SuperMemo is useless!

If you believe SuperMemo is useless (at least beyond learning vocabulary), you are not alone. Please have a look at the following e-mail. It was received at SuperMemo World on January 25, 2000 and seems quite representative. Many people share similar opinions about SuperMemo:

1. Your program is useless. In today's world, information is so abundant and can be accessed so readily that it is hardly necessary to lumber one's memory with it. Tell me: What's the difference between persons A and B? Person A can recall thousands of facts in 5 seconds. Person B can look up those facts in 30 seconds. Clearly, person B is more effective, because he/she hasn't invested 30 minutes every day in appending the facts and then repeating them.
2. Your program is useless. You claim it makes one memorize 95 percent of the stored knowledge. That means one doesn't remember 1 item in 20! Given the time cost of using your program, and the psychological barriers one has to overcome (e.g. forcing oneself to do repetitions every day), the effectiveness you offer is not worth it.
3. Your program is useless. Being successful is not about knowing all the answers. Rather, it's about the ability to ask the right questions. Your program has nothing to do with that.
4. Your program is useless. People lead highly successful lives without it, because the brain is impressively good at remembering knowledge which is important to us or knowledge we use frequently - in short, knowledge we really need. SuperMemo may be good for memorizing a bunch of useless trivia to show off in front of other people.

Do you share similar convictions?

Please note:

• this text has been pasted literally from a customer's e-mail without abbreviations warts and all
• it was written by a person with a remarkably high IQ (150 falls right above Einstein -- although still below Sharon Stone)
• it was written by a soon-to-be graduate of a prestigious university
• it was not written by a prejudiced bystander. It was written by a long-term user of SuperMemo 6 for DOS!
• later we will refer to the above e-mail as "SuperMemo is useless" e-mail and refer to its author as SuperMemo Critic 

SuperMemo World's opinion: "SuperMemo is useless" e-mail bolsters the Socratic maxim: All troubles come from ignorance. The usefulness of SuperMemo is so obvious that one might argue, we should not spent time contesting this e-mail. However, there are many people who would share the convictions of SuperMemo Critic and our mission is and must be to work on changing similar misconceptions! 

This article is intended to point out errors in "SuperMemo is useless" and as such also demonstrate that the greatest barrier on the way towards the widespread use of SuperMemo is the difficulty in overcoming prejudicial convictions of which many are deeply entrenched in tradition of many cultures. If you disagree, please feel free to voice your opinions in SuperMemopedia.

The official rejoinder below is divided into the following sections:

• common sense reasoning against the presented four accusations (for those who like examples and popular scientific approach)
• falsification of the accusations using a more scientific approach (for those who do not mind some more hermetic language)
• comments to the above text gathered in a quick user poll  
  • opinions in favor of SuperMemo
  • opinions against SuperMemo
• articles: For and Against, written by users of SuperMemo

Common sense refutation

Let us first contest individual points of "SuperMemo is useless" e-mail using common sense examples. Later, using more strict reasoning, we will point out well-known concepts of science that falsify all four quoted points

Point 1 - Memorization is not needed

Is it true that we do not need memorization? With constant improvement in access to knowledge, we can obtain answers to important questions with ever increasing easiness

False!

