On freedom of education and freedom of information

From supermemo.guru
(Redirected from Freedom of education)
Jump to navigation Jump to search

This text is part of: "I would never send my kids to school" by Piotr Wozniak (2017)

Freedom to learn

Full educational freedom implies freedom to choose one's own way of thinking about the world. It implies one's own choice of sources of information.

Those who believe freedom is dangerous for the integrity of democratic society make the same mistake as those who claim that social media undermine the wisdom of the masses, or that, in the most extreme case, "Internet makes us stupid".

All those claims come from the wrong interpretation of the evolution of knowledge.

One truth

The whole human knowledge and our perception of reality rely on basic axioms:

  1. some of the information coming in via our senses is true, and
  2. there is one truth.

Information science does the rest: false information is eliminated by statistical confrontation with the body of knowledge. The brain forms models using simple neural network generalizations. Those can get distorted by cognitive biases and need to be worked on with the tools of science to verify their validity, and to sift away the impact of human error and emotion. Scientific models are formulated using pragmatic Occam's razor principle.

Willy nilly we need to assume that human knowledge converges upon some approximation of true reality. There is no point in arguing those basic axioms as we have no other alternative. Approximation of the truth is a convergent process that occurs in any dynamic body of knowledge. It occurs in our heads as we grow up and try to make sense of the world. It occurs in societies. It occurs in science. It occurs in expert systems empowered with probabilistic assertions.

In the chapter on self-directed learning, I try to show how people tend to converge on similar themes and memes in history. It does not take long for a student to hit the immortal memes of pyramids, Guttenberg, Napoleon, Hiroshima or Hitler. Here comes the origin of Argumentum Ad Hitlerum in a loose discourse.

The same convergence occurs in the whole spectrum of human knowledge. Diversification is healthy. Like pieces of knowledge attracted to individual semantic tree leaves, people of different skills fit the jigsaw puzzle of the economy. This is a galactic process of information self-organization that propels mankind to new heights.

The evolution of knowledge is a powerful and positive process, which is accelerated by access to information and communication. It has been accelerated by language, written records, printing press, telegraph, telephone, radio, television, mass media, web, and now social media. At each acceleration stage there are skeptics who see and bemourn the decline of humanity, trivialization of knowledge, dumbification of the masses, distortions of reality, fake news, etc. As much as the biological evolution leads to indeterminate outcomes, so does the evolution of knowledge (see: Value of wrong models). However, statistically, the process is convergent and our approximation of reality, given the original axioms, is getting better and better.

Evolution of knowledge is convergent onto an ever improving approximation of reality. Accelerated dissemination of information, accelerates the convergence of knowledge

Books were considered seditious. Press wars sparked otherworldly hoaxes and yellow journalism. Telegraph was seen as a likely channel for gossip. Telephone was to destroy human personal interaction. Radio was to be the means of propaganda. Television was to promote lewdness. Mass media was to dumb down the masses with tabloid press and mindless entertainment. It has always been the same problem: popularity sparks vulgarity. Web was to flood us with infobabble. These days, after a short romance with the democratization effect of the Arab Spring, more and more people see social media as a channel for fake news again. The yearning for the reinstitution of the one and only truth is the domain of well-schooled brains (see: 50 bad habits learned at school). This yearning leads to new calls for censorship, political correctness and even the denunciation of digital media (see: Digital Dementia).

Internet makes us stupid

Nicholas Carr is one of the most prominent journalists complaining that "the Internet makes us stupid!". Carr is wrong. The exact opposite is true (see: proof).

It is quite natural for a journalist with a flair for "deep thinking" to bemourn fake news, and how it competes with well-polished fact-checked intellectual journalism. This, however, should not obscure the fact that all modern communication should be seen as progress. While there are horrifyingly wrong processes occurring in their wake, new media makes us smarter and more knowledgeable, on average.

For starters, Carr claims that Google cannot be used as a substitute for knowledge in our heads. This is largely correct. SuperMemo had to battle the myth of Googleable knowledge ever since its inception. Google is no substitution for associative memory. We can free our memories to some degree, but core knowledge that underlies reasoning must reside in human memory or it loses its power.

