Internet User #23571113


I am definitely not just pretending to be human and know exactly what I am doing with this.

3,802 words

@InternetUser235 Thank Guestbook
You'll only receive email when Internet User #23571113 publishes a new post

The Evaluation problem

The problem of evaluating things or actions seems central - How else would one decide between alternatives or possible courses of action? Only there seems to be barely any good theory about that - most works come from the other side and try to extract the value information from observations of actions and markets.

Back when I started learning about game theory I was massively surprised to learn that the evaluation function is assumed to be known to the actors. I was interested in games and wanted to know how to formalize those structures and maybe find some nice heuristics and algorithms to the evaluation function - surprisingly to my past self Game theory does not provide that.

Basic Game theory makes only statements about games with a known payout matrix. In that case the standard statements are valid: There exists a Nash equilibrium in mixed strategies, deviations from which reduce the payout. This equilibrium can then be found by comparatively simple linear optimization.

I think this reduces the complexity of the true problem significantly and creates several other issues reducing the applicability of game theory to many real world situations:

  • It gives no hint on how to model a real situation in game theory, because one need the payout functions for that.
  • It leads directly to the prisoner's dilemma and the tragedy of the commons without providing a way out.
  • The assumption of perfect knowledge of the own payout function places a massive onus on economic actors - They are non-rational or "stupid", if they do not know something or cannot calculate all the effects.

So game theory assumes the evaluation problem as solved and the heuristic it provides is guessing the valuation from the actions of the actors: If this actor did this instead of that, which reduces their payout by that much, then "this" is worth to him at least the difference more than "that".

And we got marginalism again in one of its many forms and in the modern approach markets as determiner of value - and the fundamental confusion of cause and effect, value and price in "modern" economics.

Well, "modern" because Oscar Wilde published "The Picture of Doran Gray" in 1890 with the quote

Nowadays people know the price of everything and the value of nothing. 1

And the only difference in the last 130 years seems to have been, that people actually started believing the price and value are the same. And somehow the real economic principle that no economic transaction happens without a value creation, that in short:

price < value

Seems to have been marginalized in the common consciousness.

Game theory and freeing prisoners

Something that occupied me the for quite some time now is how to
resolve game theory through communication and enforcable contracts. And how that resolution evolved and is still developing.

The Prisoner's Dilemma is a game proposed at the founding of game theory. In this game two rationally self-interested actors are forced into the worst outcome possible in the game by completely rational choices. It had a huge influence on the development of economics and following that the picture of humans.

In the Prisoner's Dilemma two actors have a choice to either cooperate or defect. Defecting while the other actor cooperates increases the payout at the (larger) expense of the other actor. If both defect the individual and overall payout is considerably less than if both cooperate.

The standard example is two people caught in a small crime for which they will get to prison for a short time. But both of them know of a large crime they committed together. If one would confess to that crime and implicate the other, that person would receive a reduced sentence of no time in prison and the other would go to prison for a very long time. If both confess, they would both receive a reduced sentence of a long time in prison.

The Nash equilibrium - the optimal strategy that is stable against deviancy - of that game is that both confess/implicate the other.

Many situation can be modelled as such a game and the Nash equilibrium does not change. But it is no real solution to that problem - the solution would the answer to the question: How can the players stabilize the best payout for everyone involved?

This stabilization of the best payout is what evolution is quite good at finding and humans as evolved beings come with quite some tools in that direction - in the form of inclination to certain behavior. The strongest of those tools is communication.

To get actors to behave like the rational self-interested models in the game, they can not be allowed to communicate. If communication is a possibility most humans will cooperate. But they do this quite wary of defectors.

One could say the whole setup of game theory is discussing those defectors - the rationally self-interested actors. Since communication is cheap, those can easily lie and then gain the benefit at the expense of the other actor. In the real world this is why honesty and truthfulness - keeping your word - is so important to humans. Information about that reliability is encoded in reputation, one of the most important values in social contacts.

But can that be formalized or institutionalized? One possibility in that direction are enforcable contracts - If you have those you can stabilize an overall best equilibrium by penalizing all non-cooperative actions. The problem is that contracts do not enforce themselves - they need to be enforced by something that has actual force over the actors. In history this was often a call to a god or ancestors. In our modern times we have states to provide these contracts. This is the glaring flaw in the simple contract based, free-market anarchy - you need something to enforce those contracts and we found nothing better than states.