• If this were true, we could travel around the world without learning languages. After all, finding out words in a dictionary takes mere seconds
• If this were true, students of medicine would not need to cram details of human anatomy and physiology. Instead they would learn to use state-of-the-art expert systems with all answers built in. If you asked: What's up doc? You will hear: I have no idea but wait ... I will check it out on my computer
• If this were true, all exams such as SAT, GRE, TEOFL, FCE, GCSE, USMLE, etc. would be a great waste of human time and resources. Students should rather come to exams with their link to the web and figure out answers ad hoc. Or they would just read out relevant paragraphs from a textbook
• If this were true, nobody should poke fun at George Bush Jr. for his lack of knowledge of the heads of state of Chechnya, Taiwan or Pakistan. After all, he can find those names in seconds on his palmtop. If using external storage was permissible, nobody should blame George W. for mixing up Slovakia with Slovenia, or refer to Kosovars as Kosovarians, East Timorese as East Timorians, or Greeks as Grecians (NB: the word Grecians is considered correct too)
• If this were true, you could be a rocket scientist at NASA tomorrow! After all, rocket scientists follow well-know rules written in well-know manuals organized in a very well-know manner. If you need to plot the trajectory for Galileo to beam images of Europa back to Earth ... no problem ... take the Advanced Calculus textbook, figure it out, and send your billion dollar mission on course
• Last but not least: Two ideas do not come together to produce a great invention unless they sit in the same head. They either have to be called up at the same time or one has to come unexpectedly from the outside:
  • Archimedes stepped into a bath and immediately associated the fact with all he knew about weight and buoyancy. Thousands of Greeks at his time failed to make the link. This was not solely because of their less lively minds. It was mostly because of their lack of understanding of hydrostatics
  • Isaac Newton is said to have been hit by an apple that produced an immediate use of his knowledge of physics and mathematics to enhance his three famous laws of motion published in Principia by the laws of gravity
  • James Watt is rumored to have watched a kettle boil to combine this inspiration with his knowledge of engineering to improve upon a Newcomen's steam engine. This association changed the industry upside down in the course of the hundred years that followed

  SuperMemo can help you be creative not only by combining various pieces of knowledge in your mind. It can help you generate new ideas while making repetitions! Genius associations do not come from the mere existence of two pieces of knowledge in your brain. Those two pieces of knowledge have to light up at the same time. Only this way can your brain make a connection. Strangely, a repetition related to genetically modified tomato can light up memories related to tomato juice, cucumbers, genetic disorders, take-home dinner, flu or even the silent Mars Polar Lander. If your collection combines knowledge pertaining to different subject domains, the stream of new ideas and unexpected associations coming to your mind may surprise you!

  The advantage of keeping knowledge in your head as compared to keeping it in external sources can metaphorically be compared to the advantage of going from primary through secondary to university education as opposed to getting a week-long course on digging info from external sources. Nearly all parents seem to prefer to choose the former for their kids 

Point 2 - Remembering 95% of knowledge still leaves 5% forgotten which is not worth the cost of repetitions

Is it true that SuperMemo makes us pay too much for knowledge? 

False!

1. SuperMemo gives you a choice to determine how much you are ready for pay for knowledge and how much knowledge you want to keep in your mind. If 95% is too little, you can increase the forgetting index. If 95% cost you too much work, you can decrease the forgetting index. You can decrease it so much that you will hit the level of maximum speed of knowledge per time cost! If SuperMemo were dumb enough, it would even let you decrease it further so that your retention would drop down to the level you normally reach in learning without SuperMemo!
2. The complaint on the loss of precious 5% of knowledge stands in contrast with Point 1 from which we might conclude that 5% is not a big deal if it can be accessed in 30 seconds from external sources
3. It is enough to ask any user of SuperMemo who has passed TOEFL or GRE: Were hours of repetitions worth the resulting knowledge? All those who have previously invested weeks and months or even years into language courses will not hesitate to testify in favor
4. One of the greatest benefits of SuperMemo in learning is that building new knowledge often requires a solid foundation. With a good foundation, new concepts easily slot in an make a harmonious entirety. Contradictions and discrepancies are easily located. The time invested in a good foundation will pay back manifold in the course of passing years

Point 3 - Success is based on posing right questions; not on knowing the answers

Is it true that SuperMemo requires cramming instead of reasoning?

False!