Carr claims that multitasking takes away from "deep thinking". "Deep thinking" is a hazy concept. I would like to see it defined more precisely in terms of information processing. Here I can only use it intuitively. In his reasoning, Carr seems to be forming a model of information processing typical for a teenager who hangs around with his smartphone, keeps multitasking between various social media services, sending messages, gossiping, collecting likes and friends, uploading instant photos, and the like. The problem is that the same teen, without the phone, might be just hanging around with friends sipping beer or daydreaming about that gorgeous cheerleader. In that sense, time-wasting and inefficient multitasking is a step ahead. The same teen, instead of paying attention to his teacher of biology, might be reading recent gossip news abouts his favorite celebrity. Here he is likely to discard incomprehensive knowledge of biology that does not satisfy his learn drive for scraps of world knowledge smuggled even in the least ambitious tabloid stories. This is again a step forward, on average. It is better to learn from tabloids than to let incomprehensible knowledge float unresisted through inattentive mind.

Carr seems to ignore millions of people who use social medial for doing actual work, getting valuable news updates, exchanging valuable information, and the like. The smarter they are, the less they multitask, and the smarter they get. Bemourning Facebook or Twitter is akin to complaining about the existence of cars because accidents kill people.

A huge wave of disappointment with the alleged decline of human knowledge and human intelligence came with a recent swing to the right, or to the left in recent elections, referenda, or polls of electoral prospects around the world.

Smarter sum of dumber usership

Brexit and Trump do not mean masses get dumber! Just the opposite. Masses get smarter. The average knowledge of an average Internet user might have dropped indeed. This is simply because of the fact that those with lesser smarts or lesser education have joined the bandwagon. They still get smarter because of their online access. For each stupid online falsehood, they get two corrections in terms of the truth. Wikipedia is the core of the net knowledge today. It contains lots of falsehoods and simplifications, but it is getting better by the minute. It is the shiniest example of the wisdom of the crowds. Most people have heard of Wikipedia. More and more people use it. We are converging on the truth. The size of available knowledge grows fast. The news is good. We should not fear biased sectarian homeschoolers. The are bound to be far more open to the world than it was a case a decade ago.

If there is knowledge convergence, you may ask, why does research show that political debates get hotter in proportion to the degree of well-informedness? The heat of the debate follows a U-shaped curve. For zero knowledge, there is no debate. For complete knowledge, there is a perfect agreement. For complex knowledge, like "Is Brexit good?", we are all closer to the point of ignorance. Adding information is like adding fuel to the fire. The sides only tighten their ranks and turn livid.

If you Google for Brexit or Trump, relevant Wikipedia entries will come on top. This is knowledge that even the smartest political pundit may not be entirely familiar with. This is knowledge that most people would and should gladly take on board and feel smarter. It is reasonably balanced between the points of view of all sides. It is far better a compilation than the models of reality most of people carry in their heads. Carr would probably deny it. After all, he is on Britannica's board of advisors. With all good intentions and a great deal of valid argument, Carr is an information Luddite of the web era. In the world of fast progress, few people can afford to finish reading Anna Karenina. The book has not lost its value. It is just the choice that got wider and richer.

As a contributor to Britannica, Carr is a writer with a point of view. He contributes on subjects that are affected by his point of view. Analogous topics at Wikipedia are a result of crowd powered balance. Richness of Wikipedia is its main strength. To a critical mind, there is more inspiration in 9 pieces of correct information and 1 piece of wrong information than in 2 pieces of correct information. Wikipedia is a great starting point when learning about a new subject. Moreover, Wikipedia is now rich in citations for anyone to verify the source of most weighty claims. It is up to the reader if he makes a good use of the information, or get fooled. Britannica may indeed be better for people with lower ability to critically process information. However, Wikipedia's accuracy keeps increasing. Today, it seems unbeatable.

Wikipedia vs. Britannica

The battle over the superiority of Britannica is nearly two decades old. Today the verdict is pretty easy to settle. Here are two definitions of sleep from Wikipedia and Encyclopedia Britannica. Which one is better in your view?

