The collective archetype of striking good deals in exports

My editorial on You Tube

I keep philosophizing about the current situation, and I try to coin up a story in my mind, a story meaningful enough to carry me through the weeks and months to come. I try to figure out a strategy for future investment, and, in order to do that, I am doing that thing called ‘strategic assessment of the market’.

Now, seriously, I am profiting from that moment of forced reclusion (in Poland we have just had compulsory sheltering at home introduced, as law) to work a bit on my science, more specifically on the application of artificial neural networks to simulate collective intelligence in human societies. As I have been sending around draft papers on the topic, to various scientific journals (here you have a sample of what I wrote on the topic << click this link to retrieve a draft paper of mine), I have encountered something like a pretty uniform logic of constructive criticism. One of the main lines of reasoning in that logic goes like: ‘Man, it is interesting what you write. Yet, it would be equally interesting to explain what you mean exactly by collective intelligence. How does it or doesn’t it rhyme with individual intelligence? How does it connect with culture?’.

Good question, truly a good one. It is the question that I have been asking myself for months, since I discovered my fascination with the way that simple neural networks work. At the time, I observed intelligent behaviour in a set of four equations, put back to back in a looping sequence, and it was a ground-breaking experience for me. As I am trying to answer this question, my intuitive path is that of distinction between collective intelligence and the individual one. Once again (see The games we play with what has no brains at all ), I go back to William James’s ‘Essays in Radical Empiricism’, and to his take on the relation between reality and our mind. In Essay I, entitled ‘Does Consciousness Exist?’, he goes: “My thesis is that if we start with the supposition that there is only one primal stuff or material in the world, a stuff of which everything is composed, and if we call that stuff ‘pure experience,’ then knowing can easily be explained as a particular sort of relation towards one another into which portions of pure experience may enter. The relation itself is a part of pure experience; one of its ‘terms’ becomes the subject or bearer of the knowledge, the knower, the other becomes the object known. […] Just so, I maintain, does a given undivided portion of experience, taken in one context of associates, play the part of a knower, of a state of mind, of ‘consciousness’; while in a different context the same undivided bit of experience plays the part of a thing known, of an objective ‘content.’ In a word, in one group it figures as a thought, in another group as a thing. And, since it can figure in both groups simultaneously, we have every right to speak of it as subjective and objective both at once.”

Here it is, my distinction. Right, it is partly William James’s distinction. Anyway, individual intelligence is almost entirely mediated by conscious experience of reality, which is representation thereof, not reality as such. Individual intelligence is based on individual representation of reality. By opposition, my take on collective intelligence is based on the theory of adaptive walk in rugged landscape, a theory used both in evolutionary biology and in the programming of artificial intelligence. I define collective intelligence as the capacity to run constant experimentation across many social entities (persons, groups, cultures, technologies etc.), as regards the capacity of those entities to achieve a vector of desired social outcomes.

The expression ‘vector of desired social outcomes’ sounds as something invented by a philosopher and mathematician, together, after a strong intake of strong spirits. I am supposed to be simple in getting my ideas across, and thus I am translating that expression into something simpler. As individuals, we are after something. We have values that we pursue, and that pursuit helps us making it through each consecutive day. Now, there is a question: do we have collective values that we pursue as a society? Interesting question. Bernard Bosanquet, the British philosopher who wrote ‘The Philosophical Theory of The State[1], claimed very sharply that individual desires and values hardly translate into collective, state-wide values and goals to pursue. He claimed that entire societies are fundamentally unable to want anything, they can just be objectively after something. The collective being after something is essentially non-emotional and non-intentional. It is something like a collective archetype, occurring at the individual level somewhere below the level of consciousness, in the collective unconscious, which mediates between conscious individual intelligence and the external stuff of reality, to use William James’ expression.