1. The question-vs-answer dilemma is not defined well-enough to be of philosophical challenge here! After all: why ask the question if the answer is known. If we knew how to proceed with cold fusion, why ask the question How?
2. Memorizing the answer as opposed to understanding the ability to derive it cannot be rejected outright either. It is obvious that it is better to learn to multiply rather than memorize the product of all possible combination of numbers. Yet few students are encouraged to derive formulas for finding the solutions to quadratic equations each time they encounter such an equation. It is simpler to memorize the formula than to derive it in need. It is only a matter of balancing the cost and the benefit. The derivation may be useful to mathematicians as a testing ground. An ordinary engineer will rather recall the formula or even use a computer program 
3. The ability to ask right questions is also based on knowledge which has to be learned. A primary school student will usually not be able to ask right questions about the origins of the universe as the concept of the universe may be hard to grasp without understanding the earth, the Solar System, the galaxies, distances measured in light years, time measured in millions of lengths of human life and zillions of lengths of a lunch break
4. Without relevant knowledge, you might never ask the question in the first place! If you have no knowledge about greenhouse gases and environmental degradation, why ask questions about the greenhouse effect?

Point 4a - Memory has an excellent ability to retain important information

Is it true that our brain is extremely effective in remembering knowledge that is important?

False!

1. If this were true you would not tremble before an exam and confusingly run through the notes to be sure that at the zero hour you won't suffer from a proverbial blackout. At the same time, you could easily recall details of a Schwarzenegger movie seen last evening (or even weeks before the exam). Clearly, Arnie beats the ups and downs of the Ottoman Empire. And if you think the Ottoman Empire had much greater an impact on humanity than the island shootout in Commando, you are still likely to remember the muscle and the machine gun far better than the timeline of the sultans
2. If this were true you would never have problems with recalling the date of your mother-in-law's birthday -- a piece of data often critical to your marital harmony
3. If this were true, you should instantly forget the Olympic champions in football in Atlanta 1996 or Munich 1972. After all this might be a classic case of unimportant knowledge. Yet few Africans would forget how Nigeria beat Argentina 4:3 in 1996. Similarly, few Poles would forget the most memorable moment in the history of Polish football: Olympic championship in Munich

The truth is that we excellently remember only things that are both easy to remember and repeated frequently enough. The brain does not have an internal measure of importance (other than limited volitional control or control via repeated exposure)! Your memory storage ruthlessly deletes your career-critical knowledge with same ease as it ravages the traces of last year's golf scores or contributing names listed at the end of a boring soap opera.
SuperMemo provides you with the ultimate pigeonhole for mission critical data. What is in -- stays in your memory. What is out -- is free to go

Point 4b - SuperMemo is useless! People have lived successfully for ages without SuperMemo

Is it true that Albert Einstein, Ivan Pavlov, Norbert Wiener, or Adam Smith were successful enough to never have benefited from spaced repetition in learning (i.e. SuperMemo)?

False!

1. By analogy one could claim: The Internet is useless! People have lived successful lives before the net. All that thanks to their remarkable skills of verbal communication!
2. Einstein might have failed to step from special to general theory of relativity had he been armed with extra knowledge in physics or mathematics that would make his brain stick to the beaten tract. Equally well, this additional knowledge might have helped him arrive at grand unified theory. All pieces of knowledge in our mind determine its ultimate response to questions posed. However, the stronger your mindware, the greater the probability of generating educated results

Refutation based of facts of science 

The short version of the SuperMemo Critic's thesis is: SuperMemo is useless. By using external references and posing the right questions we can be successful at problem solving using the traditional methods (i.e. without spaced repetition in learning)

All external sources of knowledge so readily available to the author of the above thesis in the era of electronic communication, have not helped him accurately answer the question: Is SuperMemo useful?

