportements à choisir et leurs implications futures produisent une copie imparfaite d’elles-mêmes dans l’esprit du décideur.

La conclusion qui vient tout droit de ce raisonnement, mais qu’Herbert Simon formule très, très prudemment est qu’en présence d’une réalité donnée, dans un marché ou dans un système social complexe, le chemin pris par les décisions économiques est conditionné en premier lieu par ces représentations de choix réels et non pas par la réalité elle-même. Information accessible au sujet de l’environnement est aussi importante que l’environnement lui-même. Par exemple, la différence entre A et Ar (donc entre les alternatives réelles d’une part et ce qui nous semble être le répertoire de nos alternatives) peut avoir un impact beaucoup plus significatif sur un marché que l’ensemble A en tant que tel. Imaginons une situation, où un grand nombre d’entrepreneurs espèrent, à tort, qu’une nouvelle technologie est possible à mettre en marche en moins d’un an. Ils sont donc convaincus, à tort, que la stratégie du type « on démarre en janvier et on a des premières pièces vendues en novembre » est une alternative accessible de comportement. Ça ne marche pas, bien évidemment, parce que ça n’avait aucune chance de marcher. La fausse conclusion que les entrepreneurs peuvent en tirer est que cette technologie en tant que telle vaut rien – et non pas que c’était le timing irréaliste qui ne valait rien – et cette fausse conclusion influencera le choix futur entre des technologies possibles.

L’article de Herbert Simon semble être une passerelle entre la théorie des jeux et l’évolutionnisme. En fait, il me semble être une sorte de discussion avec toute la série d’articles que John Nash a publié dans les années 1950 et qui lui ont valu le prix Nobel en économie en 1994 (Nash 1950a[4] ; Nash 1950b[5] ; Nash 1951[6] ; Nash 1953[7]). Dans sa théorie d’équilibre dynamique dans un jeu, John Nash utilisait la notion de « stratégie dominante » ou stratégie attachée à la plus haute récompense entre toutes les récompenses possibles. Nash postulait que l’équilibre dynamique dans un jeu peut émerger seulement sur la base des stratégies dominantes de la part de tous les joueurs. Seulement voilà, il a fait comme des esprits vraiment fins le font souvent : il a formulé une idée intéressante mais ambigüe et il a laissé les autres se démordre avec les ambigüités. Doit-ce être une stratégie dominante objectivement ou suffit-il, pour former un équilibre dynamique, que les joueurs pensent qu’ils ont des stratégies dominantes ? Si quatre joueurs sur cinq ont leurs stratégies dominantes et le cinquième pas vraiment, ce cinquième participera-t-il à l’équilibre dynamique dans ce jeu ou pas ? C’est juste un échantillon de questions que John Nash a élégamment laissé sans réponses et c’est probablement la raison pour ce prix Nobel qu’il a finalement obtenu. Il a obligé un sacré monde à penser dans les lignes de sa théorie et c’est l’un des critères principaux pour le prix Nobel. Par ailleurs, Herbert A. Simon a eu son prix Nobel à lui en 1978, seize ans avant John Nash.

Si je considère Herbert Simon comme le patient zéro de l’approche évolutionniste, c’est donc la piste générale de la rationalité limitée à laquelle je dois attacher ma propre recherche. Mon modèle de changement technologique comme adaptation intelligente par sélection dans une population sexuée devrait être donc généralisé comme un cas de rationalité limitée et de ses conséquences macroéconomiques. Bon, pigé.

[1] Lo, A.,W., 2005, Reconciling Efficient Markets with Behavioral Finance: The Adaptive Markets Hypothesis, The Journal of Investment Consulting, Volume 7, no. 2, pp. 1 – 24

[2] Malthus Thomas, 1798, An Essay on the Principle of Population. An Essay on the Principle of Population, as it Affects the Future Improvement of Society with Remarks on the Speculations of Mr. Godwin, M. Condorcet, and Other Writers, London

Printed for J. Johnson, in St. Paul’s Church-Yard, 1998, Electronic Scholarly Publishing Project, http://www.esp.org

[3] Simon A.,H., A Behavioral Model of Rational Choice, The Quarterly Journal of Economics, Vol. 69, No. 1 (Feb., 1955), pp. 99-118

[4] Nash, J.F., 1950, Equilibrium Points in n – Person Games – Proceedings of the National Academy of Sciences of the United States of America, vol. 36, no.1, pp. 48 – 49

[5] Nash, J.F., 1950, The Bargaining Problem, Econometrica, vol. 18, no.2, pp. 155 – 162

[6] Nash, J.F., 1951, Non – Cooperative Games, The Annals of Mathematics, Second Series, vol. 54, issue 2, pp. 286 – 295

[7] Nash, J.F., 1953, Two – Person Cooperative Games – Econometrica, vol. 21, issue 1, pp. 128 – 140

Equilibrium in high esteem

My editorial

I am still at this point of defining my point in that article I am supposed to hatch on the topic of evolutionary modelling in studying technological change. Yes, it takes some time and some work to define my point but, man, that’s life. I think I know things, and then I think how to say what I know about things, and it brings me to thinking once again what is it that I know. If, hopefully, I come to any interesting conclusions about what I know, I start reading literature and I discover that other people know things, too, and so I start figuring out what’s so original in what I know and how to say it. You know those scenes from Jane-Austin-style movies, where people are gossiping in a party and they try to outgossip each other, just to have that momentary feeling of being the most popular gossiper in the ballroom? Well, this is the world of scientific publications. This is what I do for a living, very largely. I am lucky, mind you. I don’t have to wear one of those white wigs with a fake tress. This is a clear sign that mankind is going some interesting way forward.

Yesterday, as I was gossiping in French (see ‘Deux lions de montagne, un bison mort et moi’ ), I came to some conclusions about my point. I think I can demonstrate that the pace and intensity of technological change we have been experiencing for the last six or seven decades can be explained as a function of intelligent adaptation, in the human civilisation, to a growing population in the presence of scarce food. This is slightly different an angle of approach from those evolutionary models I have been presenting on my blog over the last few weeks, but what do you want: blog is blog, and scientific gossip is scientific gossip. This deep ontological distinction means I have to adapt my message to my audience and to my medium of communication. Anyway, why this? Well, because as I turned and returned all the data I have about technological change, I found only one absolutely unequivocal gain in all that stuff: between 1992 and 2016, the human population on the planet has doubled, but the average food deficit per person per day has been cut by half, period. This is it. Of course, other variables follow, of similar nature: longer life expectancy, better access to healthcare and sanitation etc. Still, the bottom line remains the same: technological change occurs at intensifying a pace, it costs more and more money, and it is correlated with improvements in the living conditions much more than with increased Total Factor Productivity.

There is a clan of evolutionary models, which, when prodded with the stick labelled ‘adaptation’, automatically reply with a question: ‘Adaptation to what?’. Wrong question, clan. Really. You, clan, you have to turn your kilts over, to the other side, and see that other tartan pattern. Adaptation is adaptation to anything. Otherwise, if we select just some stressors and say we want to adapt to those, it becomes optimization, not adaptation, not anymore. The right question is ‘How do we adapt?’. Oh, yes, at this point of stating my point I suddenly remember I have to do some review of literature. So I jump onto the first piece of writing about intelligent adaptation I can find. My victim’s name is Andrew W. Lo and his article about adaptive markets hypothesis (2005[1]).  Andrew W. Lo starts from the biological assumption that individuals are organisms, which, through generations of natural selection form so as to maximize the survival of their genetic material.

Moreover, Andrew Lo states that natural selection operates not only upon genetic material as such, but also upon functions this genetic heritage performs. It means that even if a genetic line gets successfully reproduced over many generations, so if it kind of goes intact and immutable through consecutive generational turns, the functions it performs can change through natural selection. In a given set of external conditions, a Borgia (ducal bloodline) with inclinations to uncontrolled homicide can get pushed off to the margin of the dynasty by a Borgia (ducal bloodline) with inclinations to peaceful manipulation and spying. If external conditions change, the vector of pushing off can change, and the peaceful sneaky kind may be replaced by the violent beast. At the end of the day, and this is a very important statement from Andrew W. Lo, social behaviour and cultural norms are also subject to natural selection. The question ‘how?’, according to Andrew Lo, is being answered mainly as ‘through trial and error’ (which is very much my own point, too). In other words, the patterns of satisfactory behaviour are being determined by experimentation, not analytically.

