Only coerced labor can keep humans accountable

There is a theory about how people would just do drugs, have sex, and do no good in general if they did not need to be accountable by their jobs and the pressure of society to keep them under control. Note: not that doing drugs and having sex are bad things, but they are only good (for individuals and society) when in moderation and under control.

I have always thought I was not that kind of person. I have always thought that if I had some basic income, for example, then I would spend my days doing free software, publishing some research, and also having a healthier life: exercise every day, respecting sleep hours, enjoying more time with family and friends, etc. But I found myself wasting most of the time that I am not working.

A man who dares to waste one hour of time has not discovered the value of life.

Charles Darwin

Back to the point, certainly a job is not a sufficient condition to keep humans accountable, if that was so, work addiction would not be a problem. All sorts of volunteers, NGOs, parents,… prove that it is not a necessary condition either. A fulfilling life with a purpose is a much better way to keep humans accountable. Sometimes it is not about what you want from life, but what “life” wants from you, and caring about that.

Therefore, the theory mentioned in the first paragraph is clearly wrong. It is a quite harmful model, and something to forget. At this point in time, I do not think that life wants much from me. But I know that I do not want to be the kind of person described in the theory. Therefore, for various reasons, I am going to start to write some blog posts. They are going to be short and fairly crappy, to be honest, but they are better than nothing. This may be one of those rare cases of information published on the Internet where only the author reads it, and I am fine with that.

But more importantly, this may be one of those rare cases where a person (or group of them) can prove a theory wrong, intentionally. So if you say: “people will not do anything useful unless you pay them”, I respond: “you cannot stop me”.

To try to give a take-home message: try to do something that you genuinely enjoy and that you think that it is valuable, because that may be what gives you energy (motivation, not stamina), a sense of purpose in your life. Some things provide instant gratification (like Twitter), some things take a bit longer (like WordPress), and some much longer (like GitHub). I am going to try to train myself to get more and deeper gratification by doing gradually harder and more useful things. Maybe you should try to do the same thing.



FMA:B, science, philosophy, value, and bullshit

tl; dr: FMA:B does not criticize science, nor scientism, it criticizes taking something that is not deserved, the greatest sin committed by the homunculus. The core of the series is truth and value. A similar sin would be proselytism on ideas worth nothing or with a negative value, bullshit vending, which seems to be the main occupation of too many philosophers nowadays. Be skeptical on ideas, and loving on people, nothing is more valuable, as people are the ends that provide value to any means.

FullMetal Alchemist: Brotherhood may be my favorite series of all time, e.g. I took my Twitter avatar from it. [Spoilers in the whole post]. The series explains how wrong the concept of “human resources” is as well as concepts like anti-fragility in simple and beautiful ways.

As a consequence of my reflection on that last link, I was considering to go (in some far future, i.e. never) through the series commenting each episode or scene, and the “truth behind the truth” in each of them. I saw much truth and value in the series, maybe because there are apt and consistent metaphors for real phenomena, or maybe because of my interpretation. To find out which one of the two could be the case, I decided to check what other people had done. I found an analysis on the eyes and their meaning, quite right and partial (to the eyes part) IMHO, and an analysis of the philosophy in the series, and my reply to it as a comment became so long that it is now this blog post [probably only] you are reading.

There are two problems with the latter video:

  • The lesser problem: it does not accurately describe the series, numerous references to Feyerabend clearly indicate an epistemological anarchism bias, opposed with the series.
  • The greater problem: it masks criticism towards science through scientism from epistemological anarchism. While verificationism has great flaws, they have been interpreted as a blank cheque by all kinds of bullshit vendors to promote relativism and get rid of the so uncomfortable truth.

In the remainder of this post, I will go through both of them.

About the series

There are two main concepts in the series: truth and value. The main truth is the law of equivalent exchange (especially highlighted in the previous series), and Truth is even a character of the series.

The law of equivalent exchange resembles conservation of mass and conservation of energy laws, but it refers to value, it is therefore directly linked to economics. Equivalent exchange means zero sum games, breaking it as the homunculus means negative sum, and the final proposal from Alphonse Elric is the creation of positive sum games. Red, blue, (and possibly green) are associated with this meaning in general culture, like the Western stock markets, the Jedi lightsabers, and the colors of transmutation energy.