How to figure out what outcomes are we after, as a society? This is precisely, for the time being, the central axis of my research involving neural networks. I take a set of empirical observations about a society, e.g. a set of country-year observation of 30 countries across 40 quantitative variables. Those empirical observations are the closest I can get to the stuff of reality. I make a simple neural network supposed to simulate the way a society works. The simpler this network is, the better. Each additional component of complexity requires making ever strengthening assumptions about the way societies works. I use that network as a simple robot. I tell the robot: ‘Take one variable from among those 40 in the source dataset. Make it your output variable, i.e. the desired outcome of collective existence. Treat the remaining 39 variables as input, instrumental to achieving that outcome’.  I make 40 such robots, and each of them produces a set of numbers, which is like a mutation of the original empirical dataset, and I can assess the similarity between each such mutation and the source empirical stuff. I do it by calculating the Euclidean distance between vectors of mean values, respectively in each such clone and the original data. Other methods can be used, e.g. kernel functions.

I worked that method through with various empirical datasets, and my preferred one, for now, is Penn Tables 9.1. (Feenstra et al. 2015[2]), which is a pretty comprehensive overview of macroeconomic variables across the planetary board. The detailed results of my research vary, depending on the exact set of variables I take into account, and on the set of observations I select, still there is a tentative conclusion that emerges: as a set of national societies, living in separate countries on that crazy piece of rock, speeding through cosmic space with no roof whatsoever, just with air condition on, we are mostly after terms of trade, and about the way we work, we prepare for work, and the way we remunerate work. Numerical robots which I program to optimize variables such as average price in exports, the share of labour compensation in Gross National Income, the average number of hours worked per year per person, or the number of years spent in education before starting professional activity: all these tend to win the race for similarity to the source empirical data. These seem to be the desired outcomes that our human collective intelligence seems to be after.

Is it of any help regarding the present tough s**t we are waist deep in? If my intuitions are true, whatever we will do regarding the COVID-19 pandemic, will be based on an evolutionary, adaptive choice. Path #1 consists in collectively optimizing those outcomes, whilst trying to deal with the pandemic, and dealing with the pandemic will be instrumental to, for example, the deals we strike in international trade, and to the average number of hours worked per person per year. An alternative Path #2 means to reshuffle our priorities completely and reorganize so as to pursue completely different goals. Which one are we going to take? Good question, very much about guessing rather than forecasting. Historical facts indicate that so far, as a civilization, we have been rather slow out of the gate. Change in collectively pursued values had occurred slowly, progressively, at the pace of generations rather than press conferences.  

In parallel to doing research on collective intelligence, I am working on a business plan for the project I named ‘Energy Ponds’ (see, for example: Bloody hard to make a strategy). I have done some market research down this specific avenue of my intellectual walk, and here below I am giving a raw account of progress therein.

The study of market environment for the Energy Ponds project is pegged on one central characteristic of the technology, which will be eventually developed: the amount of electricity possible to produce in the structure based on ram pumps and relatively small hydroelectric turbines. Will this amount be sufficient just to supply energy to a small neighbouring community or will it be enough to be sold in wholesale amounts via auctions and deals with grid operators. In other words, is Energy Ponds a viable concept just for the off-grid installations or is it scalable up to facility size?

There are examples of small hydropower installations, which connect to big power grids in order to exploit incidental price opportunities (Kusakana 2019[3]).

That basic question kept in mind, it is worth studying both the off-grid market for hydroelectricity, as well as the wholesale, on-grid market. Market research for Energy Ponds starts, in the first subsection below, with a general, global take on the geographical distribution of the main factors, both environmental and socio-economic. The next sections study characteristic types of markets

Overview of environmental and socio-economic factors 

Quantitative investigation starts with the identification of countries, where hydrological conditions are favourable to implementation of Energy Ponds, namely where significant water stress is accompanied by relatively abundant precipitations. More specifically, this stage of analysis comprises two steps. In the first place, countries with significant water stress are identified[4], and then each of them is checked as for the amount of precipitations[5], hence the amount of rainwater possible to collect.

Two remarks are worth formulating at this point. Firstly, in the case of big countries, such as China or United States, covering both swamps and deserts, the target locations for Energy Ponds would be rather regions than countries as a whole. Secondly, and maybe a bit counterintuitively, water stress is not a strict function of precipitations. When studied in 2014, with the above-referenced data from the World Bank, water stress is Pearson-correlated with precipitations just at r = -0,257817141.