I found an interesting passage to quote in this article: ‘Individuals make choices based on experience and their best guesses as to what might be optimal, and they learn by receiving positive or negative reinforcement from the outcomes. If they receive no such reinforcement, they do not learn. In this fashion, individuals develop heuristics to solve various economic challenges, and as long as those challenges remain stable, the heuristics eventually will adapt to yield approximately optimal solutions’. From that, Andrew Lo derives a general thesis, which he calls ‘Adaptive Markets Hypothesis’ or AMH, which opposes the Efficient Market Hypothesis (EMH). The way it works in practice is being derived by close analogy to biology. Andrew Lo makes a parallel between the aggregate opportunities of making profit in a given market and the amount of resources available in an ecosystem: the more resources are there, the less fierce is the competition to get a grab of them. If the balance tilts unfavourably, between the population and the resources, competition becomes more ruthless, but ultimately the population gets checked at its base, and declines. Declining population makes competition milder, and the cycle either loops in a band of predictable variance, or it goes towards a corner solution, i.e. a disaster.

The economic analogy to that basic biological framework is that – according to AMH and contrarily to EMH – ‘convergence to economic equilibrium is neither guaranteed nor likely to occur at any point in time’. Andrew Lo states that economic equilibrium is rather a special case than a general one, and that any economic system can either converge towards equilibrium or loop in a cycle of adaptation, depending on the fundamental balance between resources and population. Interestingly, Andrew Lo manages to supply convincing empirical evidence to support that claim, when he assumes that profit opportunities in a market are the economic equivalent of food supplies in an ecosystem.

I find that line of thinking in Andrew Lo really interesting, and my own research, that you could have been following over the last weeks on this blog, aims at pinning down the ‘how?’ of natural selection. The concept is being used frequently: ‘The fittest survive; that’s natural selection!’. We know that, don’t we? Still, as I have that inquisitive ape inside of me, and as that ape is being backed by an austere monk equipped with the Ockham’s razor, questions abound. Natural selection? Splendid! Who selects and how? What do you mean by what do I mean by ‘who selects?’? (two question marks in one sentence is something I have never achieved before, by the way). Well, if we say ‘selection’, it is a choice. You throw a stone in the air and you let it fall on the ground, and you watch where it falls exactly. Has there been any selection? No, this is physics. Selection is a human concept and means choice. Thus, when we state something like ‘natural selection’, I temporarily leave aside the ‘natural’ part (can there be unnatural selection?) and I focus on the act of selecting, or picking up from a lot. Natural selection means that there is a lot of items, produced naturally, through biology (both the lot in its entirety and each item separately), and then an entity comes and chooses one item from the lot, and the choice has consequences regarding biological reproduction.

In other words, as long as we see that ‘natural selection’ as performed by some higher force (Mother Nature?), we are doing metaphysics. We are pumping smoke up our ass. Selection means someone choosing. This is why in my personal research I am looking for some really basic forms of selection with biological consequences. Sexual selection seems to fit the bill. Thus, when Andrew Lo states that natural selection creates some kind of economic cycle, and possibly makes the concept of economic equilibrium irrelevant, I intuitively try to identify those two types of organisms in the population – male and female – as well as a selection function between them. That could be the value I can add, with my model, to the research presented by Andrew Lo. Still, I would have to argue with him about the notion of economic equilibrium. He seems to discard it almost entirely, whilst I hold it in high esteem. I think that if we want to go biological and evolutionist in economics, the concept of equilibrium is really that elfish biscuit we should take with us on the journey. Equilibrium is deeply biological, and even physical. Sometimes, nature is in balance. This is more or less stationary a state. An atom is an equilibrium between forces. An ecosystem is an equilibrium between species and resources. Yes, equilibrium is something we more frequently crave for rather than have, and still it is a precious benchmark for modelling what we want and what kind of s*** we can possibly encounter on the way.

[1] Lo, A.,W., 2005, Reconciling Efficient Markets with Behavioral Finance: The Adaptive Markets Hypothesis, The Journal of Investment Consulting, Volume 7, no. 2, pp. 1 – 24

Deux lions de montagne, un bison mort et moi

Mon éditorial

Je suis en train de bâtir les fondements théoriques de mon article sur l’approche évolutionniste du changement technologique. Dans tout article scientifique bien formé point de vue style, au moins en sciences économiques, il y a ce passage où l’auteur explique pourquoi il a pris le chemin qu’il ait pris. Cette explication bifurque, en fait, en deux questions distinctes : pourquoi est-ce le sujet important et qu’est-ce qui donne de la pertinence à la méthode employée par l’auteur ? Ici, quelques mots d’explication sur les finesses d’écriture scientifique en sciences économiques. Nous, les économistes, un peu comme les juristes, on assume, en général, que ça dépend. Toute question peut être approchée sous des angles différents : classique, néoclassique, institutionnel ancien, institutionnel nouveau, keynésien, postkeynésien, monétariste, théorie des jeux, évolutionniste, behavioriste, NCM etc. en encore, j’en passe quant aux mésalliances. L’exercice de style, dans une publication en sciences économiques, consiste très largement à annoncer quelle approche méthodologique ou bien quelle combinaison d’approches ai-je choisi pour la recherche empirique présentée dans l’article, ainsi qu’à expliquer pourquoi juge-je cette approche pertinente.

D’abord, donc, pourquoi est-ce que tout ce truc de changement technologique est-il important ? Ben voilà : on assume que nous formons une civilisation technologique avancée. Comme nous ne connaissons vraiment aucune autre civilisation technologique – les dernières tentatives de contact effectuées par Thor n’étaient pas vraiment prometteuses point de vue échange culturel – il est difficile de dire avec certitude comment avancés sommes-nous. En revanche, on peut étudier la manière dont on avance. Toujours ça de gagné. Les données empiriques en ce qui concerne l’effort d’innovation d’une part (inventions, cadence d’amortissement des technologies établies etc.) et ses résultats d’autre part (productivité, consommation d’énergie etc.) suggèrent qu’on ne peut pas être tout à fait sûr si on avance du tout. Je veux dire, oui, je sais qu’il y a du nouveau, seulement toutes ces nouveautés, ça semble ne pas apporter grand-chose en termes de progrès. Bon, je sais que le déficit alimentaire moyen dans l’économie mondiale s’est rétréci de presque 172 kilocalories par jour par personne en 1992 jusqu’à peine plus de 88 kilocalories en 2016. Oui, d’accord, couper le déficit alimentaire moyen par deux tout en doublant la population est quelque chose d’intéressant. Néanmoins, ce n’est pas vraiment ce qu’un économiste respectable traiterait de progrès technologique comme il faut.

A partir de là, ça devient clair et logique : peut-être bien qu’il faut réviser ce qu’un économiste respectable peut espérer d’un changement technologique décent. Ça, ça pourrait bien être l’importance scientifique de mon article : montrer une façon alternative d’appréhender le comment du changement technologique. Ici, je sens que je devrais retourner à la fonction de production, telle qu’elle a été formulée originellement par Charles W. Cobb et Paul H. Douglas en 1928[1]. Ces deux chercheurs ont construit in modèle, appelé plus tard « la fonction de production », où ils ont dérivé un produit intérieur brut modèle de l’accumulation des facteurs de production, capital et travail. Ce PIB modèle a montré une capacité étonnante à tenir le pas au PIB réel, et c’en est resté comme ça : le modèle est robuste, donc on le garde. Un modèle théorique, vous savez, on le traite parfois un peu comme un chien égaré : il suffit qu’il ne pisse pas dans la chambre à coucher et qu’il ne vous morde pas (tout de suite) pour que vous cédiez aux supplications de vos enfants (ou bien de votre subconscient) et que vous le gardiez. J’ai déjà développé une analyse de la fonction de production ( jetez un coup d’œil sur “Un modèle mal nourri” ) donc maintenant je vais me concentrer sur les intentions déclarées par Charles W. Cobb et Paul H. Douglas. Dans la partie finale de leur article ils soulignent très fort qu’ils considèrent leur modèle comme pas tout à fait apte à mesurer le progrès technologique en tant que tel et que ce qu’ils avaient construit est, dans leur dessein, plutôt un modèle structurel, fixe dans le temps. En d’autres mots, prendre les coefficients de leur modèle, donc les exposantes de capital et de travail, plus le coefficient commun « 1,01 » du début de l’équation, et les grouper tous sous la même enseigne de « productivité » porte toutes les marques d’un malentendu scientifique. Remarquez : les malentendus, ça débouche parfois sur des trucs beaucoup plus intéressants que ce que nous avions initialement à l’esprit. Néanmoins, l’usage que fait Joseph Schumpeter du concept théorique de fonction de production, dans ses « Business Cycles », dix ans après l’article de Charles W. Cobb et Paul H. Douglas, fut du freelance théorique complet, sans aucun lien vraiment solide avec le modèle initial.

