Hacker News Reader

Comment by sombremesa | Hacker News Reader

Anecdata, but I really don't think thinking rate is fixed. An easy example is learning a new language - at the beginning you will have a lot of trouble expressing yourself as you grasp for vocabulary and struggle with grammar, but over time these tasks become second nature and you'll be able to think and express yourself much faster.

You can argue that your thinking rate hasn't changed, you've just internalized some concepts and no longer have to actively think about them, but that seems like a disingenuous argument. You could certainly repeat this process for other disciplines and indeed 'think faster' for all intents and purposes.

Of course, this is a clickbait article and I'm wasting my time even responding to this, but I had a visceral reaction to this that I felt I shouldn't ignore.

View on HN

> You can argue that your thinking rate hasn't changed, you've just internalized some concepts and no longer have to actively think about them, but that seems like a disingenuous argument.

I think this is an important distinction actually.

For instance let's take the example of choosing a tool for a particular job. Over time you teach yourself that tool X is best for a particular type of task, so any time you are faced with that type of task you automatically pull put tool X, without thinking about it. On the surface it looks like you're thinking and deciding very quickly but really you're just taking a shortcut.

Eventually you are faced with a task that looks on the surface like the type of task you use tool X for, so you start working. After a while it turns out that it would be best to use tool Y. Now you've wasted a bunch of time using the wrong tool because you used a shortcut to avoid thinking, instead of taking the time to think.

It's a dumb example but I think it illustrates the difference between "thinking faster" and shortcutting your thinking to reach a conclusion faster.

The reason this is a disingenuous argument is because it endeavors to restrict the definition of 'thinking' strictly to cognition - or further still, to cognition in the absence of recognition, which is actually quite absurd if you think about it. However, by definition thought has no such restriction - recognition and spontaneous ideas should hold as much weight, and can in fact speed up the process of cognition itself [0].

> The difference between "thinking faster" and shortcutting your thinking to reach a conclusion faster.

I don't see any difference between these two things - perhaps you can elaborate. In my opinion if you can shortcut your thinking to reach the same conclusion faster, you are in fact thinking faster for all intents and purposes.

[0] https://www.youtube.com/watch?v=9Zm9sTfWnQ0

I think a definition of thinking close to how the article is using the term is more like:

Thinking is a process of considering options and reaching a conclusion.

Using that definition, or something like it, then I think it's pretty evident that decision making shortcutting is an absence of thinking.

> Thinking is a process of considering options and reaching a conclusion.

Okay, let's take this definition of thinking as our basis. Presumably, a chess grand master is able to do this faster than a novice.

However, by your own argument, you're also calling this an absence of thinking. So, which is it?

A chess grandmaster knows better than to consider every option to the same depth. Pruning is choosing what not to think about. I wouldn't expect grandmasters improve by increasing their positions evaluated/second throughout their career.

As an amateur chess player (2000/rapid/lichess) I wouldn't say the speed or amount of thinking I do has changed as I've gotten better in terms of the number of moves I evaluate. In fact it feels like I frequently think less now than when I was worse.

My brain has compressed/chunked patterns so that I can immediately see what the optimal move or set of candidate moves are instead of having to determine them more computationally.

I'm not sure how that fits into the framework of this article. Something like the number of calculations remains static but total work done per calculation increases.

Much of the debate in this comment thread is a result of these concepts being fuzzily defined in the article itself.

You memoized some of the computation.

"Think rate is fixed" glosses over activities with compounding returns. Learning!

Better abstractions and heuristics can have an enormous impact on problem solving efficiency.

First-principles thinking is a powerful hammer, but it's expensive. Over-use it and you'll get nothing done. Under-use it, and you'll do nothing original. Knowing where to use it is the trick, and that's _also_ a skill that can be improved over time.

Chess GMs certainly prune better, but can also evaluate moves faster because they can easily hold multiple boards in their head. It's common for strong GMs to play 10+ blindfold simul games. Nakamura solving chess puzzles is a sight to see.

How is knowing what to prune not thought?

Sometimes it is, then it would be thinking in this context. Other times it's pattern matching which isn't analytical thinking.

I’m still having a hard time understanding the distinction. How is pattern matching not thinking?

This distinction is captured in "Thinking Fast and Slow"

https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow

Doing less thinking to reach a conclusion quickly isn't the same thing as thinking faster.

It's a meaningful distinction because there are physical and mental costs (calories and stress) to trying to "think faster" that don't accrue the same when making decisions based on learnings.

You've abandoned "thinking is a process of considering options and reaching a conclusion" now.

> making decisions based on learnings

This is the ONLY type of thinking there is. Ergo, you can 'think faster' for any given discipline with practice.

> You've abandoned "thinking is a process of considering options and reaching a conclusion" now

This is a big claim to make without any explanation of why you think that.

I'm forced to conclude you're not discussing in good faith. Bye.

If the process of considering options and reaching the correct conclusion concludes faster and that is our definition of thinking, it should follow that the thinking is faster. Somewhere in your last comment you must've pivoted your definition, because you described it rather as thinking 'less' and saying that it is NOT the same thing as thinking faster. That's perfectly fine, of course, but it left me confused as to what definition you were now using.