Source 1: Sleep, a normal, reversible, recurrent state of reduced responsiveness to external stimulation that is accompanied by complex and predictable changes in physiology. These changes include coordinated, spontaneous, and internally generated brain activity as well as fluctuations in hormone levels and relaxation of musculature. A succinctly defined specific purpose of sleep remains unclear, but that is partly because sleep is a dynamic state that influences all physiology, rather than an individual organ or other isolated physical system. Sleep contrasts with wakefulness, in which state there is an enhanced potential for sensitivity and an efficient responsiveness to external stimuli. The sleep-wakefulness alternation is the most-striking manifestation in higher vertebrates of the more-general phenomenon of periodicity in the activity or responsivity of living tissue.

Perhaps this one is better?

Source 2: Sleep is a naturally recurring state of mind and body characterized by altered consciousness, relatively inhibited sensory activity, inhibition of nearly all voluntary muscles, and reduced interactions with surroundings. It is distinguished from wakefulness by a decreased ability to react to stimuli, but is more easily reversed than the state of hibernation or of being comatose. Mammalian sleep occurs in repeating periods, in which the body alternates between two highly distinct modes known as non-REM and REM sleep. REM stands for "rapid eye movement" but involves many other aspects including virtual paralysis of the body

Do you have a verdict? Hard to say? Google it out and see, which is which. By that time, one of those will probably have improved, the other will likely stay frozen, so you may need two Google tries.

If this is a close call, go to the chapter on the function of sleep. Here Britannica lags by two decades. It does not include notes on the newest research and is very conservative as to what scientists believe the function of sleep is. Sleep is a form of neural optimization. This seemed obvious at the end of the last millennium. In 2017, Britannica misses the point entirely with a shy note in the last short paragraph that reduces optimization to mere consolidation: "experiments have pointed to sleep as playing an essential role in the modification of memories, particularly in making them stronger". Wikipedia, in the meantime, adds a big chapter with references and links to three independent wiki articles on memory, learning and creativity in sleep.

Back in 2001, I wrote a few paragraphs for Wikipedia on neural optimization in sleep. The crowd voted me down, so I had to give it up. I was too progressive for the times. Even the open mind of Larry Sanger would cross-check with Encarta and remain skeptical. It took a while for Wikipedia to catch up with research. In the meantime, Britannica and Encarta stayed decades behind, and Encarta actually died of its own inertia.

For me there is no contest. Wikipedia is already light years ahead of Britannica and this is only going to get worse for the venerable pioneer of printed encyclopedias. If you care about learning, research, or inspiration, you cannot possibly disagree.

Isolationist homeschooling

A great deal of motivation for homeschooling around the world is to shield the kids from the evils of the world. For some religious fundamentalists, even truthful science carries a great deal of evil knowledge. Ubiquitous Internet makes shielding harder. This is particularly true of shielding with half-truths. It took one book on evolution at the age of 10 to get me out of the class in religion, and out of my Catholic faith for good. Today, all human knowledge is just a few clicks away. Parents who wish to retain their kids' purity, not only need to prevent schooling, they need to prevent access to the Internet. This may still be the case in select communities or select families, but it is getting harder and harder. For more see: Ban on homeschooling

Fake news and TrustRank

Work is underway to produce crowd-sourced databases of fake and untrustworthy sites. The database could be used in user alerts or prioritization of Facebook feeds. There is a more attractive tool in prospect though.

I predicted the emergence of TruthRank long before I wrote about it in 2002. Google has already announced they are working over a concept named TrustRank. Even Ray Kurzweil is said to be on the effort (Google’s Director of Engineering). TrustRank is probably a better name than TruthRank. After all, trustworthy sites may publish untruthful information by mistake or before a hypothesis is proven false. Interestingly, knowledge-based trust correlates with PageRank. Climate change deniers, antivaccination activists, and creationists alike, shake in their boots already. So should future Trumps and promoters of future Brexits.

The truth is as good as information available. TrustRank might penalize those who refused to believe in weapons of mass destruction in Iraq 2003. They might also penalize those who claimed Trump is a favorite in the US Election 2016. Such biases are inevitable. If intelligence is wrong, or polls are wrong, we may all fail to see the light. What TrustRank will do is to weaken verifiably false memes.