The only way to meet Truth is through the portal of Truth, composed by the knowledge in alchemy. This frontally opposes to epistemological anarchism. In the end, Edward Elric exchanges his portal for his brother, this does not mean that there are alternative paths to reach to the truth, but that there are some things that are more valuable for Edward Elric than truth and alchemy, e.g. his brother. Friendship being more valuable than knowledge, and similarly medicine (main focus of von Hohenheim) being more valuable than epistemology (main focus of Father).

The philosopher stones allow “ignoring” the law of equivalent exchange. In fact, they are exchanging human lives for something. This is something that science does not usually do, but economy exchanges time for money (is that equivalent exchange?) and gets things done through the use of men-month. It is very clear in the reference of humans as resources, something that does not concern science, but is the core to economy, capitalism, and businesses. Similarly, Tucker is not pushed by science, but by economy.

Godwin is right in this one, though, and coincidentally the greatest force stopping the Third Reich was Communism (on the other side of Briggs). Nevertheless, associating Nazism with scientism is a bit shallow. Nazism and wars in general are often associated to economic causes, like hyperinflation, and the oil that we would find in Ishbal if it was not for the location for the transmutation circle. Father even has the monopoly of alchemy in Central, being able to block it.

By focusing so much on truth and value, the series focus mostly on axiology, ethics, and economy. Criticizing faith is a bit out of place when God (as Truth) appears in the second episode and then some more. What we find is mostly a critique of bullshit vendors that will get something valuable from people (their time, effort, money, or something else) in exchange for something delusional in nature. Nevertheless, when asking for a philosophical perspective, it is easy to find some criticism to science and any other competing epistemological approach that there could be, more on that in the remainder of the post.

About science and philosophy

Mad people committing atrocities in the name of science is not what science is. Epistemological anarchism is an easy position to defend, as it is solipsism, nihilism, and in general any denial. We can also say that, from the perspective of pragmatism or functionalism, lies may work just as well as truths. But in the end, there can only be one truth, and we try our knowledge to get as close as possible to such truth, then falsifiability, verifiability, and in general any check of the verisimilitude of a particular claim is required to accept something as true. Doing otherwise means failing, and while different speculative bubbles may last for different periods of time, the price of errors does similarly increase over time. It is not possible to escape the truth for ever, but it is possible to deny it for long periods of time, always with awful consequences.

There are some things that we do not know how to measure or observe (directly or indirectly), but we can do so and we must strive to do so, as the only things that cannot be observed indirectly have no effect in our world, and either they do not exist or are irrelevant, innocuous.

There are many problems in the knowledge acquired with science, mostly because it relies on humans, errare humanum est. But science is more than a body of knowledge, it is a mindset, and a very positive one. No other discipline has achieved such a great success and improved human life quality as much as science. While science has been corrupted by capitalism, this is not a problem of science itself, but of lack of science to apply in areas like economy, ethics, axiology, game theory, politics, and utility theory.

The scientific method has been defined in such rigid ways that not even scientists follow it. Many variations and new methods emerged to deal with problems in more scientific ways, and they should be recognized as such. Ultimately, it is a form of reinforcement learning, that tries to make predictions about observable outcomes, if the predictions are right, then we say that it works; and if they do not, then we have to revise the theory or model. Even if all models are wrong, some are useful, and not all are equally useful. For example, consider a person that is blind. Philosophy may help that person to accept it and deal with it, while science may help that person to see. Among the two options, I think that one is infinitely more useful than the other.

Finally, nothing among what is relevant and philosophy or any other discipline does cannot be done with science. The only property that philosophy has over science is the capability of accumulating bullshit that does not go through the hard scrutiny and skepticism of science, i.e. discuss what is not observable and verifiable. That is why science works, and others cannot say the same. In fact, I would go as far as to assert that, if it works, then it is science, because being able to observe that it works (so as to assert it) is enough to apply science to its study, and check how it works, when it works, and in which ways it works best.

PS: In retrospective, after spending so many hours to write this post, I will certainly not comment on each episode separately. In addition, I plan on sticking with 500 words posts. There are many points that could have been elaborated further, but that would mean doing three or more posts. This is just too much, for me and for you (someone got this far? Please comment to confirm).