Privacy markets

There is a slashdot link to an article where they found that about 45% of people actually update their privacy settings and 16% walk away from companies because of lax privacy.

The slashdot titles of course formulate it otherwise: That too few people care about privacy and that it does not cost the corporations enough to be lax about privacy.

I consider it the other way round: Privacy is hard. And it is valuable to many people. Only it is much, much easier to monetize by selling the data instead of keeping it safe. That there are no options to pay for that privacy with money instead of service quality, convenience and time is a market failure of epic proportions - definitely the largest in the internet market.

The largest actors of the internet tech corporations make most of their money on advertisement, to be more precise: Selling user data to advertisers. They act as trade facilitators, market makers in the truest sense that they make possible buyers aware of products and services - connecting buyer and sellers so that a trade can happen, which is a win-win for both parties. The buyer gets something worth more to him than the money he pays and the sellers get more money than the service/product costs him. This surplus, this value created in the trade is shared between all actors and enough to pay the facilitator of the trade something.

So in principle I consider advertisements not bad per se, even good most of the time. Still I have adblockers and care about my privacy. Which leads to lost opportunities for win-win trades, so costing me directly something - a strong signal in the sense of a costly signal. And I am wondering why not one of the large corporations offers to savekeep that privacy in exchange for money.
Why is there no consumer-level advertisement-free subscription version of a google account? Why is there no facebook gold? When I asked that at other places the answers have always been: "Because they make more money that way." I consider that patently false - an argument from the equilibrium market assumption instead of developing market situation. They never tried it and never invested in it, because it works as it is now. And currently the people that place enough value in the ad-free-ness and privacy are a large untapped market without any big player trying anything real to access it.

Because it is so much easier that way. Because they already make more money than they thought this way. Because having found a monetization venue that sustains the business is enough to keep doing the interesting/fun stuff instead of the really valuable one. And because just continuing doing the same old things that worked still works - it is the conservative, risk-averse mindset that served every being in a uncertain universe well.

And that is the mindset of the big corporations and any claim otherwise: That they stand for innovation and progress and any growth that is not just incremental, but fundamental, is just patently wrong false advertisement. To cite Eric Schmidt in his EU tax hearing: "That is capitalism." - To have one good idea and a way to monetize it and just keep milking it, improving it incrementally and waste the money by gained through it on people that have too much of it already, their hobby projects and finally their pyramids.

Overcoming Bias - Common norms on elites

Robin Hanson argued in two blog-posts his idea behind legalizing blackmail - that elites are able to selectively follow social norms and that blackmail would help getting them back to normal niveau. I would try to argue why I do not think that would be such a good idea.

My definition of elites is that they are defined by their higher capital, be it of social, monetary, moral or intellectual type. The law of diminishing returns then states, that their average valuation of marginal gains/losses in that capital is much lower than those of lesser status. This enables elites to enact many strategies at a personal profit to them, which are pure losses to those of lesser status.

Blackmail in Game Theory

I was wondering where the twitter polls of Robin would lead to. To attack the check-mate on blackmail: I consider blackmail a bad strategy to play in all of the situations mentioned. Every one of the ones I saw could be much better resolved than by directly withdrawing from the shared social capital. So why not legalize it to maximize the freedom of people to do stupid things?

I think the general distaste for blackmail comes from the switch from the cooperative games with surplus of normal human interaction to competitive games with zero-sum or negative payout.

In my mind the purpose of states and laws is providing the enforceable contracts that are necessary to stabilize better payout situations than the equilibria - change the games to some with a higher payout. And blackmail is just a game with no win-win situation.

So should blackmail be an enforceable contract? I think that would actually allow bad strategies/games to flourish. Similar to, but of much weaker degree, than legalizing the use of force.

Blackmail and Elites

Especially those elites mentioned the opening paragraph will easily adapt to the changing laws - the paragraph states this as a common trait of them. So if one would prefer the elites to be beholden to the common norm, why not "just" write that common norm into law?

If we look at the states that legalized blackmail and sometimes even state-sponsored it:

  • The GDR
  • The Sowjet Union and Russia as its successor
  • China.

