Michael Nielsen, standing on one foot

A highly sophisticated MR reader demanded a dose of Michael Nielsen.  I wrote to Michael, and he was kind enough to oblige.  Everything that follows is from Michael, here goes:

I started with the question “What might amuse Tyler?”, and it became very easy.

Three opinions that may amuse MR readers:

1. Peter Thiel has said: “We wanted flying cars, instead we got 140 (280) characters.” Thiel is wrong: 280 characters are much, much better than flying cars. Twitter is misunderstood as being an online service; it’s merely the online component of a much improved offline experience. Twitter DM’s are a superpower, one of the most valuable ways of connecting people ever invented. More on one way of using Twitter here.

2. Movies are primarily a visual form; movie criticism and the popular conversation about movies are primarily a literary form, and informed by literary sensibilities. This is why good movies such as Transformers are so underrated. People who dismiss such movies are mostly revealing their own ignorance.

3. Many corners of the internet have a culture of judgement or argument. Typical subtexts in online conversation are: is this good or bad? What’s wrong with it? But until and unless healthy conversational norms are formed, argument and judgement are mostly useless status-seeking by participants. Much better is a “Yes, and” culture.

Three books or papers which should be better known:

1. Elinor Ostrom’s book Governing the Commons.  Ostrom dismantles the market / government dichotomy, sketching out ways common pool resources (and, to some extent, public goods) can be provided using non-market, non-government solutions.

2. Alex Tabarrok’s paper introducing dominant assurance contracts. Cryptocurrencies have huge potential as a way of creating entirely new types of market, using ideas like this. This potential is mostly unrealized to date.

3. Bret Victor on Media for Thinking the Unthinkable.

Blog posts don’t really get going until about 5,000 words in. Here are three favourites of mine:

1. Thought as a Technology, on how imaginative designers invent fundamentally new modes of thought.

2. If correlation doesn’t imply causation, then what does?

3. Using Artificial Intelligence to Augment Human Intelligence (with Shan Carter).

Despite the fact I’m well short of 5,000 words, I’ll stop here.

You can follow Michael on Twitter here.


I think I remember Bret Victor from a previous MR link. I just found this video which is nice:


Correlation doesn't "prove" causation.

The claims about movie criticism and conversation are just staggeringly ignorant. The idea of looking at cinema as visual was widely disseminated through the essays of Sergei Eisenstein, who died in 1948, and who made rather significant contributions to the visual language of film himself. At the very latest, the essential role of visual language in cinema was established for critics and the more educated kind of conversation by the Nouvelle Vague directors/essayists in France in the 60s, who drew attention to the merits of Hollywood films which were not obviously 'literary', particularly those of Hawkes, Ford and Hitchcock and which have since become firmly established as great classics.. Bay is just a very uncreative director and Nielsen is scrabbling about for excuses to think himself frightfully cultured for enjoying Transformers, though it is widely derided. If you enjoy it, enjoy it, don't pretend it's more than it is.
Yes twitter can be great, but not because of DMs which are clearly secondary in impact as a form of messaging compared with email, Facebook messaging, WhatsAPP, SMS, iMessage and maybe a few other similar services. Nielsen seems like a good thinker on the other topics mentioned, but on cinema, embarrassingly bad, and on twitter, oddly muddled though not totally wrong.

What nonsense. It's not "ignorant" to think Transformers is a good movie. It just means your taste diverges from that of most movie critics and other members of the cultural elite (but not necessarily from the wider population).

Personally, I think he has a point here. I haven't seen Transformers, but Avatar was a movie that was not "good" and yet was one of the most enjoyable I have seen due to the stunning visual experience. I thought it should have won Best Picture.

I’ve not seen Avatar, im curious, does it hold up? Is it still worth watching now that sfx are dime a dozen?

Avatar is very, very bad. I suppose you could watch it for the "visual experience" if it is on mute, but there are still many better ways to spend your time.

I saw Transformers. It was not very good, and I like a good action movie. Mr. Nielsen sounds like the kind of fellow that enjoys striking provocative, contrarian poses, which is a hit or miss game, but it makes for an interesting Twitter persona. Worth a follow.