Make neural networks better, please

1. Add noise in the training to avoid overfitting

One common problem in deep learning, artificial neural networks (ANNs) and in general machine learning is overfitting. Due to overfitting, machine learning algorithms focus on a few specific features of the data and do not generalise well to new cases of the data.

As a consequence, by analysing how the ANN works, it is feasible to trick it into mistakes. Having some noise that precisely makes the ANN mistake a panda for a vulture is unlikely statistically, but shows that they can make big stupid mistakes. If ANNs are used more and more often, then this kind of mistakes becomes more and more statistically probable, and eventually happens more and more often.

A simple solution is to add noise in the training set, so that ANNs are more resilient to noise and less prone to overfitting. Noise can come in many different varieties, from white noise, to anything else. In the particular case of ANNs used for image processing, due to the spatial component of the dimensions, different types of distortions can be applied, e.g. lens distortions. This is not common, AFAIK.

2. Analyse, synthesise, generalise

Currently, ANNs do some analysis of the data, even if implicit in their adjustment of the weights. They are very specific, hence the overfitting.

Approaches like self-organising maps allow synthesising what has been learned from the examples, and then make the ANNs more simple, by removing neurons and connections that perhaps were not so useful. This is particular relevant for the optimisation, as deep learning models can become very complex and computationally expensive.

Self-organising maps seem to be more concerned with adding new neurons and connections, though. If this is used for the analysis then nothing good is going to result from that, but a worse overfitting. This capability of making an ANN more complex has to be used to generalise the network to more cases and more complex problems.

3. Learn from few examples

AlphaGo could defeat Lee Sedol in a quite consistent way (four out of five games). On the one hand Lee Sedol learned a lot from those games, especially the four he lost, but probably from all five. On the other hand, the learning that AlphaGo performed on those five games is probably negligible, and it is very likely that it did not learn anything at all in the games it won.

AlphaGo learned from millions of games before facing Lee Sedol, more games than humans could ever play in their lives (or would probably want to play, given the chance to live just for that). Five games are not going to make a big difference, and probably they should not after the extensive experience. While this has a positive side in our chances to defeat Skynet, it is disappointing when we expect AI to do more complex and general things.

There are approaches that use fewer training cases and learn more from them, support vector machines and case based reasoning being two of them. In the end, connectionists and deep learning experts may find something useful in other AI approaches, if they are open for collaboration.

How to make a safe self-improving AI

Philosophy might be the key that unlocks artificial intelligence, but I would say that we should not look at epistemology, but axiology. We can pseudo-mathematically define axiology as a function that maps outcomes with a value (a number) depending on the values of the people (what they appreciate).

First, we do not want to encode axiology in a set of rules, we want machine learning to learn from humans whatever our axiology is, currently, and keep it updated. The sum of axiologies in every human will be inconsistent, but should be easy for the machine to have a good grasp of the most crucial parts, which will be most probably quite commonly agreed. This is not so different from the current state of the art, e.g. sentiment analysis. The artificial intelligence does not need to take a position in most controversial topics, and we do not want that. If it did, how safe the AI is would be equally controversial.

Once that we have the axiology module ready, we can use it as a supervisor for an otherwise unsupervised machine learning algorithm. The machine learning algorithm can get incredibly complex, involving image processing and information from all kinds of sensors. In general, anything mapping inputs and outputs using any of the machine learning technologies that we have currently available or some better future ones. The axiology module keeps learning and providing feedback to the main machine learning module about how good or bad it is performing. Safely and ad infinitum, for any new outcome.

Certainly there are other aspects to consider in the architecture. Epistemology could be useful to guide the artificial curiosity of the AI and to make it more efficient. Please note that artificial curiosity is not safe per se, as the AI could decide that experimenting with humans provides a lot of interesting information. Axiological considerations are crucial in this context. Similarly, we could provide a better starting point to the seed AI, e.g. simulate before performing an action, for faster and safer learning.

There are also many aspects to consider in each one of the modules, especially in the main machine learning module. In all likelihood it should display a form of algorithmic universal intelligence. But the state of the art is a bit further from that than the previous example of sentiment analysis, time will tell whether theory is the tip of the spear or practice goes before as serendipity. There is some work to do before reaching to the “master algorithm“, but we might be closer than it seems.