Ce que j’essaie si laborieusement de dire est qu’il est peut-être erroné d’espérer la productivité totale des facteurs, ou TFP pour ses amis anglophones, de croître tout le temps, d’année en année, comme la technologie change. Donc, il pourrait bien être erroné, aussi, de modeler le changement technologique comme action prise en vue d’un accroissement de productivité. Il se peut que le résultat essentiel à espérer de la part du changement technologique soit l’adaptation et que l’innovation définie en termes behavioristes soit une adaptation intelligente. Remarquez, s’il vous plaît, que je n’ai pas utilisé l’expression « action en vue d’adaptation » mais bien « adaptation intelligente ». Il y a une nuance à explorer, ici. Si je dis « action en vue d’adaptation », j’assume qu’il y ait une vue, un horizon glorieux avec les objectifs à long terme qui se dessinent comme une chaîne montagneuse sur ledit horizon. Seulement voilà, tous ceux qui sont ne serait-ce qu’un peu familiers avec la pratique de changement technologique sur le terrain savent que cet horizon bien net est l’une de dernières choses qu’on voit. En fait, d’habitude, cet horizon, on le voit dans le rétroviseur, quand tout a été fait et accompli. La perspective dominante au cours d’un changement technologique réel est l’incertitude.

C’est bien là que j’aborde la question d’adaptation intelligente. J’ai appris que j’utilise mon intelligence de la façon la plus intense et la plus efficace en présence d’incertitude. Les pires conneries dans ma vie, je les ai commises lorsque j’avais une certitude de fer sur un sujet. Peu importe le sujet, en fait, c’est la présence de la certitude qui a tendance à me pousser dans la stupidité. J’ai une base évolutionniste forte, là. Nous, les humains, fallait qu’on soit vraiment futés lorsqu’on voyait ce lion de montagne, de l’autre côté du bison fraichement tué et on devait remuer les méninges pour prendre la situation en main. Adaptation intelligente veut dire qu’on recombine des informations diverses et distinctes, et on teste des idées ainsi obtenues dans une séquence d’essais à issue incertaine. Plus grand est le challenge (deux lions de montagne au lieu d’un, par exemple) plus d’essais nous avons besoin de faire pour arriver à une solution viable (survie). Il en faut encore plus pour maximaliser le résultat (survie plus bison plus peau du lion de montagne).

Adaptation intelligente implique donc une séquence d’essais. Dans le monde des technologies cela veut dire l’impératif d’expérimenter avec plusieurs technologies. Là, une autre nuance apparaît. Je peux bien expérimenter avec plusieurs technologies à la fois, mais ma capacité de conduire des essais simultanés est presque toujours limitée. Elle est limitée par les ressources à portée de main, ainsi que par l’utilité que je peux dériver d’une séquence d’essais. Essais simultanés veulent dire résultats simultanés (ou presque). Je ne peux donc pas utiliser le résultat d’un essai dans un autre essai simultané. En revanche, lorsque je fais des expériences en séquence, je peux utiliser la boucle de feedback pour optimiser l’efficacité marginale du dernier essai courant. Je fais face à ces deux lions de montagne. Si je pouvais faire toute une séquence d’essais du type « si je fais un pas vers la gauche, où est-ce que tu vas aller, chaton ? », ça ma placerait dans une situation singulièrement plus avantageuse que la vie réelle, ou j’ai juste une leçon avant l’examen final.

Il est donc possible que nous, je veux dire la civilisation humaine, nous effectuons tout ce truc de changement technologique comme un séquencement délibéré d’un processus d’adaptation sans objectif fixe, juste avec un impératif, celui de manger à notre faim. Une chose qu’on accomplit, pas à pas, comme civilisation, est la réduction de notre déficit alimentaire. Comme c’est un déficit moyen par capita, il peut être traduit dans la vie réelle comme une fraction de la population qui est tellement mal nourrie qu’elle est incapable de prendre pleinement part dans la vie collective. Lorsque j’étudiais la corrélation entre le déficit alimentaire et les indicateurs de cadence dans le changement technologique (demandes de brevet par million d’habitants, amortissement agrégé par tête d’habitant), nos technologies semblent changer le plus vite dans les pays avec un déficit alimentaire vraiment modéré, moins de 88 kilocalories par jour par personne. Le truc intéressant est qu’à l’échelle globale, notre déficit alimentaire moyen commence à entrer précisément dans cette intervalle. Il se peut que comme civilisation, nous sommes maintenant à l’apogée du changement technologique.

C’est donc ça, la relevance scientifique de mon modèle évolutionniste : tracer l’esquisse d’un chemin de recherche (je n’aurais pas l’arrogance de tracer un chemin définitif) ou les changements technologiques dans l’économie mondiale peuvent être représentés comme une adaptation intelligente de notre espèce à l’impératif de nous nourrir, tout simplement.

[1] Charles W. Cobb, Paul H. Douglas, 1928, A Theory of Production, The American Economic Review, Volume 18, Issue 1, Supplement, Papers and Proceedings of the Fortieth Annual Meeting of the American Economic Association (March 1928), pp. 139 – 165

I cannot prove we’re smart

My editorial

I am preparing an article, which presents, in a more elegant and disciplined form, that evolutionary model of technological change. I am going once again through all the observation, guessing and econometric testing. My current purpose is to find simple, intelligible premises that all my thinking started from. ‘Simple and intelligible’ means sort of hard, irrefutable facts, or, foggy, unresolved questions in the available literature. This is the point, in scientific research, when I am coining up statements like: ‘I took on that issue in my research, because facts A,B, C suggest something interesting, and the available literature remains silent or undecided about it’. So now, I am trying to reconstruct my own thinking and explain, to whomever would read my article, why the hell did I adopt that evolutionary perspective. This is the point when doing science as pure research is being transformed into scientific writing and communication.

Thus, facts should come first. The Schumpeterian process of technological progress can be decomposed into three parts: the exogenous scientific input of invention, the resulting replacement of established technologies, and the ultimate growth in productivity. Empirical data provides a puzzling image of those three sub-processes in the modern economy. Data published by the World Bank regarding science, research and development allow noticing, for example, a consistently growing number of patent applications per one million people in the global economy (see http://data.worldbank.org/indicator/IP.PAT.RESD ). On the other hand, Penn Tables 9.0 (Feenstra et al. 2015[1]) make it possible to compute a steadily growing amount of aggregate amortization per capita, just as a growing share of aggregate amortization in the global GDP (see Table 1 in the Appendix). Still, the same Penn Tables 9.0, indicate unequivocally that the mean value of Total Factor Productivity across the global economy has been consistently decreasing since 1979 until 2014.

Of course, there are alternative views of measuring efficiency in economic activity. It is possible, for example, to consider energy efficiency as informative about technological progress, and the World Bank publishes the relevant statistics, such as energy use per capita, in kilograms of oil equivalent (see http://data.worldbank.org/indicator/EG.USE.PCAP.KG.OE ). Here too, the last decades do not seem to have brought any significant slowdown in the growth of energy consumption. The overall energy-efficiency of the global economy, measured with this metric, is decreasing, and there is no technological progress to observe at this level. A still different approach is possible, namely that of measuring technological progress at the very basic level of economic activity, in farming and food supply. The statistics reported by the World Bank as, respectively, the cereal yield per hectare ( see http://data.worldbank.org/indicator/AG.YLD.CREL.KG ), and the depth of food deficit per capita (see http://data.worldbank.org/indicator/SN.ITK.DFCT ), allow noticing a progressive improvement, at the scale of global economy, in those most fundamental metrics of technological performance.

Thus, the very clearly growing effort in research and development, paired with a seemingly accelerating pace of moral ageing in established technologies, occurs together with a decreasing Total Factor Productivity, decreasing energy efficiency, and just very slowly increasing efficiency in farming and food supply chains. Now, in science, there are basically three ways of apprehending facts: the why, the what, and the how. Yes, I know, there is a fourth way, the ‘nonsense!’ one, currently in fashion as ‘this is fake news! we ignore it’. Still, this fourth way is not really science. This is idiocy dressed fancily for an electoral meeting. So, we have three: the why, the what, and the how.