It's a shame you've decided to abandon the argument.

> If the process of considering options and reaching the correct conclusion concludes faster and that is our definition of thinking, it should follow that the thinking is faster.

That doesn't follow because once you've mastered something you consider less options to reach a conclusion than a novice. One reaches the correct conclusion whit less thinking, not faster thinking.

Following your logic a person taking a shortcut from point A to point B is doing the same amount of driving as somebody not taking the shortcut, they are just driving faster.

Think of it like simple decisions. X=>Y

You start with some of them: A=>B, C=>D, B=>C, E=>F, B=>G. You now get A, and need to decide which tool is better: D or F. You start the reasoning: A=>B=>G&C=>D. After 4 reasonings you arrive at the solution: D. If you need to do this reasoning several times, you 'learn' A=>D. Now, when you need to make the same reasoning, you do A=>B&D. You get the same output but twice as fast. You 'learned'. Muscle memory is also a very good example of this.

The objection is that spending more time thinking about something doesn’t guarantee more accurate responses. Often people’s first reaction is correct and they second guess themselves into making the wrong choice, thus so called shortcuts are just as valid as any other kind of thinking.

For the muscle memory example try really paying attention to walking and your going to be less stable.

> Eventually you are faced with a task that looks on the surface like the type of task you use tool X for, so you start working. After a while it turns out that it would be best to use tool Y. Now you've wasted a bunch of time using the wrong tool because you used a shortcut to avoid thinking, instead of taking the time to think.

I'd caution that for many tasks, using a close enough tool can be good enough, and an experienced user of a "close enough" tool may very well get results rapidly.

Regular Expressions are a great example of this. I've done stuff with regular expressions that a proper parser or tool modifying ASTs would be "correct" for[1], and if I learned to use those specialized tools I'd be better off, for that one particular task. But the fact is regular expressions serve to sub-optimally solve a lot of different problems. I'd make the argument that the total time I've spent learning regex's and then solving a lot of problems with them at 80% speed has been a more efficient use of time than if I'd learned a bunch of one off "ideal" tools.

[1] I have a ~bad~ habit of generating large repetitive chunks of code using regexs.

Agreed, you can see it as operating on a compressed version of the original domain. Your symbolic reasoning may not be faster but your symbols change, through your own dictionary optimization.

> It's a dumb example but I think it illustrates the difference between "thinking faster" and shortcutting your thinking to reach a conclusion faster.

As in the old joke:

Q: What's the difference between a novice and an expert?

A: The novice thinks twice before doing something stupid.

I think your instinct is both correct and missing something. At some level, 'thinking' involves transmitting information about the world. Suppose you had a poor model of a new task - the code used to encode information would necessarily be inefficient, since you don't know the probability distribution of anything related to the task. Once you do the task for a while, you get a better sense of the probability distribution of stimuli, and (in theory at least) could develop a more efficient code. Adaptive Huffman coding does exactly this - builds codes on the fly that gradually get more efficient. It's possible that the brain does something qualitatively similar. If so, it would 'look like' information transmission is happening faster.

From this perspective, 'information transmission rate' could actually be constant during learning, when defined as a reduction in subjective uncertainty over time. But because you start with a poor task model, your initial uncertainty is higher when starting learning, due to unfamiliarity with the task, so to an outside observer it looks like information transmission is slower. As your task model improves, your prior for each message matches the environmental statistics better, and information transmission speeds up -- again, just from the perspective of the observer, as your subjective uncertainty now starts low and just goes lower.

Well said. Your comment reminds me of a conversation from 30+ years ago with a work colleague who opined - or shared the opinion, I cannot remember which - that our "input bit rate" was fixed. Young and argumentative, I disputed this, but....

I don't remember all of the arguments in favour, but he pointed out reading: We start sounding out individual letters and eventually progress to scanning whole words or sentences or paragraphs.

It's not that our bit rate changes, it's that our "bit" changes.

Yeah, handwavy I know, I am struggling to recall the conversation.

Essentially, though, consider your point another way: We can only process so many concepts/notions per unit time, and initially, in any subject, our concepts/notions are smaller. As our experience grows, they grow, so our throughput increases, even at the same processing speed.

Sort of like widening a highway to handle more volume at the same speed. Sort of.

In machine learning lingo, you'd have better 'embedding vectors' for the world at large.

This is what Barbara Oakley (of the popular Coursera course "Learning how to Learn") calls chunking. I think of it like changing the word size based on the topic and my understanding of it.

I tend to be a proponent of the resource model of cognition. E.g. you have a limited budget of willpower in a given day, and when you run out of it, it becomes much harder to resist that piece of chocolate cake in the fridge.

By the same token, I would find it plausible to believe you have a certain budget of decisions to be made, or problems to be solved in a given period of time before quality goes down.

I think it's likely this is also something like a muscle that can be trained over time.