I thought the first Transformers movie was well done and engaging. It was a good action block buster movie. The sequels were poor in my opinion.

makes for an idiot's idea of what a smart person is like. to be fair, that is a *great* description of Tyler, so it fits commenters here.

Transformers (2007) has an 85% audience score. That’s quite high.

I don't think Transformers should be under-estimated. Michael Bay managed to get a reasonable performance out of Shia LaBouef and a not-half-bad performance out of Megan Fox. That is an achievement in itself.

Even if both were grossly out-acted by some large pieces of chromed scrap metal, one of whom would not even speak.

That's uncharitable, the metal is not scrap.

Barry is correct. Watching images dance across a screen for 90+ minutes is only entertaining if you're on acid. Otherwise, the images need to tell a story to keep the audience engaged. The story needs to have characters and a narrative arc. Hate to break it to you, but you need literary sensibilities in order to craft something with these elements that audiences will find compelling.

Agreed. I personally find better storytelling and get more enjoyment from the better TV series. I'm guessing that these series employ the better writers and can provide them with somewhat long-term employment. The budgets likely create the need for better story-telling and the series gives them the time to tell it.

I generally agree with this but I think its more accurate to say "narrative sensibilities" or "story telling sensibilities" than "literary sensibilities." Movies, books and plays all need to tell a good story (i.e., they need good story telling sensibilities) but they have different tool kits for doing so.

FWIW, I feel like this comment from Nielsen makes me want to ignore him. It shows no interest in learning about a topic before spouting off in arrogant show-offy rant.

You didn't explain what was ignorant. You just said a bunch of names.

I mean, maybe you didn't intend to explain what was ignorant and only intended to signal that _you_ knew it was ignorant because you know the work of _people_.

It is not really an exaggeration to say that Twitter would be an amazing web site even if Michael Nielsen were the only user.

Opinion 3 is good! Oops, I mean Yes, and...!?

Unlike some here, I found Michael Nielsen's comments about Twitter and Transformers to be both fascinating and intellectually stimulating. I would like to hear Nielsen's opinions on other subjects, for example, what does he think about professional wrestling? Perhaps Tyler Cowen could ask Nielsen about Monster Truck Rallies. Does he agree with Professor Heinrich Von Blatter that Monster Truck Rallies are a metaphor for the ultimate destruction of human civilization?

Back to Transformers. Is there really more than meets the eye, or is that just an illusion?

Wow, I wish I knew about Michael Nielsen earlier. His blog that got linked is amazing.....I wonder if MR would ever consider an "econ polymath project" like what Nielsen champions in his work on OpenScience (http://www.michaelnielsen.org/papers/mcm.pdf)

Wow! I never knew I was a 'highly sophisticated MR reader'!

Thanks, Tyler!

Cryptocurrencies have huge potential as a way of creating entirely new types of market, using ideas like this. This potential is mostly unrealized to date.

The SEC has launched a lawsuit against a guy who pulled off a fraudulent ICO: https://twitter.com/stephendpalley/status/1001561673864548352

The cool thing about it though is the buzzphrases that the fraudster was claiming to offer: "Company as a Service", "Bring Your Own Cloud", "Mining as a Service" ... phrases that sound just like the ones uttered by the "Blockchain Thought Leaders" and are equally meaningless.

Fraud works because the fraudster mimics honest people. The archetypal wolf (fraudster) in sheep's clothing (buzzphrases) case.

If you focus on the sheep's clothing you'll never spot the wolf ☺

Yesterday's TC cryptocurrency post was by a partner at Ycombinator and stated "One day we’ll have some type of proof that is truly dispersed by nature. My two favorite candidates are (a) a coin based on DNA or (b) true proof of location"

Is Daniel Gross really have a somewhat developed idea for a coin based on DNA? And have some convincing argument that it is likely to be more successful than other coin ideas? And does he have some notion of "proof of location" and understand why it will work better than PoW and PoS?