The why, or ‘Why are things happening the way they are?’, is probably the oldest way of starting science. ‘Why?’ is the intuitive way we have of apprehending things we don’t quite understand, like ‘Why is this piece of iron bending after I have left it close to a furnace?’. Probably, that intuitive tendency to ask for reasons reflects the way our brain works. Something happens, and some neurons fire in response. Now, they have to get social and to inform other neurons about that something having happened. Only in the world of neurons, i.e. in our nervous system, the category ‘other neurons to inform’ is quite broad. There are millions of them, in there. Besides, they need synapses to communicate, and synapses are an investment. Sort of a fixed asset. So, neurons have to invest in creating synapses, and they have a wide choice as for where exactly they should branch. As a result, neurons like fixed patterns of communication. Once they make a synaptic connection, they just use it. The ‘why?’ reflects this predilection, as in response we expect ‘Because things happen this way’, i.e. in response to this stimulus we fire that synaptic network, period.

The problem with the ‘why?’ is that it is essentially deterministic. We ask ‘why?’ and we expect ‘Because…’ in return. The ‘Because…’ is supposed to be reassuringly repetitive. Still, it usually is not. We build a ‘Because…’ in response to a ‘why?’, and suddenly something new pops up. Something, which makes the established ‘Because…’ look a little out of place. Something that requires a new ‘Because…’ in response to essentially the same ‘why?’. We end up with many becauses being attached to one why. Picking up the right because for the situation at hand becomes a real issue. Which because is the right because can be logically derived from observation, or illogically derived from our emotional stress due to cognitive dissonance. Did you know that the experience of cognitive dissonance can trigger, in a human being, stronger a stress reaction than the actual danger of death? This is probably why we do science. Anyway, choosing the right because on the illogical grounds of personal emotions leads to metaphysics, whilst an attempt to pick up the right because for the occasion by logical inference from observation leads to the next question: the ‘what?’. What exactly is happening? If we have many becauses to choose between, choosing the right one means adapting our reaction to what is actually taking place.

The ‘what?’ is slightly more modern than the ‘why?’. Probably, mathematics were historically the first attempt to harness the subtleties of the ‘what?’, so we are talking about settled populations, with a division of labour allowing some people to think about things kind of professionally. Anyway, the ‘what?’ amounts to describing reality so as the causal sequence of ‘because…’ is being decomposed as a sequence. Instead of saying ‘C happens because of B, and B happens because of A’, we state a sequence: A comes first, then comes B, and finally comes C. If we really mean business, we observe probabilities of occurrence and we can make those sequences more complex and more flexible. A happens with a probability of 20%, and then B can happen with a probability of 30%, or B’ can happen at 50% odds, and finally we have 20% of chances that B’’ happens instead. If it is B’’ than happens, it can branch into C, C’ or C’’ with the respective probabilities of X, Y, Z etc.

Statistics are basically a baby of the ‘what?’. As the ‘why?’ is stressful and embarrassingly deterministic, we dodge and duck and dive into the reassuringly cool waters of the ‘what?’. Still, I am not the only one to have a curious ape inside of me. Everyone has, and the curiosity of the curious ape is neurologically wired around the ‘why?’ pattern. So, just to make the ape calm and logical, whilst satisfying its ‘why’-based curiosity, we use the ‘how?’ question. Instead of asking ‘why are things happening the way they are?’, so instead of looking for fixed patterns, we ask ‘how are things happening?’. We are still on the hunt for links between phenomena, but instead of trying to shoot the solid, heavy becauses, we satisfy our ambition with the faster and more flexible hows. The how is the way things happen in a given context. We have all the liberty to compare the hows from different contexts and to look for their mutual similarities and differences. With enough empirical material we can even make a set of connected hows into a family, under a common ‘why?’. Still, even with such generalisations, the how is always different an answer from ‘because…’. The how is always context-specific and always allows other hows to take place in different contexts. The ‘because…’ is much more prone to elbow its way to the front of the crowd and to push the others out of the way.

Returning to my observations about technological change, I can choose, now, between the ‘why?’, the ‘what?’, and the “how?’. I can ask ‘Why is this apparent contradiction taking place between the way technological change takes place, and its outcomes in terms of productivity?’. Answering this question directly with a ‘Because…’ means building a full-fledged theory. I do not feel ready for that, yet. All these ideas in my head need more ripening, I can feel it. I have to settle for a ‘what?’, hopefully combined into context-specific hows. Hows run fast, and they change their shape, according to the situation. If you are not quick enough to run after a how, you have to satisfy yourself with the slow, respectable because. Being quick, in science, means having access to empirical data and be able to test quickly your hypotheses. I mean, you can be quick without access to empirical data, but then you just run very quickly after your own shadow. Interesting, but moderately productive.

So I am running after my hows. I have that empirical landscape, where a continuously intensifying experimentation with new technologies leads, apparently, to decreasing a productivity. There is a how, camouflaging itself in that landscape. This how assumes that we, as a civilisation, randomly experiment with new technologies, kind of which idea comes first, and then we watch the outcomes in terms of productivity. The outcomes are not really good – Total Factor Productivity keeps falling in the global economy – and we still keep experimenting at an accelerating pace. Are we stupid? That would be a tempting because, only I can invert my how. We are experimenting with new technologies at an increasing pace as we face disappointing outcomes in terms of productivity. If technology A brings, on the long run, decreasing productivity, we quickly experiment with A’, A’’, A’’’ etc. Something that we do brings unsatisfactory results. We have two options then. Firstly, we can stop doing what we do, or, in other words, in the presence of decreasing productivity we could stop experimenting with new technologies. Secondly, we can intensify experimentation in order to find efficient ways to do what we do. Facing trouble, we can be passive or try to be clever. Which option is cleverer, at the end of the day? I cast my personal vote for trying to be clever.

Thus, it would turn out that the global innovative effort is an intelligent, collective response to the unsatisfactory outcomes of previous innovative effort. Someone could say that this is irrational to go deeper and deeper into something that does not bring results. That is a rightful objection. I can formulate two answers. First of all, any results come with a delay. If something is not bringing results we want, we can assume it is not bringing them yet. Science, which allows invention, is in itself quite a recent invention. The scientific paradigm we know today has taken definitive shape in the 19th century. Earlier, we basically have been using philosophy in order to invent science. It makes some 150 years that we can use real science to invent new technologies. Maybe it has not been enough to learn how to use science properly. Secondly, there is still the question of what we want. The Schumpeterian paradigm assumes we want increased productivity but do we really? I can assume, very biologically, what I already signalled in my previous posts: any living species tends to maximize its hold on the environment by absorbing as much energy as possible. Maybe we are not that far from amoeba, after all, and, as a species, we collectively tend towards maximizing our absorption of energy from our environment. From this point of view, technological change that leads to increasing our energy use per capita and to engaging an ever growing amount of capital and labour into the process could be a perfectly rational behaviour.

All that requires assuming collective intelligence in the mankind. Proving the existence of intelligence is both hard and easy. On the one hand, culture is proof of intelligence: this is one of the foundational principles in anthropology. From that point of view, we can perfectly assume that the whole human species has collective intelligence. Still, an economist has a problem with this view. In economics, we assume individual choice. Can individual choice be congruent with collective intelligence, i.e. can individual, conscious behaviour change in step with collective decisions? Well, we did Renaissance, didn’t we? We did electricity, we did vaccines, we did religions, didn’t we? I use the expression ‘we did’ and not ‘we made’, because it wasn’t that one day in the 15th century we collectively decided that from now on, we sculpt people with no clothes on and we build cathedrals on pseudo-ancient columns. Many people made individual choices, and those individual choices turned out to be mutually congruent, and produced a coherent social change, and so we have Tesla and Barbie dolls today.

Now, this is the easy part. The difficult one consists in passing from those general intuitions, which, in the scientific world, are hypotheses, to empirical verification. Honestly, I cannot directly, empirically prove we are collectively intelligent. I reviewed thoroughly the empirical data I have access to and I found nothing that could serve as direct proof of collective intelligence in the mankind. Maybe this is because I don’t know how exactly could I formulate the null hypothesis, here. Would it be that we are collectively dumb? Milton Friedman would say that in such a case, I have to options: forget it or do just as if. In other words, I can drop entirely the hypothesis of collective intelligence, with all its ramifications, or construe a model implying its veracity, so treating this hypothesis as an actual assumption, and see how this model fits, in confrontation with facts. In economics, we have that assumption of efficient markets. Right, I agree, they are not necessarily perfectly efficient, those markets, but they arrange prices and quantities in a predictable way. We have the assumption of rational institutions. In general, we assume that a collection of individual acts can produce coherent social action. Thus, we always somehow imply the existence of collective intelligence in our doings. Dropping entirely this hypothesis would be excessive. So I stay with doing just as if.