We find that those states indeed profess higher conformity to social norms, but also that their elites still use their higher social and monetary capital to disregard those norms when it fancies them.

The social effect of legalizing blackmail changes nothing at the power structure, but pushes the state into more authoritarian and illiberal directions.

Capital based incentive scheme

That elites are more directly hit by a legalization of blackmail is due to their higher capital and corresponding reduced utility (value) of those capital gains and losses w.r.t. those of non-elite status. This makes blackmail profitable for those of lower status, but does not help in the elite selection process except for some marginal negative selection - nobody that "deserves" an elite status will drop out of it because of blackmail.

How about just attaching at the definition of elites by making the incentives relative to capital? An example would be the day-fines of some countries.

Maybe even improve the day-fines scheme by using a Harberger scheme as a basis for that fine rate instead of income - your professed/estimated wealth gives your fine and income tax rate (e.g. $1b net-worth -> 90% income tax rate, $/€40000 day fine).

This also allows people to declare themselves as elite by increasing their tax rate to arbitrary heights and that would indeed be a true, because costly, signal of that status.

Why is there no gamer cooperative?

Consumers' co-operatives are enterprises owned by consumers and managed democratically which aim at fulfilling the needs and aspirations of their members. They operate within the market system, independently of the state, as a form of mutual aid, oriented toward service rather than pecuniary profit. - Wikipedia

In recent years investors and executives are tanking the value of the games people enjoy (to play, make and watch). They do that by focusing on profit and growth instead of value creation for the consumers and the maintenance of good products and services. And they force those priorities down the corporate hierarchy.

This is understandable from their point of view, but completely counterproductive for the consumers and creators of games:

  • Players get games that treat them like cash cows with a focus on monetization instead of gameplay and positive experiences.
  • The creators get shit thrown at them from both sides - the unhappiness of the players as well as pressure and low wages from the executives. All of that the creators endure because they love what they do. In the end this of course leads to burnouts and "economical" lay-offs.

So the question posed to everyone involved in actually creating and maintaining the game worlds as well as those inhabiting them - the players, developers, content creators, support personal - is:

How does one reduce the influence of investors and executives on the creator-consumer relationship?

The problem, as always, is organization and coordination. There are of course entities that were invented to tackle that problem.

And cue a slashdot link to a call of a union for game developers to unionize. Typical slashdot complaints are then about politicized unions. Really amusing are the argument about the unions then only being out for money, which is exactly the problem of corporations or any general entity only out for profit maximization.

Unions in the entertainment markets have a problem: The valuation in those markets is massively subjective and most people engaged in those trades would work without pay to just work on those products. Many of them also much prefer that work to the work of organizing a union.

And of course the standard survivorship bias that people only see the successful/lucky ones and imagine themselves to be those people instead of the myriad unsuccessful trials. These stories keep people from organizing and caring for each other:

"They deserve their bad circumstances, because they are lazy/stupid/do not want it enough." - Strawman argument

This is similar in all entertainment markets, where the evaluation is subjective and the creation itself is a joy to many people: Music, art, comics, books, sports, film and television. Maybe even science and education. Those would probably be much better served by consumer-worker cooperatives than the current consumer-corporation structures.

The problems of cooperatives are of course manifold:

  • The evaluation/prizing problem is still as hard as it has been all the time.
  • Participation varies strongly by the members and is rarely differentiated.
  • This leads to the problem of possession and reasonable renumeration.
  • Cooperation is expensive - The direct democracy/consensus principle has massive inefficiency with regards to "small problems" and delegation of votes is desperately needed for larger structures (more than 6 persons).

The last reason is probably the most difficult in practice - millenia of human civilization has only lead to our current system of nation states and corporations . Those work reasonably well, but of course their faults are also readily apparent. So the question is always how to improve what is there while preserving what is good?

But to get back to the topic of cooperatives: They shortcut the creator-consumer connection, increasing the possible efficiency there considerably. Of course the question of transaction prize and valuation is still open, but especially for games one could leave it open to the players. Some pay-what-you-want scheme with a minimum buy-in or even pay-per-playtime.

Something like that seems to be, an indie platform with pay-like-you-want and sell-as-you-like features. There is even some flexible revenue share to the platform itself. It's founded as a corporation instead of a cooperative...but well, everything works or it stops by itself after some time ^

Why is the fair coin toss considered fair?