More likely that it is all jargon and signalling ...

Proof is useless unless someone reads it.

Causation: Nielsen's "other factor" has significance in every point on his list. Finding, or seeing, the "other factor" can be a challenge, ignoring it all too often being the rule. Thiel doesn't need me to defend him, but Thiel's flying cars is a metaphor. I suspect that Nielsen objects to the metaphor because he prefers complex ideas over concrete results. Autonomous cars would be a better example. As Elon Musk has learned, making quality cars is hard, and to the mind of people like Nielsen, it seems so easy, especially as compared to developing the software needed to make them autonomous. If Nielsen were Henry Ford, he would have developed the software for cars rather than the cars. I'm not sure the software would be all that useful without a car, but who are lesser mortals like me to question Nielsen's priorities. In simpler times, people like Henry Ford identified an obvious problem (such as moving people from point A to point B) and found a solution for it. Today, smart people like Nielsen find a solution first and then search for a problem where it might might be useful. For some reason blockchain comes to mind (as do autonomous cars). One might call it effect and cause (as opposed to cause and effect).

No, Transformers is a visual mess of a movie, and the sequels are even worse. Yes, I am aware that I am using words to describe visuals, but I don't have access to a movie camera to criticise the movie to Nielsen's satisfaction.

If you want an utterly superior "visual" movie, watch Fury Road.

Twitter is CB radio without the twang.

It's only usefulness is it's ability to show how stupid, venal and infantile most of the so called smart people are. We now have proof positive that journalists should have kept up the heavy drinking loush image, because that is all they really are and were. Now they are just boring over excited idiots.

Thought as a Technology

Progress in interface is a slow because we have accepted as a core principle that people are lazy.

If you said "I have a new interface that looks weird, but trust me, commit weeks to learning it in your world will be rocked," you will have no takers. It has been tried.

Apple and Microsoft discovered this long ago, and provided the masses with the same interface and small changes over and over. Changes had to be "discoverable" and that became the keyword for decades. Anyone can pick up a system that is discoverable and get it to do more or less what they want. This is the path of least resistance.

The only problem with this path is that it obviously creates path dependency. You can't get to any solution space that was not discoverable every step along way.

Who is working on radical new interfaces, ones that would require diligent initial learning by users? I would guess no one. CAD programs will be much like previous CAD programs. Even with things as creative as computer languages, users stick with favorites, and are most open to "new" languages look much like the old.

Laziness is a meta design restraint.

"If you said "I have a new interface that looks weird, but trust me, commit weeks to learning it in your world will be rocked," you will have no takers. It has been tried."

That pretty much describes every smart phone. That's a product that didn't exist 20 years ago. And yet they've been very widely adopted.

Furthermore, any kind of knitting, musical instrument playing, driving, riding a bike, or honestly anything that takes scores of hours of practice to become even nominally qualified would fit under that description.

So, no, your statement is just wrong. Clearly wrong. Not even really close to being correct.

If smartphones must be learned and not discovered, where is the user manual?

iPhone User Guide: https://help.apple.com/iphone/11/

At this point it is obvious that discoverability is a Term of Art in user interface design but you simply refuse to comprehend or accept.

Michael Nielsen and Bret Victor certainly know the history.

If you want to make a contribution, do some reading.

It's obvious at this point when the actual facts undermine your comments that you resort to attacking the other person.

I'll give you another chance.

In the beginning, computers were considered unfriendly. If you typed "help" they might say "command not found." This was fine with many early computer programmers, who invented things like the vi editor. vi is a good example of a computer program which is perfectly undiscoverable. If you don't already know which keys to push, you aren't going to make anything useful happen. But, and this feel could be a big but, if you learn first, if you study out of bandwidth, you can learn a very powerful and expressive tool. You can learn to do things they still make GUI editor users say "wow."

Discoverability came later, and as I say there's a big literature on this.

Consider my first comment again in light of this idea but no one is really doing undiscoverable program design anymore.

Well I don't really need another chance, but it does look like you are using this post to restate your argument in a better light. And I do agree with it to a large extent.