Had the author of the thesis kept five relatively simple facts of science in his memory, he would not have arrived to his wrong conclusions. Of these five facts, three belong to well-established body of science and two to the less-well-known body of knowledge related to repetition spacing. Any one of these should suffice to refute the thesis:

1. associative memory (in computing sciences known as content-addressable memory) is superior to standard random access memories that are usually found in our desktop PCs. Human brain is based on neural networks that belong to the associative memory category
2. evolution of the human brain has essentially been completed before language skills and abstract thinking have developed. The differences between the brains of early Homo sapiens from 200,000 years ago and present humans are in the context of this discussion insignificant
3. problem solving in artificial intelligence can be used to mimic those mental faculties of a human being that we associate with intelligence and creativity. There are many approaches to problem solving such as deduction, induction, abduction, reasoning by analogy or probabilistic methods, neural networks approach, searching in state space, and more. And even though some of these can generate false conclusions or uncertain responses, they can all be expressed by means of deductive methods such as those used in theorem proving in mathematics. Deduction proceeds from axioms or premises towards true formulas or assertions through logical derivation using valid inference rules. The better the selection of assertions and the selection of inference rules, the broader the reasoning capacity of an expert system or human brain. Rich inferential knowledge stored in our memory makes us fast in thinking, creative, intelligent, efficient in problem solving, etc. Yet we need factual knowledge as the raw processing material for derivation. Even the creative vein and inspiration can be expressed in terms of multithreaded derivation and backtracking well formalized in logic programming. No magic fluids or reflections of the soul are involved here. Just a plain network of firing neurons recovering the encoded patterns of facts and rules. See also: Roots of creativity and genius
4. forgetting index in spaced repetition can be used to optimize the trade-off between the knowledge acquisition rate and knowledge retention
5. lifetime minimum cost of remembering a single well-formulated piece of knowledge can easily be estimated to fall into the range from 30 seconds to 5 minutes (minimum cost is the cost that is achieved using the optimally spaced repetition schedule). This estimation can be used in cost-benefit analysis in deciding which mission-critical pieces of knowledge should be subject to learning based on spaced repetition

Here is how the above facts influence our reasoning in reference to SuperMemo: 