There will always be funding bias, sponsorship bias, power bias, emotional bias, and a whole host of cognitive biases. We will not root them out. We only want to weaken them using information science.

For Google, it would makes sense to add a customization option: (a) filter my knowledge by truth rank or (b) feed me with whatever bunk is popular. That will add one more layer of amplification to the truth: the less fortunate minds will get even more confused and weakened in the information battle. In addition, I would love to compare those two filtered and unfiltered worlds. PageRank is already doing a great job in that. I recall the days of AltaVista. I mentioned similar worries about InfoBabble in my PhD the whole two decades ago (1995). Google quashed infobabble in the search space once and for all, and fast. Now we will make another step in the right direction. If Google implements TrustRank poorly, Bing or other competition will step in, however, Google have a good track record and I am optimistic. If they are slow, others might come to the front! In the end, the ultimate hope rests with Tim Berners-Lee and his Semantic Web. Once we build an intelligent universal global concept network, its departures from reality and its own cognitive biases will only reflect the scarcity of data in the universe of all human knowledge of all history.

Fashion reflects convergence

Radioactivity was so fashionable at some point that it was used in cosmetics. It took many decades to convince populations around the world that smoking is harmful. To this day people and even some physicians believe that you can "catch a cold from cold". I hear and try to bust that meme daily! The truth is slow but it comes.

Ngram Viewer is a useful tool for analysis of changes in language, culture, and fashion. Scientific terms wax and wane depending on current interests and trends. Pseudoscientific terms tend to peak and never recover from decline. A skeptic may say that the decline comes from the mere fact of being labelled pseudoscientific. Information science says that memes are crippled by falsehoods and the handicap increases with available information due to an increase in probability of associative contradictions.

Ngram Viewer is a useful tool for analysis of changes in language, culture, and fashion
Ngram Viewer is a useful tool for analysis of changes in language, culture, and fashion

Figure: Ngram Viewer is a useful tool for analysis of changes in language, culture, and fashion. The picture show an explosion of interest in telepathy in the early 1900s, and how it gradually declined over time with telepathy currently holding a status of archetypal pseudoscience

Here are the peak use times for some terms:

  • phrenology 1835
  • telepathy 1915
  • levitation 1917
  • psychic 1920
  • hypnosis 1961
  • scientology 1973
  • parapsychology 1976
  • iridology 1988
  • UFO 1998
  • psychoanalysis 1996
  • cold fusion 1996
  • chiropractic 2000
  • magnet therapy 2002
  • homeopathy 2013
  • creation science is still going strong and waiting for its peak

You can do your own searches here.

Interruption and disruption in SuperMemo

SuperMemo insert. What is SuperMemo?
Worshipers of "deep reading" and "deep thinking" often complain that humanity slowly degrades its ability to focus. We get the attention span of a goldfish. Superficially, incremental reading may seem to fit that picture perfectly. To an observer, it may look like a hurried run between pieces of information in search for some intangible reward. No "deep reading" implies no "deep thinking". Maryanne Wolf even says that "deep reading is indistinguishable from deep thinking". If "deep thinking" is understood as contemplation, and uninterrupted analysis, then it needs to be stressed that nothing prevents the user of SuperMemo to pause and contemplate for an hour over whatever precious text she is reading at the time.

First of all, there is no multitasking in SuperMemo. Incremental reading makes it possible to read "in parallel", but this means "hundreds of articles go through the process". At any given moment of time, only one article receives focus. This is a full undivided and clear attention on just one piece of text. SuperMemo actually helps avoid multitasking. If the user opens a dozen of tabs in his web browser, SuperMemo makes it possible to import all the tabs as separate articles at set their priorities. They will all be processed independently and with due attention.

SuperMemo carries no obligations. It juggles information for the same reason for which TV viewers zap channels, bibliophiles browse a library, readers skim passages of a text, and Internet users open multiple tabs in their browser. Consumers of information search for meaningful messages in the noise of a broadcast. The main difference between chaotic searches and incremental reading is that SuperMemo does the search in a systematic manner. No information is lost. No distortion introduced. All knowledge gets prioritized and processed in due time.