I specialize in Industrial GUI design, just for the record. So your comments regarding my ignorance were more amusing than anything.

"but no one is really doing undiscoverable program design anymore."

That's true to a large extent, but that's because non-discoverable design is considered low quality and/or old school. Indeed, most books on GUI or HMI design actually emphasize that self documenting, intuitive layouts and navigation are far superior to layouts that require extensive manuals and explicit user knowledge and training.

Man, that's dishonest.

You wake up to what I was talking about originally and say that I've said something new.

If you had understood, if you had been enough of an "Industrial GUI designer" to know what the word "discoverable" means you would have been on to this originally.

Did you do some reading?

"Man, that's dishonest.

You wake up to what I was talking about originally and say that I've said something new."

No, I'm quite honest. You just don't seem to understand what I'm saying. It's quite possible I'm not being specific and the fault is on my end.

To be clear:

"a) Progress in interface is a slow because we have accepted as a core principle that people are lazy."

This is silly and wrong.

"If you said "I have a new interface that looks weird, but trust me, commit weeks to learning it in your world will be rocked," you will have no takers. It has been tried."

This is also silly and wrong. Well the last part is true, but the idea there are no takers is ridiculous.

The idea that no one reads the manual and that all new devices, etc are learned purely by discovery is ridiculous. Now granted, most people probably do rely on other people reading the manual for them. But most people always did. How many people ever read a car manual? (I've seen figures of roughly 16%.)

But it's always been that way. Most people don't read the manual and most people never did. However, that's a far cry from making the kind of outrageous comments that you started out making.

"The only problem with this path is that it obviously creates path dependency. You can't get to any solution space that was not discoverable every step along way."

Yep another statement that's wrong and silly. Just because it's good practice to make icons, buttons and new features intuitive and logical doesn't mean you can't ever introduce a new concept.

"In the beginning, computers were considered unfriendly."

Yes! This is true. As a fellow engineer once said, "Unix isn't User Friendly, it's User Hostile.".

"Discoverability came later,"

Also true. Discoverability is an aspect of good GUI design. It's a "good" idea to not make your features opaque and their functions unclear.

Look no further than Sony electronics for a company that continues to produce products that are feature rich but tends to assume the customer not only reads the manual, but references it periodically.

Do us all a favor, by thinking about what I wrote for a day, instead of immediately furiously typing out some rebuttal. And then following up with a second post 3 minutes later.

Now you are just denying the world. Everything from new smartphones to new smart TVs come with a one-page quick start and that's it. No manual. Everything is to be discovered by an experienced user in common paradigms.

This is what creates path dependence. Common paradigms may be extended by discoverable features, but that's it.

Nothing off the map, nothing that would require front end education, is attempted.

Sigh. It feels like I'm talking to a wall. First, most smart phones and smart TV's have a detailed manual online and indeed my Samsung has an e-manual that is stored on the TV itself.


Secondly, the features themselves show embedded descriptions, that clarify what the feature does. It's far better to have a smart device with actual contextual help than a paper manual that you must flip through.

"Nothing off the map, nothing that would require front end education, is attempted."

Do you really think a TV or a phone should require someone to read a manual before it's useable?

Seriously what's your whole point to this long tedious serious of posts where you repeatedly ignore reality in favor of some nebulous narrative that "this path is that it obviously creates path dependency".

You can't be serious.

The detailed manuals are online and obscure place because no one uses them.

Everyone reading this page knows that and knows that you are just trying to paper over b*******.

The better argument, which you would have reached earlier if you had been at all serious is this:

Discoverable user interfaces are great for many things, and all many people need, especially casual users.

However. The thing that we cannot be sure of is that highest productivity interfaces can be reached by this path. Perhaps millions of man hours are lost, because no one wants to waste time learning to be good at email in an unfriendly but powerful email environment. Etc.

Perhaps a better interface would have prevented this stupid thread.

By the way, "thank you" for stalling this conversation at the definition of terms, and never reaching the actual argument.

And not one word on the original argument, that discoverability brings path dependence.