1. Associative memory - understanding associative memory alone should decisively demonstrate the error in the belief that we could use external memory as a crutch in reasoning. No two facts or rules stored in two different books will come together to produce a new quality. This can only happen inside the human mind. In the future, artificial intelligence systems should also be able to do the same, and only these will ultimately stand a chance to prove SuperMemo useless (perhaps making human brain of little use at the same time). All innovation in the history of science and technology is based on association. Forgetful memory can be helped with external sources of knowledge and each major scientific breakthrough is based on a series of smaller steps, many of which will be based on hours of search, trial-and-error and experimentation. However, none of those little steps happens outside of the human mind. Hence the brain armed with extensive knowledge is more likely to come up with a selected association than a brain trained solely in scanning external references
2. Maladapted brain - the evolution of the brain proceeds too slowly to have helped us adapt its structures to abstract thinking. What was excellent for survival 200,000 years ago does not suffice to process modern abstract knowledge. Simple computational tasks such as multiplication or division proceed in a shamefully inefficient way in the human mind. After all, early humans did not need to multiply. At the same time, no state-of-the-art computer can compete with the visual cortex in pattern recognition and processing (although the brain is just about to be beaten here as well)! Recognizing the enemy or prey was critical not only for Homo sapiens but also for birds, reptiles, fish or even insects. The only measure of the importance of knowledge our brain synapses have at hand is the pattern of repetition, levels of circulating hormones at the time of exposure, and a limited impact of conscious attentive labeling of information as important at the moment of encoding. Forgetting is needed to optimize knowledge storage; hence we have to forget less important things. However, modern life has changed the hierarchy of value and importance in the area of knowledge. The link between importance and repetition has been severed. A flashy lingerie billboard we see every morning is not likely to be more important than dozens of volatile facts pertaining to our professional life. Regrettably, there is no circuit in our brain that would let us consciously etch important memories: This is important! I must not forget it! All we can do is to use the trick of reverberation or mnemonic techniques which ... still will usually not last long unless we apply spaced repetition (i.e. SuperMemo)
3. Derivation-vs-solution - we cannot blindly assume that it is better to derive answers to problems than to simply memorize the answers. It is always a matter of cost-vs-benefit. In problems with multiple instances, we will tend to learn the derivation. For example, it is easier to learn how to derive the third power of X than to memorize all useful pairs: argument-result. On the other hand, most of people memorize the multiplication table at the time when they could still derive the result from bit-operations on binary numbers. Memorizing some 40 or so combinations seems more practicable, esp. that derivation always takes time and we make use of the multiplication table thousands of times in a lifetime. Here memorization saves hours and days in the lifetime perspective. Brain armed with rich factual and inferential knowledge will associate the most remote ideas and derive the most unexpected conclusions. Through a conscious control over this rich reasoning process, we have built the present civilization
4. Forgetting index - traditional learning (i.e. learning that is not based on spaced repetition) can easily be compared to learning with a very high forgetting index (depending on the field, extent of time, etc. this number might easily reach beyond 90%). This fact shows that there is no sharp transition between retention in SuperMemo learning and retention in non-SuperMemo learning. Consequently, the only question which might be asked is What is the optimum forgetting index in learning? Is it closer to the one used in SuperMemo (10% by default) or closer to traditional learning (say 90%). The answer to this question will depend on a given student's priorities and on a given piece of information! This is why SuperMemo provides an individual forgetting index to all elements in a SuperMemo collection. Mission-critical pieces can be remembered at a high repetition cost with recall probability of 99%. Fringe knowledge can be stored with retention 90%. NB: all users who tried to compare learning with the forgetting index of 20% with the default 10% know from their own experience how frustrating the repetitions in the former case can be. This is a reflection of the traditional learning in a mild case! Only those who truly experienced SuperMemo can testify to this
5. Cost of knowledge - each piece of knowledge stored in SuperMemo contributes to the overall time expenditure. Using well-formulated knowledge, we can safely assume that a single element will be repeated from 8-20 times in a lifetime. If a single repetition time is assumed to fall into the range from 3 seconds to 15 seconds, we could conclude that the expected lifetime cost of a single item ranges from 24 seconds to 5 minutes. SuperMemo 99 adds a new statistical measure called Cost. It measures the total time cost of a single memorized element. In a well-managed long-term learning process, this cost is usually detected at 2-3 minutes/item in a 10-year bracket. This is then more than the theoretical prediction due to the fact that each collection contains a small subset of so-called leeches which dramatically increase the average cost/item (note that leeches can easily be detected and removed from the learning process).
   The only rational criterion for deciding if a piece of knowledge should be memorized by means of SuperMemo is to judge the benefits of having the given piece constantly available in memory. If the benefits do not seem to add up to more than 10 minutes, the student might simply not add an item to SuperMemo. An exemplary cost-benefit discussion of individual items can be found here: SuperMemo as a new tool increasing the productivity of a programmer

Conclusions

In 1979, Dan Bricklin and Bob Frankston introduced the first killer app: VisiCalc. The precursor of Lotus's 1-2-3, Borland's Quattro Pro and the current leader: Microsoft's Excel. The program gained a relatively high recognition at its time due to the innovative and easy-to-grasp concept popularly called a spreadsheet. Now VisiCalc is all but forgotten; however, it has for long remained a beacon of hope for would-be software developers looking for the next killer app.

When SuperMemo World was founded in 1991, it quickly attracted strong support from many who expected it to become the next great killer application. Although it has ultimately reached the magic figure of the estimated one million users worldwide, it is far from being a typical killer app. The reasons behind it are well illustrated in SuperMemo is useless e-mail. When Dan Brickling spoke about VisiCalc, he mentioned a factor of success that is missing in SuperMemo: instant gratification. In his words: Any product that pays for itself in two weeks is a surefire winner. This is not the case with SuperMemo. SuperMemo does not pay for itself even in a month.