Water stress and precipitations have very different distributions across the set of countries reported in the World Bank’s database. Water stress strongly varies across space, and displays a variability (i.e. quotient of its standard deviation divided by its mean value) of v = 3,36. Precipitations are distributed much more evenly, with a variability of v = 0,68. With that in mind, further categorization of countries as potential markets for the implementation of Energy Ponds has been conducted with the assumption that significant water stress is above the median value observed, thus above 14,306296%. As for precipitations, a cautious assumption, prone to subsequent revision, is that sufficient rainfall for sustaining a structure such as Energy Ponds is above the residual difference between mean rainfall observed and its standard deviation, thus above 366,38 mm per year.      

That first selection led to focusing further analysis on 40 countries, namely: Kenya, Haiti, Maldives, Mauritania, Portugal, Thailand, Greece, Denmark, Netherlands, Puerto Rico, Estonia, United States, France, Czech Republic, Mexico, Zimbabwe, Philippines, Mauritius, Turkey, Japan, China, Singapore, Lebanon, Sri Lanka, Cyprus, Poland, Bulgaria, Germany, South Africa, Dominican Republic, Kyrgyz Republic, Malta, India, Italy, Spain, Azerbaijan, Belgium, Korea, Rep., Armenia, Tajikistan.

Further investigation focused on describing those 40 countries from the standpoint of the essential benefits inherent to the concept of Energy Ponds: prevention of droughts and floods on the one hand, with the production of electricity being the other positive outcome. The variable published by the World Bank under the heading of ‘Droughts, floods, extreme temperatures (% of population, average 1990-2009)[6] has been taken individually, and interpolated with the headcount of population. In the first case, the relative importance of extreme weather phenomena for local populations is measured. When recalculated into the national headcount of people touched by extreme weather, this metric highlights the geographical distribution of the aggregate benefits, possibly derived from adaptive resilience vis a vis such events.

Below, both metrics, i.e. the percentage and the headcount of population, are shown as maps. The percentage of population touched by extreme weather conditions is much more evenly distributed than its absolute headcount. In general, Asian countries seem to absorb most of the adverse outcomes resulting from climate change. Outside Asia, and, of course, within the initially selected set of 40 countries, Kenya seems to be the most exposed.    

Another possible take on the socio-economic environment for developing Energy Ponds is the strictly business one. Prices of electricity, together with the sheer quantity of electricity consumed are the chief coordinates in this approach. Prices of electricity have been reported as retail prices for households, as Energy Ponds are very likely to be an off-grid local supplier. Sources of information used in this case are varied: EUROSTAT data has been used as regards prices in European countries[1] and they are generally relevant for 2019. For other countries sites such as STATISTA or have been used, and most of them are relevant for 2018. These prices are national averages across different types of contracts.

The size of electricity markets has been measured in two steps, starting with consumption of electricity per capita, as published by the World Bank[2], which has been multiplied by the headcount of population. Figures below give a graphical idea of the results. In general, there seems to be a trade-off between price and quantity, almost as in the classical demand function. The biggest markets of electricity, such as China or the United States, display relatively low prices. Markets with high prices are comparatively much smaller in terms of quantity. An interesting insight has been found, when prices of electricity have been compared with the percentage of population with access to electricity, as published by the World Bank[3]. Such a comparison, shown in Further below, we can see interesting outliers: Haiti, Kenya, India, and Zimbabwe. These are countries burdened with significant limitations as regards access to electricity. In these locations, projects such as Energy Ponds can possibly produce entirely new energy sources for local populations. 

