fairness as a necessity

There is no such thing as an observer independent event. All the major breakthroughs in physics can be attributed to the slow and painful realisation of this fact.

Relational thinking is quiet tricky. (Even Einstein could not match Mach's relational ambitions!) We instinctively believe that we share a single unified reality. This of course is a necessary illusion since each one of us is confined to a single point of view. It is only when we try to switch points of view and insist on a single underlying reality do absurdities start to emerge.

Consider the notions of fairness and justice. These concepts come automatically with relational thinking. One is forced to listen to all sides of a story because there is no such thing as the story. There are only story-observer pairs. In other words, fairness is built into the very ontology of nature.

Perhaps the difficulty of getting our heads around relational thinking has got to do with the difficulty of being fair.

flow of culture

Culture flows from top to bottom, not from bottom to top. Here are some possible explanations:

  • Hiring-decisions are either made by the top echelon or made by criteria determined by the top echelon.
  • Once an organisation starts exhibiting a certain culture, applicants self-select themselves into it.
  • Once people start working in the organisation, there is a tremendous pressure to conform to the existing culture.

Culture hardens as it scales. Hence the reason why you do not see values like openness and multiculturalism in larger organisations which need more social glue to hold people together. (Think of Google throwing out the coder James Damore for releasing a conservative manifesto.)

If you are a member of an organisation that you do not culturally belong to, try to limit the information flow as much as possible. That will hopefully decrease the cognitive strain.

hierarchy and testosterone

Hierarchies select for testosterone heavy traits. In a world where only the high testosterone people can rise to the top, decisions will be testosterone driven. Hence, if you want to make the world a little less aggressive place, you should start by making the internal structure of the decision making entities less hierarchical. But how do you proceed?

  • Keeping the size of the entities small is one option. But competition and scale effects favour consolidation. Hence this will not work out in any sensible economic regime.
  • Trying out non-hierarchical organisational structures like holacracy is another option. But these flat fantasies never last too long. None of the large entities can even hope to give them a try.
  • Waiting for artificial intelligence to mature seems to be the most feasible option at the moment. AI will dramatically decrease the need for human decision making so that even the largest entities can be run like a small entity.

math as sensitisation

In the first iteration of the study, he and the team had started with a totally naive neural network. But they found that if they began with a neural network that had already been trained to recognise some unrelated feature (dogs versus cats, say) it learned faster and better. Perhaps our brains function similarly. Those mind-numbing exercises in high school—factoring polynomials, conjugating verbs, memorising the periodic table—were possibly the opposite: mind-sensitising.

- AI vs. MD (Siddhartha Mukherjee)

I had always suspected that mathematics increases general mental sharpness. Since it is the most rigorous of all academic disciplines, it should also be the most "mind-sensitising" one in Siddhartha's sense. This creates a pragmatic ground for arguing in favor of making abstract mathematics a mandatory part of public education.

Personally speaking, my most challenging intellectual journey involved a deep understanding of category theory. The inhuman level of abstraction caused me headaches. Now, looking back at my experience, I think that pain was literally the pain of adding new layers on top of the already existing layers in my (neocortex) neural networks. I have probably been using those additional abstraction layers ever since, in all areas of my life.

mastery and audience

One gets better over time as one masters a subject. Then how come the best artists' best albums are by far their first ones? Here are two possible explanations:

  • An artist spends years preparing for the first album whose release immediately forges a large fan base which in turn demands a shorter production cycle.
  • The truly talented artist becomes famous very quickly. The sudden shock unbalances his psychology, damages his work and creates a time pressure to cash in on the fame while it lasts.

Lesson: The road to mastery should have no audience.

iyi polis kötü polis

İyi polis kötü polisi oynayabilmek için minimum iki kişi gerekir. Dolayısıyla iki kişilik bir yönetimin tek kişilik bir yönetime karşı müthiş bir avantajı vardır.*

Tüm öğrenme süreçleri hem teşvik hem de tenkit gerektirir. Şımarıklığa yol açmamak için teşviğin tenkitle, ümitsizliğe yol açmamak içinse tenkitin teşvikle dengelenmesi şarttır. Teşvik eden kişinin aynı zamanda tenkit eden kişi olduğu durumlar kafa karışıklığı ve kredibilite kaybı doğurabileceği için iyi polis ve kötü polisin ayrı kişiler olması önemlidir.

Tek öğretmenli sınıflarda öğrenciler, tek ebeveynli ailelerde çocuklar, tek ortaklı şirketlerde çalışanlar komplikasyonlu bir öğrenim sürecinden geçerler.

* Genelde ikiden üçe geçmenin faydası birden ikiye geçmenin faydasından çok daha azdır. Mesela iki dil bilmenin bilişsel gelişime önemli katkıları vardır, fakat üç dil bilmenin çok da ek bir getirisi yoktur. Benzer şekilde, başka bir ülkede yaşamaya başladığınızda, akvaryumdan çıkmış gibi aydınlanır ve kendi kültürünüzü daha iyi tanırsınız. Üçüncü bir kültürel şok aynı etkiyi yaratmaz.

İlgili yazılar: Mystery of Two and Three, Two Opposite Lives

geometry of wealth

Ownership of space has always been the paradigmatic way of displaying wealth. (See the previous post exploring the relationship between power and void.)

In modern times the geometry of status spaces has changed thanks to the development of various enabling technologies that led to greater concentrations of people living in cities. 