When 21-year-old Mark Andreessen and his pal Eric Bina wrote their Mosaic and let it spread free over the Internet in 1993, they set a new standard for a highly useful, easy-to-use and a fast spreading software application. The example of Mosaic remains a reference point to software developers who would like to see their applications approach its record-breaking levels of popularity.

What sets Mosaic, VisiCalc, MS Word, and computer games apart from SuperMemo? We are all convinced that cultural barriers and educational tradition are the greatest inhibitors of SuperMemo's further expansion. Program's complexity also discourages many in the beginning. It is not the marketing outlays that make a difference here. Simple and well-designed applications quickly raise to million-user figures in the present Internet-based software culture with widespread freeware, shareware, GNU and open source paths to success. SuperMemo freeware has always been out there on the net, yet it never approached a fraction of popularity of such handy tools as WinZip or AceReader. For ages, we all have been taught at school: Don't just memorize it! Reason it out! This is why there is so much disdain for repetitive learning. Time and again, we see reviews of SuperMemo in the press or on the net that miss the critical point: spaced repetition! Time and again, we hear from users that they tried the program years ago and did not like it the first time. Why do I have to grade myself? Why does the program not ask me to spell the answer using the keyboard? Clearly, SuperMemo ain't Mosaic.

Yet, we take the obstacles in good part. The mission is: make the technology more popular. Make people understand it. This is why you can bet SuperMemo 2010 will offer qualities that we cannot even imagine today. 

In the past we have seen major blows given to technology as a result of prejudice or ignorance. Florence issued an edict in 1299 forbidding bankers to use Arabic numerals. The city council of Danzig ordered the inventor of ribbon loom to be drowned in 1579. The print was threatened as well: book in the hands of masses can topple the king. Brisk mind did not prevent Napoleon to temporarily abolish the decision to adopt the metric system in France (ironically, the US has ratified the Metric Convention as early as in 1875). Did we not hear in the past that cars promote immorality, or radio will ruin the family life by eliminating conversation? Or that the web will bring a global infobabble?

It took a century to make a full use of the steam engine. It took years to make a good use of a telephone. Things went faster with electrification and automobiles. Radio and TV were widespread within a generation. Education is the key to such accommodations. Hence the positive feedback: we need more education, demand for education is conducive to SuperMemo, SuperMemo helps education, the more we know the more we know we do not know, and the self-propagating circle of demand closes.

There is little acceptance today for the genetically modified food, which is likely to revolutionize food science. Some people question the benefits of vaccination, which was the main contributor to the dramatically increased life expectancy in the 20th century. Some even are wary of the Internet seen as a source of pornography and hatemongering. 

E-mail with the subject: SuperMemo is useless is for us a good occasion to ponder over the future of spaced repetition in learning. A small step towards the universal acceptance of the technology.

---

USER COMMENTS

We have sent a few dozen e-mails to the most active and vocal supporters of SuperMemo. This is why the comments below are in unanimous disagreement with SuperMemo Critic. Yet we realize the sample is not representative and will post other interesting comments as soon as we receive them

We have shortened some texts, softened the language, removed personal data and placed our short comments in square brackets

---

User contributions

AGAINST SUPERMEMO

• Witold Szafraniec, Poland - No force in the world can convince me to SuperMemo!
• Criticism at SuperMemopedia

FOR SUPERMEMO 

• Luis Gustavo Neves da Silva, Brazil - It's more than just memorizing 
• Satisfied Customer, USA - How I got a job at #1 software company
• Peadar Willy de Bhuinn, Ireland - Homeschooling with SuperMemo

---

5 years later

SuperMemo Critic had these comments to this article:

Negating something is a fundamental way of verifying it. These four arguments are about the best I can do for now in the way of negating SuperMemo. If you can pulverize them, your program will have withstood the challenge. If somebody then uses these arguments, you will be prepared to fend them off. Your article can forestall the enemy's move in that direction.

By 2005, SuperMemo Critic has amazingly become closely involved in promoting SuperMemo on the Internet. He also uses SuperMemo for learning foreign language vocabulary