You can't "get to" any solution space that is not "discoverable" every step of they way, ever iteration of design, between here and there.

Actually I think the iPhone was audacious in being the first thing that was fully discoverable.

If I remember correctly even the iPhone One had no instructions. We just poked at them until they did what we wanted.

Here is an easy example of laziness as a design restraint:

If you create a new blog, it must look like all previous blogs.

Why? Visitors are lazy. They want to use your blog just as they always have. They want familiarity. If it is a bit different, they will give you two or three seconds as they try to figure it out. And then they're gone.

Any blog innovation you invent must be discoverable by users in two or three seconds. That is a pretty big path constraint.

No, you are describing the marginal utility of features or products.

Explain spreadsheets with your model. Lotus 123 drove the PC revolution. They are complicated, difficult to use, but amazingly useful hence worth learning to master.

Same with word processors. Wordstar and Wordperfect were hard to learn but for people who did real work they were amazingly productive. It was worth learning to use them. Most word processor users would do fine with a typewriter.

Everyone of my generation did the hard work of learning how to drive a car, often standard transmission, with various quirks to keep the damn thing running. Because it was profoundly useful. Easy is good, but productive is better.

The spreadsheet and word-processing were early "first draft" translation of established paper techniques to computing.

The word "spreadsheet" came from "spread" in its sense of a newspaper or magazine item (text or graphics) that covers two facing pages, extending across the center fold and treating the two pages as one large one. The compound word "spread-sheet" came to mean the format used to present book-keeping ledgers—with columns for categories of expenditures across the top, invoices listed down the left margin, and the amount of each payment in the cell where its row and column intersect—which were, traditionally, a "spread" across facing pages of a bound ledger (book for keeping accounting records) or on oversized sheets of paper (termed "analysis paper") ruled into rows and columns in that format and approximately twice as wide as ordinary paper.[14]

Let me say though, when you give examples from the 80s you might be right. There was a much more expectation of user learning in the pre-GUI and early GUI age.

The real commitment to discoverability came as GUI was accepted and user manuals cease to be.

Perhaps I should contrast "pre-learning" or previous study with in-the-moment learning by discovery.

Not everyone gets "discoverable," here is some related commentary:


Enjoyed this post. Even the stuff I disagree with (I'm not unreceptive to the visual/literary argument, but Transformers literally put me to sleep) was food for thought.

Strongly agree with the point about judgment and argument - with the big exception that, if a group identity is based on *good* judgment and argument, discussion within the group can be great. The old internet had many walled-off gardens where this was true. On the new internet, where so much discussion is funneled through public spaces (or at risk of being broadcast out-of-context there), this is less so.

I would frame the original Transformers as fun enough, without being good.

"3. Many corners of the internet have a culture of judgement or argument. Typical subtexts in online conversation are: is this good or bad? What’s wrong with it? But until and unless healthy conversational norms are formed, argument and judgement are mostly useless status-seeking by participants. "

It often seems that Tyler Cowan posts about interesting subjects and the commentors are trying to analyze it for deeper significance or trying to parse the meta-meaning of the post. Whereas, it often seems to me, that Tyler is just posting about something that's curious and he hasn't committed to any kind decision about the context.

I think that a thoughtful person should spend more time thinking than deciding. But it's certainly something I personally have to remind myself about and often fail to do.

1. I started reading the Twitter thread laying out Nielsen's argument. About 2-3 tweets in, I got sick of the Twitter format. Couldn't read the rest. Yeah, I still feel like Twitter isn't for me. Even after all these years.

2. Yeah, good stuff is good, and taste is overrated.

3. See #2. He's completely right on this point.

I can only assume he was talking about the 1986 animated Transformers movie - truly an unappreciated masterpiece.

On number 2, it's that correlation doesn't necessarily cause causation, for example, let's say (hypothetically) in the past 10 years, violent crimes have lowered and also in the same period the creations of memes have ocurred. Does that mean memes are responsible for the lowering of violent crimes? NO

I found Michael Nielsen's comments interesting and very smart.

Comments for this post are closed