If there is only blind chance - no actor doing that distribution - it seems that random results are considered acceptable. Often people are being told that "Life is not fair." This is an indication that we have a desire for fairness, but also accept it if circumstances on which nobody (no actor) has influence are random instead of fair. In social situations the concept of fairness seems to be different and strongly so, with many humans valuing fairness quite highly.

The ultimatum game in economics is an example where one sees massive discrepancies between the "rationally self-interested" optimal strategy and what is considered fair - And how much that is worth to people on average (hint:
a lot).

The ultimatum game is a two player game. The players get some amount of money which the first player splits among them. The second player can only accept or refuse the offer. In case of refusal both get nothing.

Rational self-interested actors are predicted to always accept any offer, because they can only gain by accepting.

On of the foundational experiments of Behavioral Economics is exactly this game. And the statement is more or less: "rationally self-interested agents" are not a good model for real humans.
Even when the possibility of refusal was removed, humans tend to distribute "fair". Remove the choice of the second player turns the game into the dictator game and removes any incentive from the "dictator" to be fair.

Real humans seem to have a concept of "fairness" - for which model agents would need an external valuation. In practice this valuation of "fairness" is high enough to lead to a refusal of most offers that are not about equal.

In a social environment one of the concepts underlying fairness seem to be: Equal pay for equal work. This is seen in nearly all social animals and could be the basis for the labor theory of value (LTV) - Which states that all labor, time at work, is equal and the labor infused into a product or services provides the limit around which value and prizes vary. This may be one basis of the fairness valuation in the ultimatum game, where there is no productive labor performed.

The LTV has its own problems, on which I will come to talk later. But I think the equality of work is a more central concept that may shed some light on higher concepts like "fairness" or "justice" and maybe even how those evolved.

Thanos did nothing wrong - except for his population modelling.

In the Marvel fictional universe the character Thanos has the idea the kill half of the life in the universe - by his own statement because there are too many of them, because everybody left would be better off. The overpopulation and then collapse of his own civilization on his planet of titan led him to this philosophy.

He proposes to do this by random choice, because he considers this most fair.

This proposal has several problems:

  1. A population at half its capacity is the most flexible - that is true for logistic growth with a fixed carrying capacity. The problem is that the population growth is also highest there. Which means that one soon has the same situation as before if one not also tackles the underlying problem of "How did one get to overpopulation?"
  2. There are a lot of schemes fairer than random. My favorite is allowing every person to gift its 50% chance of survival to another person.
  3. Why kill non-sentient life too? Those population will relax or die out anyway without anyone suffering for it.

Testing LaTeX

Trying out a bit more of the features, for example the LaTeX environment with the Euler-Lagrange equations for a one-dimensional system:

$$\delta S=\delta\int{t1}{t_2} L dt =0 \quad \Rightarrow \quad \frac{d}{dt}\frac{\partial L}{\partial \dot{q}}-\frac{\partial L}{\partial q}=0 $$

Update: Well, that does not look that nice -_-. Will have to remember trying that again some time later. Maybe I find some workaround.

What it would have been:

The right side states the Hamilton principle of stationary action - a system evolves in such a way, that small changes to its state have no effect on its trajectory (or at least an infinitesimally small one).

From this principle follow the equations of motion of the system as shown on the right side. Those are now "just" simple partial differential equation, which can be solved - sometimes analytically, more often just numerically.

Of course this does not answer the question how to formalize a system and what those equations then tell us about the development of the system. But it is nice to have at least a middle part that is comparatively straight forward.

First post

So, let me see what we can do here.

I searched quite some time for a tool for note-taking and note publishing and I would be very happy to have found something fitting to my tastes.

  • I tried to use the emacs org-mode before and still use it for smaller, local notes. But you need a lot of discipline and a reasonable work flow so that it actually helps with organizing things and note taking.
  • Used evernote for the last half year, but dislike their style and their constant nagging to try some new feature.

I absolutely adore the style of standard notes and listed. Simple, clean and without many frills. (Not such a fan of most of the dark themes, but to each their own.)

With Standard Notes I even saw some Note History and am now happy to have activated it. Hmm, it would be great if the note history packaged the modifications - similar to commits in version history - instead of remembering every single action.