Previously, the wealthy owned huge swaths of land. Now they own skyscrapers. This simple change of coordinates created impressive perceptional differences:

  • Horizontal spaces could only be experienced tangentially by the outsiders and therefore looked small relative to their actual sizes. Vertical spaces on the other hand are completely exposed to the gazing eyes, revealing the full glory of wealth.
  • Now the poor literally look up to the wealthy. What was once an abstract status hierarchy has acquired a physical form as well.

machine learning revolution

Take agriculture, for example. In agriculture you plant seeds, you fertilize them, you spray them, and then you wait six months and hope for the best. Once upon a time, people really understood the land, they understood the plants, and they had this intuitive feel for what would happen. Then farms got bigger and there were fewer people on them. We didn't have the ability to walk the land, and feel it, and close the loop. In the absence of information, we had to resort to monoculture. We closed the loop on an industrial scale. We went to predictability: If I use this Monsanto seed, spray it with this chemical, in this very well understood soil, I'm likely to get a pretty good outcome. In the absence of information, we went with predictability, simplicity, and a lot of chemicals.

- Closing the Loop (Chris Anderson)

We literally rationalised the shit out of the environment after we successfully formalised the inner workings of the consciousness. (We basically got super excited once we found out how to exert scalable control over the nature.)

Now we are living through another massive revolution. This time we have discovered the inner workings of the unconsciousness and it is a lot better at dealing with the sort of complexities and non-linearities exhibited by ecosystems. Intuitive control has finally become scalable.

The worst is behind us. Our relationship with the environment will slowly recover. Machine learning will allow control structures to stay beautiful and organic even as they scale.

Artificial intelligence (AI) is a misnomer. Intelligence is a notion we associate with consciousness (which works rationally / analytically), not unconsciousness (which works intuitively / statistically.)

This is not surprising, since the actual historical evolution of the idea of AI started off as an analytical project and ended up as a statistical one. (Analytical approach to automate human-level intelligence failed.)

This too is not surprising, since nature itself too invented unconsciousness first. (In other words, the straight-forward looking consciousness has to be deeper and more complex than unconsciousness.)

Side Note: Here, by unconsciousness, I am referring to the intuitional part of the reasoning mind, not to the lower level older brain parts that are responsible for things like the automation of bodily functions. Hence the remarks are not in contradiction with Moravec’s Paradox.

Notice that there is a strange alteration going on here. Human reason was built on top of human intuition. Now machine intuition is being built on top of (advanced) human reason. In the future, machine reason will be built on top of machine intuition.

The reason why our unconscious mind copes better with complexity has to do with its greater capacity to simultaneously deal with many variables. Rational mind on the other hand extracts heuristics and stories out of patterns. This process results in a drastic reduction of number of variables which in turn results in a loss of accuracy.

Take, for example, the problem of predicting the distribution of trees in a forest using only map data, such as elevation, slope, sunlight, and shade. Jock Blackard and Denis Dean did the original calculations in 1999, and left behind an enormous public database for other mathematicians to use. According to Vapnik, when computers were trained using 15,000 examples, their predictions were 85 to 87 percent accurate. Pretty good. But when they were fed more than 500,000 examples, they developed more complex rules for tree distribution and boosted the accuracy of their predictions to more than 98 percent.

“This means that a good decision rule is not a simple one, it cannot be described by a very few parameters,” Vapnik said. In fact, he argues that using many weak predictors will always be more accurate than using a few strong ones.

- Teaching Me Softly (Alan S. Brown)

What distinguishes excellent service (and maintenance) from very good service (and maintenance) is anticipatory power. Machine learning techniques are increasing our predictive capacity. We will soon be able to fix problems before they even occur. All problems, including those related to customer unhappiness and mechanical breakdowns, exhibit early signs which can be picked up by machine learning frameworks.

Life is about to become significantly happier and safer for our children.

monolithic monopolies

Artists are angry at streaming services for turning into monopolies, but what they are really angry at are their decreasing profit margins. Streaming services like Spotify are not monopolies. They are utility apps that are finding it increasingly harder to distinguish themselves. There is an intense competition and that is precisely why they (and in turn the artists) have low profit margins.

The real monopoly is Apple who owns the platform which hosts all the music platforms which in turn host all the artists. Two years ago Apple launched its own streaming service, severely undercutting the incumbents' prices. It could afford to do so because any move increasing their hardware sales justifies itself due to the fat profit margins there.

Nested market structures do not lead to polylithic monopolies, they lead to monolithic monopolies where each level is dominated by the same entity which dominates the lowest level.

teaching and expertise

Measuring quality of education is very difficult. University rankings are biased towards research rather than teaching. They mainly focus on the easily measurable parameters like the number of papers, citations etc.

Harvard, for instance, is a horrible place to get educated. When I studied there, my best teachers were the teacher assistants. Professors who are hired on the basis of their research skills generally do not make good teachers:

  • Researchers are not interested in teaching. They often see it as a burden. 
  • Researchers are cut off from the foundational material. They spend most of their time doing technical stuff that only a PhD level student can hope to understand.
  • Researchers are quiet competitive. They generally do not care about others' learning difficulties and do not want to remember their own neither. They just want to march on.

On a related note, I think that all textbooks about the foundational subjects should be co-authored by young PhD students, not old professors.

“It often happens that two schoolboys can solve difficulties in their work for one another better than the master can. […] The fellow-pupil can help more than the master because he knows less. The difficulty we want him to explain is one he has recently met. The expert met it so long ago he has forgotten. He sees the whole subject, by now, in a different light that he cannot conceive what is really troubling the pupil; he sees a dozen other difficulties which ought to be troubling him but aren’t.”

- C. S. Lewis