[1] Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer (2015), “The Next Generation of the Penn World Table” American Economic Review, 105(10), 3150-3182, available for download at http://www.ggdc.net/pwt

Les grimpeurs, les dormeurs et les foutus

Quelques mots pour lancer la discussion

Je suis en train de rédiger un article bien sage et bien scientifique pour présenter mon modèle de changement technologique comme un processus évolutif, plus exactement comme un processus de reproduction sexuée avec sélection. Le truc un peu barbant qu’il faut faire dans un article scientifique est ce qu’on appelle « la revue de la littérature ». Je connais ces situations quand mon article est passé en revue avant la publication, il y a deux critiques qui le font et l’un d’eux lève l’argument que « la revue de la littérature n’est pas suffisamment exhaustive », pendant que l’autre dit quelque chose d’exactement opposé : « la revue de la littérature est excessivement large et en grande partie peu pertinente ». Alors voilà, lorsque je prépare un article pour la publication, il faut que je démontre une dose de clairvoyance et que je prévoie (clairevoie) les préférences des critiques en ce qui concerne l’exhaustivité de la revue de littérature.

Bon, alors moi, maintenant, je tends vers la perfection en ce qui concerne ma littérature. La perfection, c’est l’unité. Je prends un livre, juste un, mais un bon : l’œuvre monumentale d’Arnold Toynbee, intitulée tout simplement « Study of history ». Publié pour la première fois en 1939, dans sa version originale ce livre était monumental. Six volumes, pas moins. A en croire les revues publiées juste après, l’admiration pour le contenu était étroitement liée aux suggestions d’écrire une version abrégée. Seulement voilà, c’était la guerre. Il fallait attendre 1946 pour que cette version plus abordable soit écrite et ceci d’une façon originale, par un autre chercheur et un enseignant passionné d’histoire : David Churchill Somervell. A en croire l’introduction écrite par Arnold Toynbee pour ce volume abrégé, ce travail titanesque a été effectué par D.C. Somervell tout à fait en privé. Ce n’est qu‘après avoir fini d’écourter l’œuvre monumentale d’origine que Somervell a informé Toynbee, par lettre, qu’il a fini.

De toute façon, moi, dans ma recherche de perfection, j’en reste à la version abrégée que vous pouvez trouver sur la version Word Press de mon blog par ce lien hypertexte-là . Quand le lis ce livre d’Arnold Toynbee, je montre la révérence que je dois à un grand classique, ce qui veut dire que pour une fois je commence ma lecture par le début et non pas par la fin. Je vais donc bien sagement vers la table des matières. Un titre attire mon attention. Dans le chapitre II, « Les genèses des civilisations », le premier sous-chapitre est intitulé « Le problème est comment ne pas le formuler ». Voilà qui a l’air intéressant. J’y saute donc, et alors qu’Arnold Toynbee introduit une idée bien fascinante : la mimesis, ou imitation, est un trait essentiel de la vie sociale et la différence entre les civilisations avancées et les sociétés primitives réside dans la direction que prend la mimesis. Dans les sociétés primitives, chaque génération imite les générations précédentes et de cette façon, le même schéma de vie sociale est reproduit toujours à nouveau. Dans les civilisations avancées la force principale de mimétique est dirigée vers l’imitation des grandes personnalités créatives, qui apportent du changement.

Plus loin j’avance, plus intéressant ça devient, car Arnold Toynbee dit ensuite que bien que ce trait spécifique fait la différence la plus marquée entre les sociétés primitives et les civilisations avancées, il n’est ni permanent ni même fondateur. Apparemment, beaucoup de sociétés que nous connaissons aujourd’hui comme primitives avaient dans leur passé des épisodes de développement rapide, lorsque leur moteur mimétique était orienté vers les leaders créatifs. Ce n’est qu’à un certain moment de leur histoire qu’elles se sont visiblement dit « La barbe avec ce progrès. On arrête. Eh, tout le monde ! M’entendez, vous ? On arrête de progresser, entendu ? Désormais, on s’imite nous-mêmes en boucle. Monsieur Bjruhhudjis, ne m’avez-vous pas entendu correctement ? On a-r-r-ê-t-e. Veuillez bien jeter ce compas dans la poubelle. En fait, cachez cette poubelle quelque part, aussi. Jetez là dans l’océan, ce sera la meilleure idée. Dans quelques siècles, il faudra qu’on soit bien primitifs et ce n’est pas avec un compas caché dans une poubelle qu’on pourra y arriver ».

Toynbee approfondit cette idée générale avec une métaphore juteuse : « Les sociétés primitives, telles que nous les connaissons, peuvent être comparées aux gens qui sont étendus, en torpeur, sur un rebord rocheux situé sur une montagne, avec un précipice au-dessous et un autre au-dessus ; les civilisations peuvent être comparées aux compagnons de ces dormeurs, qui se sont levés et commencé à grimper la falaise au-dessus ; tandis que nous, pour notre part, pouvons être comparés aux observateurs dont le champ de vision est limité à ce rebord rocheux ainsi qu’à la partie inférieure de la falaise au-dessus ; nous sommes venus sur la scène au moment quand les membres différents du groupe ont pris leurs positions respectives. Au premier abord, nous pouvons être tentés de tirer une distinction absolue entre les deux groupes, en admirant les grimpeurs comme athlètes et en regardant les dormeurs comme des paralytiques ; encore, après réflexion, nous trouverons qu’il est plus prudent de suspendre notre jugement. Après tout, les dormeurs ne peuvent pas être vraiment des paralytiques, car ils ne pouvaient pas avoir été nés sur ce rebord rocheux, et les seuls muscles qui avaient pu les hisser jusqu’à cet arrêt au-dessus du précipice, avaient dû être les leurs. D’autre part, leurs compagnons qui grimpent maintenant viennent de quitter ce rebord et de commencer l’escalade ; et comme le rebord prochain est hors notre vue, nous ne connaissons ni la hauteur ni l’ardeur de l’étape prochaine. Nous savons seulement qu’il est impossible de s’arrêter et se reposer avant que le rebord prochain soit atteint, où que ce soit. Alors, même si nous pouvions évaluer la force, l’habileté et le nerf de chaque grimpeur à présent, nous ne pouvons pas juger si n’importe lequel d’entre eux à une chance quelconque d’atteindre le prochain rebord, ce qui est bien le but de leur entreprise présente. Néanmoins, nous pouvons être sûrs que certains d’entre eux n’y arriveront jamais. Nous pouvons observer aussi que, pour chacun qui grimpe maintenant avec effort, deux (nos civilisations disparues) avaient tombé sur le rebord au-dessous, vaincus. »

Ça donne à penser, cette métaphore d’Arnold Toynbee. Nous-mêmes, comme civilisation, ou en sommes-nous ? Est-ce que nous sommes en train de nous reposer sur une étagère rocheuse, après avoir escaladé un précipice ? Ou bien sommes-nous déjà assoupis sur ce support étroit et inconfortable ? Question rhétorique : avez-vous jamais essayé de roupiller sur du roc, abîme à côté ou pas ? Je vais vous dire : ce n’est pas évident du tout. Si on est une civilisation entière, il vaut mieux d’avoir inventé des matelas portables au préalable. Sérieusement, mon singe interne s’est mis en alerte après avoir lu cette citation de d’Arnold Toynbee. Il y a cette distinction entre grimpeurs temporairement en jeu, dormeurs temporairement hors-jeu, et finalement les vaincus, tombés, écrasés sous leur propre poids et définitivement hors-jeu. Et si cette distinction s’appliquait aussi bien à l’intérieur d’une civilisation donnée ? Et si la direction et la cadence d’évolution des technologies dépendait justement des proportions entre ces trois groupes : les coriaces, les dormeurs et les foutus ? Intéressant, surtout que dans ma propre recherche empirique sur la fonction évolutive de sélection des technologies j’ai bien trouvé des fonctions distinctes pour des structures sociales distinctes (consultez par exemple “Je n’en ai pas fini avec ce truc d’évolution” ).