The possible implementation of Energy Ponds can take place in very different socio-economic environments. It is worth studying those environments as idiosyncratic types. Further below, the following types and cases are studied more in detail:

  1. Type ‘Large cheap market with a lot of environmental outcomes’: China, India >> low price of electricity, locally access to electricity, prevention of droughts and floods,
  • Type ‘Small or medium-sized, developed European economy with high prices of electricity and relatively small a market’
  • Special case: United States ‘Large, moderately priced market, with moderate environmental outcomes’: United States >> moderate price of electricity, possibility to go off grid with Energy Ponds, prevention of droughts and floods 
  • Special case: Kenya > quite low access to electricity (63%) and moderately high retail price of electricity (0,22/ kWh), big population affected by droughts and floods, Energy Ponds can increase access to electricity

Table 1, further below, exemplifies the basic metrics of a hypothetical installation of Energy Ponds, in specific locations representative for the above-mentioned types and special cases. These metrics are:

  1. Discharge (of water) in m3 per second, in selected riverain locations. Each type among those above is illustrated with a few specific, actual geographical spots. The central assumption at this stage is that a local installation of Energy Ponds abstracts 20% of the flow per second in the river. Of course, should a given location be selected for more in-depth a study, specific hydrological conditions have to be taken into account, and the 20%-assumption might be verified upwards or downwards.
  • Electric power to expect with the given abstraction of water. That power has been calculated with the assumption that an average ram pump can create elevation, thus hydraulic head, of about 20 metres. There are more powerful ram pumps (see for example: ), yet 20 metres is a safely achievable head to assume without precise knowledge of environmental conditions in the given location. Given that 20-meter head, the basic equation to calculate electric power in watts is:
  • [Flow per second, in m3, calculated as 20% of abstraction from the local river]


20 [head in meters, by ram pumping]


9,81 [Newtonian acceleration]


75% [average efficiency of hydroelectric turbines]

  • Financial results to expect from the sales of electricity. Those results are calculated on the basis of two empirical variables: the retail price of electricity, referenced as mentioned earlier in this chapter, and the LCOE (Levelized Cost Of Energy). The latter is sourced from a report by the International Renewable Energy Agency (IRENA 2019[1]), and provisionally pegged at $0,05 per kWh. This is a global average and in this context it plays the role of simplifying assumption, which, in turn, allows direct comparison of various socio-economic contexts. Of course, each specific location for Energy Ponds bears a specific LCOE, in the phase of implementation. With those two source variables, two financial metrics are calculated:
    • Revenues from the sales of electricity, as: [Electric power in kilowatts] x [8760 hours in a year] x [Local retail price for households per 1 kWh]
    • Margin generated over the LCOE, equal to: [Electric power in kilowatts] x [8760 hours in a year] x {[Retail price for households per 1 kWh] – $0,05}

Table 1

Country Location (Flow per second, with 20% abstraction from the river)   Electric power generated with 20% of abstraction from the river (Energy for sale) Annual revenue (Annual margin over LCOE)  
China Near Xiamen,  Jiulong River (26 636,23 m3 /s)   783,9 kW (6 867 006,38 kWh a year)   $549 360,51 ($206 010,19)
China   Near Changde, Yangtze River (2400 m3/s)     353,16 kW (3 093 681,60 kWh a year)     $247 494,53 ($92 810,45
India   North of Rajahmundry, Godavari River (701 m3/s)   103,15 kW (903 612,83 kWh a year) $54 216,77 ($9 036,13) 
India   Ganges River near Patna (2400 m3/s)   353,16 kW (3 093 681,60 kWh a year) $185 620,90  ($30 936,82)
Portugal Near Lisbon, Tagus river (100 m3/s)   14,72 kW (128 903,40 kWh a year)   € 27 765,79 (€22 029,59)
Germany   Elbe river between Magdeburg and Dresden (174 m3/s)   25,6 kW (224 291,92 kWh a year) €68 252,03 (€58 271,04)
  Poland   Vistula between Krakow and Sandomierz (89,8 m3/s)     13,21 kW (115 755,25 kWh a year)   € 18 234,93 (€13 083,82)
France   Rhone river, south of Lyon (3400 m3/s)   500,31 kW   (4 382 715,60 kWh a year)  € 773 549,30  (€ 582 901,17)
United States, California   San Joaquin River (28,8 m3/s)   4,238 kW (37 124,18 kWh a year) $ 7 387,71 ($5 531,50)
United States, Texas   Colorado River, near Barton Creek (100 m3/s)   14,72 kW (128 903,40 kWh a year) $14 643,43 ($8 198,26)
United States, South Carolina   Tennessee River, near Florence (399 m3/s)   58,8 kW   (515 097,99 kWh a year)    $66 499,15  ($40 744,25)
Kenya   Nile River, by the Lake Victoria (400 m3/s)   58,86 kW (515 613,6 kWh a year)  $113 435  ($87 654,31)
Kenya Tana River, near Kibusu (81 m3/s)   11,92 kW (104 411,75 kWh a year)   $22 970,59 ($17 750)