To sum up and conclude, the key is the role of axiology, something that I have not seen mentioned before. I finish with this conclusion, after all this is a blog post and I try to keep them below 500 words. Let me know if you are interested, especially if you have funding and could provide me a salary to elaborate more on this. I am looking forward to work on such future lines.

Perceived reality

The most reputed scientists (and even worse, philosophers) decided that the KPI for the epistemic endeavour of the humanity should be reputation impact, i.e. visibility, which has become part of the bias of meta-research. No wonder that the field it is now full of posers, impostors and attention whores. This doesn’t play well with rigorous research, which may be overlooked in favor of some research with more impressive results, for the greater impact.

It may be the case that, as societies become more complex, what other people think may become more relevant for the well being (i.e. salary) of the person than actually making something. As a consequence, people strive to become competent at politics (i.e. managing their image), because nobody will be able to judge the work done. That why people could technically work from home but so few are allowed to do so, because of sheer incompetence in actually judging the difficulty of the work done and the quality of the solution.

Eventually, the work will be done by some underpaid interns, because quality cannot be measured, and being out of the metrics nobody in management cares about it (it does not exist in their Powerpoint presentations). The interns will comply with this underpaid work with a smile because they want to forge their image of competent friendly sympathetic people, so that they can move on. No reason to worry about that, eventually this work will be done by artificial neural networks, I mean, the real work, not social interactions. Then people will be able to focus on these social interactions. In fact, it doesn’t matter if some are not good at social interactions either, the point of more and more jobs is just to keep people entertained to prevent revolutions, so they are boring and repetitive, the way they are meant to be. People that comply with the status quo to a greater extent are the ones rewarded, especially the ones that feed economic bubbles of fictional value completely disconnected with reality.

This is perfectly exemplified by actual politicians. Most people voting cannot understand the kind of problems involved in managing a country, neither can they decide which candidate is proposing the best solutions. They don’t vote to whoever may really be the best candidate, but whoever looks like the best option. Promising something impossible eases the way to the government more than being realistic in the expectations. People just want to be deluded, unconsciously, yet they will always say that they value honesty over everything else.

At this point in time you may be wondering what is the take-home message. Simply put: The revolutionary idea that some rants may be right. They are usually overlooked, misheard, paid no attention. It’s not pleasurable to read unexciting research, but that may be the most trustworthy research. We may not like negative political discourses, but those may be the only ones that are honest and realistic. Finally, about the work and the workplace, distrust friendly people, let the work of each one talk by itself, if you think that it does matter, if something needs to be done, if something matters. This is not epistemology but axiology, and a matter for another post.

What is the task of all higher education?

From a doctoral examination. — “What is the task of all higher education?” To turn men into machines. “What are the means?” Man must learn to be bored. “How is that accomplished?” By means of the concept of duty. “Who serves as the model?” The philologist: he teaches grinding. “Who is the perfect man?” The civil servant. “Which philosophy offers the highest formula for the civil servant?” Kant’s: the civil servant as a thing-in-itself, raised up to be judge over the civil servant as phenomenon.

Friedrich Nietzsche, Twilight of the Idols, or, How to Philosophize with a Hammer

Starting the blog

Sometimes I’ll feel like saying something too long for a tweet, or maybe not obvious enough from just a bunch of links with no further explanation (but they will always be obvious to a great extent). I’ll then post it here and reasonably expect nobody to read it, the posts will not be particularly optimistic, you will not be happier after reading the blog, thus it is unlikely that

The posts will be short, so that I do not spend too much time writing them and you don’t waste too much time reading them. I don’t have great aspirations of pretensions for this blog, but to write something and get it out of my head. The topic may include everything, but will probably focus on information processing, from artificial intelligence to cognitive biases, including emergent behaviors from natural and artificial laws, rules, programs, etc. Someone is giving a good use to, hence the narcissistic address.

Posts will probably contain many links. There are several million caffeinated apes spending many hours typing in non completely random ways, and most know more than me about what they type, therefore I find few reasons to waste resources in the cloud (with all that is implied by that) adding worse contents. Try to read one post or two to see if any of this makes any sense.