Mon singe interne devient insupportable. Arnold Toynbee a tellement excité sa curiosité qu’il n’arrête pas de me pousser du coude et me rappeler un autre esprit brillant, mon compatriote, Alfred comte Korzybski, le père-fondateur de la sémantique générale, cette discipline à mi-chemin entre la science et la philosophie. Dans ces deux livres – “Manhood of Humanity” et surtout dans “Science and Sanity” – il expose une théorie, où l’espèce humaine a cette capacité unique de faire ce que Korzybski appelle « lier le temps » et qui veut dire la capacité d’apprendre de génération en génération de façon à ce que chaque génération consécutive sache plus que la précédente. C’est bien le même Korzybski que l’un des fondateurs de la Programmation Neurolinguistique, Richard Bandler, aimait citer eu égard à la différence entre la réalité et la représentation que nous en faisons dans notre tête.

Lorsque je combine les intuitions d’Arnold Toynbee avec celles d’Alfred Korzybski, je vois cette distinction « les grimpeurs, les dormeurs et les foutus » comme une distinction fondamentale entre des différentes façons d’appréhender la même réalité. Mon singe curieux pose alors une question : s’il y a ce truc de fonction évolutive de sélection, qui crée et impose une hiérarchie d’idées de technologies nouvelles, comment peut-elle induire une distinction à l’intérieur de la société, entre des groupes sociaux qui grimpent, ceux qui dorment et ceux qui ont complétement lâché prise ? Puis-je expliquer ce côté des phénomènes avec ma fonction de sélection ? La réponse la plus évidente qui me vient à l’esprit – donc pas nécessairement la bonne mais on fait avec ce qu’on – est que les groupes sociaux qui grimpent sont ceux qui attachent, d’une façon ou d’une autre, leur capital, leur organisation de travail et leur systèmes monétaires aux technologies les plus haut placées dans la hiérarchie, c’est-à-dire les technologies les plus préférées par les entités femelles du mécanisme de reproduction. Plus loin nous sommes du sommet de la hiérarchie, donc plus loin parmi les idées pas-tellement-préférées par les entités femelles de ce modèle, plus on a de chances de s’assoupir ou même de lâcher prise.

Being the giant mitochondria of that planet we call ours

A few words in the editorial

I had one day of break from writing (family occasion), and now I am plunging my brain back in that evolutionist bath I have been taking for the last few weeks. I am thinking about the fundamental mechanisms of social change. In terms of social construction and demolition, we have three basic tools, as a social species. Firstly, we can define and attribute social roles; secondly, we can collectively produce group identity; thirdly, we can arrange patterns of access to the most valuable resources. As strange as it might seem, this is it. This is our entire toolbox, as it comes to engineering one of the most vital systems in our existence, namely our society. Now, as I take that evolutionary stance I have been developing recently, it is all about sex and reproduction. In the first place, we reproduce ourselves: human organisms. A new human organism is born through the interaction of two older organisms: the female and the male. The female organism mixes its own genetic material with that supplied by the male organism, and grows a new specimen out of this mixed genetic material. The male is specialized in packaging and disseminating its own genetic code.

Preferences make structures. The more stable and complex are those preferences, the more complex and stable are the resulting structures. I explain. We, humans, as many other social species, form networks and hierarchies. It is possible to form a hierarchy by sheer force, where the stronger simply impose their will to the weaker. Still, hierarchies made just by brute violence are usually unstable, see gangs in prisons, for example. For a hierarchy to last, and thus for a whole social structure to maintain its cohesion, we need something more. In management and sociology that something is frequently designated as ‘leadership’. There is a whole evolutionary theory, represented, for example, by professor Jordan Peterson from the University of Toronto, whose point is that hierarchies are based on sexual and reproductive preferences. In the process of sexual reproduction, males and females get chosen for mating. That choice has a pattern in it: certain individual characteristics are much more preferred than others. Individuals who possess a relatively large repertoire of desirable characteristics become more valued as members of the society, and acquire higher a position in social hierarchies.

Whilst this is a coherent theory, there are some impolite facts, which just come and contradict it. I know it, believe me. There are impolite facts who come and contradict my own, beautiful theories, all the time. Facts just don’t know how to behave. Facts are weakly socialized. Anyway, the most important fact that comes and puts its muddy shoes on this precise table is the multitude of hierarchies. If we are in a tribe of chimps, the hierarchy is simple and unique. Chimps can do with a single mechanism of hierarchy-forming, based on a single preference function in mating and reproduction. On the other hand, we, humans, we tend to live in much larger communities than chimps, and in those large communities we have many hierarchies in parallel.

Interestingly, the long historical perspective shows us a gradual change in the multitude of human hierarchies. Hunters-gatherers lived in small groups, essentially structured in a single hierarchy. As they settled, farming cultures appeared, and hierarchies started to breed. In the very beginning, it was a distinction into religious hierarchies and the military ones. Then, business started to generate its own hierarchy. At a certain point, the ability to maintain a relative stability in those basic hierarchies gave rise to another, different hierarchy, the political one. Political leaders have usually been selected for their ability to maintain law, order, and smooth running of the business. Political hierarchies combine the capacity to muster complex social support, mostly observable as legitimation (consent that this is the Best President of All Times, earlier having existed as divine support for the ruler) combined with economic power. The feudal society was probably one of the most internally differentiated in terms of coexisting hierarchies. With the advent of absolutism, that flurry of coexisting and mutually overlapping hierarchies started to simplify. The constitutional state brought with it a further simplification, based on the essential distinction between formal hierarchies, regulated by positive law, on the one hand, and the informal ones, based essentially on custom and private preferences, on the other hand.

In some respects: yes, we are a simplified version of the most complex society known, the feudal one. Still, hierarchies are not all of society. There are networks of cooperation, too. Here, we come back to social roles, group identity, and access to resources. In a complex social structure, every individual has like three sets of coordinates: a set of social roles, another set of appurtenances to social groups, and finally the set of effective and enforceable claims on resources. Social change means, at the most fundamental level, that social roles, group identities and access to resources get reshuffled. Social change is also a process of transformation in the set of hierarchies in place. I am going to take a risk and connect these two dots: social change means a double rearrangement, one inside hierarchies, and another between hierarchies. By the way, isn’t it fascinating, how hard it is for us, humans, do describe our own social structures? In physics and chemistry, we make those sharp distinctions between different particles and molecules. Even in biology, when we essentially describe the living matter we are made of, we easily get to distinctions between types of cells and tissues. Still, when it comes to apprehending our social structures, we leave the world of simple distinctions. It looks as if defining the elementary particles of a society was much harder than defining the elementary particles of universe. Intriguing.

Anyway, the whole point of evolutionary theory in social sciences seems to be the possible translation of the evolutionary mechanisms, observable at the biological level, into the social reality. There is a twofold intuition behind it. Firstly, we can coherently assume that the fundamental biological mechanisms of selecting the mating partner, and thus forming an elementary hierarchy of mating partners, can be reproduced in more complex forms. We use the fundamental patterns, wired in our brains, to form big social structures. Secondly, most cases of social change are those of gradual emergence and progressive diffusion, ultimately ending in absolute domination, of one structural pattern (individual roles, group identities and access to resources) to the expense of another one. It suggests some kind of generational change, on the one hand, as well as some kind of rivalry between species, where the fastest breeder wins, on the other hand.

There is some kind of blurred landscape in the evolutionary theory: the concept of purpose. My private impression is that back in the day, when the Darwinian concepts, together with the modern empirical science, were elbowing and biting their way through the nearly fossilized world of philosophy based on Aristotle and Thomas Aquinas, some kind of temporary compromise emerged between the two camps. The conservatives agreed to admit the existence of evolution in species, but there just had to be some kind of metaphysical cherry on the top o that cake, and the cherry was purpose. OK, we agree that them apes evolved, but they got to have a purpose in this. Our Creator must have assigned some purpose to all that process. This is probably how the bizarre, hybrid notion of purposeful evolution has emerged. The common-folk approach to evolution is precisely that: we evolve with a purpose in view. We evolve towards something. Today, the concept of intelligent design is probably the most manifest expression of this compromise between empirical science and metaphysics.