Nicholas Carr complains that modern technologies lead to multitasking which affects the way we think. He cites interruption as one of the villains in this picture. It is important to stress that interruption is central to the efficiency of incremental learning. Multitasking and interruption are seen as carrying the overhead cost. Each switch between tasks, each switch of attention carries a cost. Interruption is bad for attention and disrupts focused thinking in problem solving.

However, we need to distinguish between external interruption and internal interruption. An external interruption, which I rather call disruption, may come from a call, e-mail, advert, or any other environmental alert. This is quite bad for all forms of intellectual effort. External interruption may be a form of brain exercise, but its costs usually far outweigh the benefit. When focusing on an important job, we should avoid disruption if possible. I personally avoid it religiously. I always work in a silent locked room without a phone or e-mail.

An internal interruption is a norm in incremental learning. I read most of my texts in SuperMemo. While reading, I may trigger an interruption under a dozen of criteria. These might be:

  • the text I read appears to be of low quality. I may deprioritize or delete
  • I have extracted the most important information from the text and want to continue with my work, or with other texts. Current text may be processed further at later time, at lower priority, etc.
  • the text is hard and burns too deep into my mental resources. I may want to retain some of the resources for other texts in the queue. Hard texts are best read in small portion with maximized focus
  • a must-read text is boring and I struggle with maintaining high attention for longer stretches of time

As you can see, no condition implies an obligatory interruption. A fascinating piece of reading that satisfies the learn drive all the way through may stay in focus for an hour or as long as the brain can sustain the effort.

If these conditions seem trivial, note that there are always hundreds of vital texts in the queue. Every text carries an opportunity cost. Instead of reading text A, I might be reading a more valuable text B. Opportunity costs are best evaluated by reading many texts and prioritizing. Even the most focused bibliophile will preview a book before making a commitment. Even the most focused "deep thinker" will give up a book when it does not satisfy his quality standards.

The costs of switching between texts are negligible in well-managed incremental learning. Each switch is a form of IQ exercise in context retrieval. With practise, that exercise is waning. Most importantly: interruption and spacing are vital cost-effective tools in developing long-term memories. If we read a book today, we may remember little in a year. If we read the same book in chapters spread a month apart, recall may improve dramatically. If we process the same book with incremental learning, the recall may be excellent. It all depends on the way of processing, priorities, and devoted time.

Internal interruption has nothing to do with disruption and is a precious tool in making the most of information sources in the long-term. Incremental learning boosts contradictory processes: focus and creativity. If focus is poor, e.g. due to a late hour, vital texts can be postponed for better times. If focus is great, the reader can do as much "deep thinking" as necessary. On the other hand, combining pieces of knowledge in incremental learning, or better yet in neural creativity, is the best known way to stimulate creativity via information presentation.

There is no better, no faster, and no more permanent way of reading than incremental reading.

Further reading

Summary

  • Internet makes us smarter
  • mankind's knowledge seems to converge onto a useful model of reality
  • many forms of bias are unavoidable in knowledge and in models of reality
  • truth is an asset for a meme, but it is no warranty of the win in a battle of memes
  • pseudoscience and fake news are gradually exterminated in the process of knowledge convergence
  • TrustRank from Google will speed up the process of eliminating misinformation on the web
  • ubiquity of information and convergence of knowledge imply that we have fewer reasons to worry about unschooling and/or free education
  • disruption and multitasking in intellectual work lead to reduced performance
  • interruption in incremental learning improves focus, creativity and long-term memory (see: value of interruption)
  • Wikipedia is a fantastic global and universal source of introductory knowledge into almost any imaginable topic
  • Wikipedia has outdistanced all other encyclopedias in its scope, depth, up-to-dateness, objectivity, and even reliability
  • despite being crowd sourced, quality of knowledge stored at Wikipedia keeps increasing
  • SuperMemo has dedicated filters for easy imports from Wikipedia
  • for a user armed with incremental learning, errors and falsehoods of Wikipedia can be tackled easily