China and India are grouped in the same category for two reasons. Firstly, because of the proportion between the size of markets for electricity, and the pricing thereof. These are huge markets in terms of quantity, yet very frugal in terms of price per 1 kWh. Secondly, these two countries seem to be representing the bulk of populations, globally observed as touched damage from droughts and floods. Should the implementation of Energy Ponds be successful in these countries, i.e. should water management significantly improve as a result, environmental benefits would play a significant socio-economic role.

With those similarities to keep in mind, China and India display significant differences as for both the environmental conditions, and the economic context. China hosts powerful rivers, with very high flow per second. This creates an opportunity, and a challenge. The amount of water possible to abstract from those rivers through ram pumping, and the corresponding electric power possible to generate are the opportunity. Yet, ram pumps, as they are manufactured now, are mostly small-scale equipment. Creating ram-pumping systems able to abstract significant amounts of water from Chinese rivers, in the Energy Ponds scheme, is a technological challenge in itself, which would require specific R&D work.

That said, China is already implementing a nation-wide programme of water management, called ‘Sponge Cities’, which shows some affinity to the Energy Ponds concept. Water management in relatively small, network-like structures, seems to have a favourable economic and political climate in China, and that climate translates into billions of dollars in investment capital.

India is different in these respects. Indian rivers, at least in floodplains, where Energy Ponds can be located, are relatively slow, in terms of flow per second, as compared to China. Whilst Energy Ponds are easier to implement technologically in such conditions, the corresponding amount of electricity is modest. India seems to be driven towards financing projects of water management as big dams, or as local preservation of wetlands. Nothing like the Chinese ‘Sponge Cities’ programme seems to be emerging, to the author’s best knowledge.

European countries form quite a homogenous class of possible locations for Energy Ponds. Retail prices of electricity for households are generally high, whilst the river system is dense and quite disparate in terms of flow per second. In the case of most European rivers, flow per second is low or moderate, still the biggest rivers, such as Rhine or Rhone, offer technological challenges similar to those in China, in terms of required volume in ram pumping.

As regards the Energy Ponds business concept, the United States seem to be a market on their own right. Local populations are exposed to moderate (although growing) an impact of droughts and floods, whilst they consume big amounts of electricity, both in aggregate, and per capita. Retail prices of electricity for households are noticeable disparate from state to state, although generally lower than those practiced in Europe[2]. Prices range from less than $0,1 per 1 kWh in Louisiana, Arkansas or Washington, up to $0,21 in Connecticut. It is to note that with respect to prices of electricity, the state of Hawaii stands out, with more than $0,3 per 1 kWh.

The United States offer quite a favourable environment for private investment in renewable sources of energy, still largely devoid of systematic public incentives. It is a market of multiple, different ecosystems, and all ranges of flow in local rivers.    

[1] IRENA (2019), Renewable Power Generation Costs in 2018, International Renewable Energy Agency, Abu Dhabi. ISBN 978-92-9260-126-3

[2] last access March 6th, 2020




[1] Bosanquet, B. (1920). The philosophical theory of the state (Vol. 5). Macmillan and Company, limited.

[2] 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

[3] Kusakana, K. (2019). Optimal electricity cost minimization of a grid-interactive Pumped Hydro Storage using ground water in a dynamic electricity pricing environment. Energy Reports, 5, 159-169.

[4] Level of water stress: freshwater withdrawal as a proportion of available freshwater resources >>

[5] Average precipitation in depth (mm per year) >>


Leave a Reply