With time, I have become cautious in discarding metaphysics. At the first scientific sight, they just seem to be ignorance dressed elegantly. Still, many a metaphysical reasoning, like the philosophical notion of logos, for example, turn out to have a deeply rational side, with just too little empirics to make it fully accountable. On the other hand, any amount of empirics requires some linguistic meaning to be intelligible and there, man, metaphysical thinking just pops up from under every stone you turn. The notion of ‘matter’ as opposed to ‘energy’, or the distinction between ‘space’ and ‘time’, are quite illustrative examples. Thus, whilst the concept of purposeful evolution is obviously metaphysical, I am cautious in rejecting it, as there might be a sensible bottom line to it. I discard that notion to the extent that the purpose we usually ascribe to evolution is somehow strangely concordant with the current social ideas in fashion. When conservatives win the elections, they claim the purpose of evolution is a better world based on religion. When progressists have the upper political hand, they claim evolution has the purpose to create a better world based on science. All that sounds nice. I love the idea of a better world. Only one question: how the hell can we know what that better world will look like?

I am giving two examples, to illustrate my doubts about purpose in evolution: cereals, and electronics. This is a patent fact that cereals – rice, wheat, rye, and company – have become the foundation of settled societies, the prime food of farmers. We use to say that we farm cereals. Now, an annoying fact comes in and puts its dirty feet on the table, once again. None of the cereal plants is a champion of survival in the wild, undomesticated form. Wheat or rice, without human help, would have never stood a chance to become those huge monocultures, wiping out entire ecosystems made of other organisms. We made those particular plants much more successful, in purely biological terms, than they could ever become without farming. We use to think we farm cereals, but maybe cereals are farming us? Absurd? Well, review the facts once again. They feed us, we feed them. Additionally, we have put together a whole system of breeding them. I had an interesting insight once, in a museum. I tried to grind rye in a quern, a hand-propelled device to grind grains into flour. Bloody exhausting, really. The guide to that particular museum told me that before the introduction of mills, the water-powered or the wind-powered ones, grinding grains in querns was a tricky trade-off: you could very well expend more energy than you gained from flour. So we invented mills, which allowed us to grind grains more efficiently, which pushed us to grow more grains. Who farms whom, here?

Similar story with electronics. We use to think and say that we make and use electronics. Still, if you just compare the amount of energy we put into making and maintaining those microscopic microchips with the amount of energy we get out of their usage, the bottom line is not really obvious. Take China. They are literally flooding the world with electronics, but as a society they still starve slightly, 74 kilocalories per person per day, on average. You take Vietnam, and the gap between the output of electronics and the caloric deficit goes even deeper. The more electronics they make, over the last decades, the less they starve. They need to make electronics in order to have food. We consider electronics as a tool in our development, but it could very well turn out that we are a tool in the development of electronics. Sounds Matrixish and Terminatorish, I know. Still, look at the facts and think.

What is our evolutionary purpose, if any? In science, it is commonly assumed that purpose is rationally defined by function. We can set a purpose regarding something we are really good at. I would say that we, humans, are champions of energy. We excel at starving, for example. A human can outstarve most non-hibernating animals of comparable size, and even some microorganisms. Do you know that when we really mean business in fasting, I mean when we fast for many days, we can kill some bacteria in our bodies, and survive until the next meal? We can literally starve better than them. How would you call a machine, which can reduce its own consumption of fuel to a very low level, virtually zero, for many days, and still keep running, and still be able to jump, on demand, onto any new portion of fuel available? Very energy-efficient, probably. Class ‘A++’, in refrigerators. And smart, on the top of it.

We can starve and invent new ways of harnessing energy from our environment. Historically, we have been constantly meddling with combining many ways of transforming energy. Our muscle plus ox muscle, plus horse muscle, plus wind, plus water, plus Red Bull etc. It is all energy being transformed by our ingenuity. In the first year of undergraduate studies, students of psychology learn that emotions are very literally the energy system of our psyche. No emotions, no activity, sorry pal. Emotions are channelled chemical energy, on the other hand. Some neurons fire, they call some others on synapses, they gang up together, and they generate a complex biochemical state we call ‘alertness’, which then turns into ‘activity’. In order to do that, they need the emotional equipment, or the injection of hormones. Evolutionarily, the human frontal cortex, the source of individual intelligence, together with the power unit called ‘body’, could very well turn out to be among the most efficient tools in combining and channelling energy from many different sources. That could be our evolutionary purpose, if we have any: being the giant mitochondria of that planet we call ours. How do you like such a purpose? Frightening? Maybe. Still, if we are stubborn in assigning purposes to functions, this is one of the avenues we can very plausibly take.

Je n’en ai pas fini avec ce truc d’évolution

Mon éditorial pour aujourd’hui

Je pense au contexte de mon modèle évolutionniste de changement technologique. Je peux représenter les changements technologiques dans l’économie mondiale comme trois évolutions parallèles, en quelque sorte : celle du capital, celle du travail, ainsi que celle de l’argent (voyez “Money, essentially doesn’t give a s*** “  pour une explication détaillée). A chaque fois, la logique de l’hypothèse de base est similaire : chaque structure sociale distincte produit du changement technologique comme sélection faite par un élément femelle (capital physique, travail, ou argent liquide), dans un ensemble d’inventions possibles, et par la suite de cette sélection, des technologies nouvelles naissent. Ce qui évolue donc, c’est soit le capital physique, soit le marché et l’organisation du travail, soit le système monétaire.

En fait, j’ai tenté de mettre tous les trois organismes-mères dans le même modèle économétrique, pour voir leur poids relatif. C’est une procédure analytique qui est en fait à la limite de l’acceptable en statistique. Je sais que capital, travail et argent liquide sont corrélés dans leur incidence à l’intérieur du même système économique. C’est l’une des raisons pour qu’on appelle cette structure un système. Si je fourre ces trois valeurs agrégées dans la même équation, j’aurai, à coup sûr, ce qu’on appelle la cointégration entre variables explicatives. En principe, c’est une chose qu’on devrait éviter. Néanmoins, si je fais ça, je peux voir si l’un de ces facteurs est définitivement plus fort que les autres dans son influence sur la variable expliquée, qui est, en l’occurrence, le nombre de demandes de brevet dans un pays donné, en une année donnée.

Alors, j’avance, pas à pas, le long de cette frontière de l’acceptable, et je pose l’hypothèse que le changement technologique est une évolution conjointe des structures capitalistes, organisations d’emploi et systèmes monétaires, sous l’impulsion de sélection que ces trois types de structures sociales font dans l’ensemble de technologies nouvelles, possibles à développer sur la base d’idées scientifiques. Bien sûr, le lecteur attentif voit déjà la grosse lacune dans ce raisonnement : comment évoluent les structures qui génèrent les inventions, donc toute la sphère de recherche et développement ? Eh bien, l’idée un peu simpliste que j’ai est que toute organisation scientifique est une combinaison de capital physique (labos), travail (les gens vêtus de blanc) et argent liquide (comptes bancaires de ces gens vêtus de blanc). C’est simpliste, je l’admets, et néanmoins je reste bien dans le cadre de la théorie évolutionniste, là : les organismes mâles, qui génèrent le flot de l’information génétique indispensable à la reproduction, sont de la même espèce que les organismes femelles, donc ils ont la même substance génétique. Les organismes mâles ont juste une fonction différente.

De toute façon, si je veux faire ce test économétrique à la limite de l’acceptable, je reformule l’hypothèse ci-dessus comme une looongue équation :  ln(Nombre de demandes de brevet) = a1*ln(Capital physique) + a2*ln(Offre agrégée de travail, en heures travaillées) + a3*ln(Offre agrégée d’argent) + a4*ln(Taux d’amortissement) + a5*ln(Consommation d’énergie par tête d’habitant) + a6*ln(Densité de population) + a7*ln(Déficit alimentaire) + constante résiduelle. Une remarque, avant que je passe au test économétrique. Vous pouvez vous souvenir qu’à un moment donné j’ai ajouté la part de rémunération agrégée de travail dans le PIB, comme variable explicative à côté du capital physique, car elle s’est avérée significativement corrélée avec la constante résiduelle de l’un des modèles transitoires que j’avais utilisé. Ceci est vrai, mais cette fois, j’ajoute l’offre agrégée de travail dans l’équation et cette variable-là est inévitablement corrélée avec la part des salaires dans le PIB : cette dernière résulte soit du nombre d’heures travaillées (offre de travail), soit du niveau de salaire horaire moyen. J’ai donc deux valeurs agrégées qui sont évidemment liées l’une à l’autre et qui, en même temps, décrivent largement la même chose. C’est une redondance statistique qui, d’autre part, n’apporte aucune distinction qualitative, comme celle entre le capital et le travail. J’ai donc éliminé la part des salaires dans le PIB et je laissé le travail agrégé, comme nombre total d’heures travaillées en une année dans l’économie du pays donné.

De toute façon, je teste cette équation dans ma base des données composite, faite de Penn Tables 9.0 (Feenstra et al. 2015[1]) et de données additionnelles de la Banque Mondiale. Et comme je m’apprête à tester, vlam !, y a u truc qui me cogne droit entre les yeux :  dans les pays développés, le déficit alimentaire est zéro et comme le logarithme naturel de zéro, ça n’existe pas. Je ne peux pas appliquer cette équation, avec le déficit alimentaire dedans, aux pays qui génèrent, à première vue, deux tiers du nombre total de demandes de brevet dans ma base de données. En plus, comme j’avais déjà inclus le déficit alimentaire dans d’autres modèles étudiés plus tôt, il faudra que je revienne sur mes pas. J’avais en fait testé ma fonction évolutive de sélection dans un échantillon restreint des pays, où toutes les variables sont bien répertoriées et le déficit alimentaire est non-nul : Argentine, Brésil, Chile, Colombie, Equateur, Inde, Indonésie, Mexique, Pakistan, Pérou, Philippines, Corée du Sud, République Sud-Africaine, Thaïlande, Turquie. Zut ! Je déteste ces moments quand je découvre que je suis décalé de la réalité.

Bon, on se calme. Je me calme. D’abord, il faut évaluer l’échelle des dégâts. Sans faire de retours hâtifs en arrière, je prends mon modèle général, celui dont l’équation est présentée ci-dessus, deux paragraphes en arrière, et je la fais évoluer. Je la mute. La mutation richarde englobe les pays sans déficit alimentaire répertorié est donc sans déficit alimentaire dans le modèle. Taille d’échantillon : n = 1 404 observations, responsables pour 14 206 819 demandes de brevet au total. Le modèle explique R2 = 0,734 de la variance observée dans ce même nombre des demandes de brevet, d’année en année et de pays au pays. La mutation malnutrie de mon équation raconte le sort des pays avec déficit alimentaire officiel et officiellement inclus dans le modèle. Ces pays-là ont généré quelques 6 550 000 demandes de brevets, dans un échantillon de n = 317 observations valides, qui explique R2 = 0,812 de variance du nombre de demandes de brevets.

Malgré la taille très disparate de deux échantillons, les deux rendent une capacité explicative très similaire. Ça promet. Allons voir les paramètres. Alors, le modèle valide pour les pays sans déficit alimentaire répertorié, est détaillé ci-dessous :

Variable Coefficient Erreur standard Statistique t p-valeur
ln(Capital physique) 0,532 0,093 5,688 0,000
ln(Taux d’amortissement) 1, 0,244 4,092 0,000
ln(Consommation d’énergie par tête d’habitant) 1,425 0,08 17,728 0,000
ln(Densité de population) 0,049 0,023 2,126 0,034
ln(Offre agrégée de travail, en heures travaillées) 0,654 0,059 11,131 0,000
ln(Offre agrégée d’argent) -0,223 0,055 -4,097 0,000
constante résiduelle -11,818 0,874 -13,528 0,000

En revanche, le modèle appliqué aux pays avec déficit alimentaire officiel se présente comme dans le tableau suivant :

Variable Coefficient Erreur standard Statistique t p-valeur
ln(Capital physique) 0,176 0,252 0,7 0,484
ln(Taux d’amortissement) -1,539 0,566 -2,721 0,007
ln(Consommation d’énergie par tête d’habitant) 1,755 0,142 12,339 0,000
ln(Densité de population) 0,497 0,078 6,354 0,000
ln(Déficit alimentaire) -0,349 0,067 -5,19 0,000
ln(Offre agrégée de travail, en heures travaillées) 0,537 0,16 3,369 0,001
ln(Offre agrégée d’argent) 0,308 0,197 1,563 0,119
constante résiduelle -23,754 2,053 -11,568 0,000

Bon, maintenant je suis bien sage et j’applique la méthode de John Stuart Mill : similitudes d’une part, différences de l’autre. La similitude la plus frappante est l’importance de la consommation d’énergie par tête d’habitant. Plus exactement, ce sont deux variables – la consommation d’énergie par tête d’habitant et le taux d’amortissement – qui, de façon un peu surprenante, deviennent les musiciens de front de l’ensemble. Leurs coefficients de régression, combinés avec leurs p-valeurs respectives (qui mesurent la solidité de la corrélation), leur donnent une dominance très visible sur d’autres variables. Si je considère cette équation économétrique comme un modèle économique, c’est un modèle basé sur deux équilibres relatifs au nombre des demandes de brevet : l’équilibre énergétique et celui du cycle de vie des technologies. Là, une fois de plus, je me sens obligé à faire ce petit échange d’idées avec moi-même : comment est-ce possible, avec toute cette pression sur l’économie d’énergie, que le système économique tende, en fait, vers la maximisation de consommation d’énergie ?

Oui, je sais, ça pourrait être un de ces trucs purement statistiques, un concours d’accidents numériques sans signification véritable. J’admets que c’est possible, seulement ça arrive dans chaque modèle que j’ai testé durant la semaine dernière. Quelle que soit le cocktail des variables explicatives pour déterminer la variance du nombre des demandes de brevet, la consommation d’énergie toujours vient avec un coefficient positif et une p-valeur hautement significative. Plus on consomme d’énergie par tête d’habitant, plus on invente. Il y a un truc qui me vient à l’esprit, maintenant, une sorte d’observation sociologique approximative : il est bien vrai que pratiquement chaque technologie nouvelle inventée de nos jours est plus économe en énergie que ses prédécesseurs mais en même temps nous accumulons de plus en plus de ces technologies. A un moment donné, le ménage moyen avait un frigo, une télé et une machine à laver. Un frigo plus économe est arrivé, tout comme une machine à laver avec une classe énergétique meilleure. Seulement, entretemps, il y a eu le sèche-linge, le climatiseur et deux ordinateurs qui viennent en scène etc. Plus économe, chacune de ces bestioles prise séparément, mais leur compilation en un même endroit pompe en haut la consommation totale des ménages.

Plus on consomme d’énergie par tête d’habitant, plus on invente de technologies nouvelles. Même s’il y a l’impératif d’économiser l’énergie, moins d’énergie consommée pourrait bien vouloir dire moins d’inventions. Ça a tout l’air d’un cercle vicieux. Maintenant, je passe au cycle de vie des technologies, qui semble jouer des rôles opposés dans mes deux échantillons. Lorsque l’habitant moyen est repu, la fonction évolutive de sélection génère d’autant plus d’inventions brevetables que le cycle de vie des technologies en place est court. Plus vite ça vieillit, plus on invente du nouveau. C’est logique et c’est comment les deux tiers d’inventions brevetables ont l’air de naître. En revanche, lorsque l’habitant statistique (qui n’existe pas, bien sûr) pourrait bien profiter d’un repas de plus dans la journée, ça fonctionne à l’envers : plus vite les technologies en place vieillissent, moins on en invente de nouvelles. La seule explication logique que je vois est que dans les pays avec déficit alimentaire, la dépréciation des technologies en place prend littéralement le pain de la bouche des chercheurs et limite la capacité de générer des inventions. Un mâle malnutri, qui en plus doit tenir le pas à un stress social prononcé, produit moins de spermatozoïdes : fait scientifique prouvé.

Il y a une deux autres différences intéressantes. La majorité d’inventions brevetables, celles générées dans les pays sans déficit alimentaire, s’associe avec un équilibre prévisible du ratio « capital par demande de brevet » et n’aime pas trop l’offre d’argent. Comme s’il y avait une contradiction entre la création de crédit et celle d’inventions brevetables. En revanche, dans l’échantillon des pays avec déficit alimentaire, la création de technologies nouvelles, exprimées comme demandes de brevet, s’accompagne d’un équilibre plutôt aléatoire du ratio « capital par demande de brevet » et semble aller bien avec la création de crédit.

Ouff, je vois que je n’en ai pas fini avec ce truc d’évolution.

[1] Feenstra, Robert C., Robert Inklaar and Marcel P. Timmer (2015), “The Next Generation of the Penn World Table” American Economic Review, 105(10), 3150-3182, available for download at http://www.ggdc.net/pwt