This is known as decision fatigue - https://en.wikipedia.org/wiki/Decision_fatigue

It leads to things like "people who tend to have to make important decisions have fewer clothing sets" ( https://www.businessinsider.com/successful-people-like-barac... )

I would contend that this is part of the reason why it is difficult for people who have to make decisions about "should I pay the rent, groceries, or utilities" to be able to join the white collar jobs - as those are about making decisions. If your will power budget is exhausted on food choices (because you don't have much money), its harder to make good white collar job decisions.

I would furthermore contend that the above is an excellent argument for UBI (though this gets deeper down into politics).

> If your will power budget is exhausted on food choices (because you don't have much money), its harder to make good white collar job decisions.

Don't you think this has a lot more to do with access? I can't imagine there are too many people saying: "I will turn down this high-paying management job because I spent my decision budget at the grocery store"

Outside of Trading Places (with Dan Aykroyd and Eddie Murphy), you don't often get the person being offered the high paying position without experience.

Access, I believe, is one part of the issue. I'm thinking more of the "my willpower budget is spent on the grocery vs utility bill" leading to other poor choices that make it more difficult to get a promotion that allows for a more care free life (not worrying about bills).

I believe that there are many parts of social and economic feedback loops that make it harder (and continue to make it harder) for people in environments that are willpower draining to move to jobs where the ability to make good decisions is the key value.

Well thought demonstrably takes effort, a ridiculous amount of your bodies resources is spent on thinking, however it's hard to pin this down in a certain 'rate'.

You could probably show that calories and certain other resources get exhausted by concentrated thought, however in my experience this has little to do with speed of thought and everything to do with depth (the only times I truly felt mentally exhausted was when I was trying to dig up deeply buried fragments of memory to solve a maths problem).

>You can argue that your processor's clock speed hasn't changed, the program has just memoized some intermediate results and no longer has to actively compute them, but that seems like a disingenuous argument.

A disingenuous argument indeed. Almost as if you took the entire premise of the argument you disagree with and put it inside a "you can argue" block which is then dismissed with no rebuttal.

The man who does not read has no advantage over the man who cannot read - Mark Twain

https://news.ycombinator.com/item?id=26308324

> Of course, this is a clickbait article and I'm wasting my time even responding to this, but I had a visceral reaction to this that I felt I shouldn't ignore.

Hey me too! I'm _phenomenally_ more creative and productive when "in the field" doing experiments, than when I'm at home / at work trying to predict problems and come up with novel solutions. I'm definitely thinking faster and better when equipment is on the line and stakes are higher.

I really do believe I am more _engaged_ in a problem when it's urgent, expensive, and imminent, rather than very far away. Engagement produces better results for me.

Maybe that's because I'm not thinking faster, but I'm thinking with a higher duty cycle on _this one problem_, but that's not a meaningful, operable insight. Like everything on fs that somehow makes HN.

I think the key is to recognize what situations help you be productive, then engineer more of those situations into your life. You don't need to squeeze your brain harder, you need to be in the field more.

> You can argue that your thinking rate hasn't changed, you've just internalized some concepts and no longer have to actively think about them, but that seems like a disingenuous argument.

The first thing I learned from a second human language teacher was that I'd know when I was becoming fluent-ish when I had a dream in Spanish. At least for second-language teachers, internalizing concepts and not thinking about them is a sign of learning progress. It doesn't seem so disingenuous to me, having internalized a lot of things that felt unfamiliar before and having experienced proficiency improvements as unfamiliarity melted away.

Huh. I've had dreams "in Russian", Greek, Latin, Japanese, and Arabic. I know none of these languages beyond a little counting or a few words, but they seemed very real at the time!

I still think that being able to crack an improvised joke based on the situation one is in is a stronger sign of fluency. But it rests on similar assumptions: having internalized the basic structures of the language to the point where you can spend your cognitive resources on execution.

"Chunking" is an important concept here.

Think about learning to read in.a script you're unfamiliar with (eg Cyrillic if your native language uses Roman script). You need to learn the shape of the letters so you have to keep in mind multiple things just to process one letter.

At some point you're familiar enough with the script that each letter becomes one "chunk" that you can recall and process mentally. This stage is equivalent to sounding out a word by sounding out each letter.

At some point each word becomes a "chunk". Fun fact: they have done tests on reading and if you scramble the letters in text but anchor the first and last letter of each word, most fluent readers can still read the text. [1]

After that groups of words become "chunks" but the barrier here becomes more nebulous.

My point is that the number of chunks your brain can process at a time is relatively fixed (around 5 IIRC).

So I'm not sure that language learning is the counterexample you may believe.

[1]: https://www.treehugger.com/why-your-brain-can-read-jumbled-l...

I think you and the parent are arguing the same thing... you're just coming to a different conclusion (whether this means that thinking rate is fixed or not.)

> You can argue that your thinking rate hasn't changed, you've just internalized some concepts and no longer have to actively think about them, but that seems like a disingenuous argument.

So the number of "chunks" may be fixed, but you are able to absorb and think about the text quicker.

The brain memoizes learned concepts just like computers often using a LFU-like algorithm at least my brain seems to

As Alan Kay put it: "A change in perspective is worth 80 IQ points."

I teach English as a second language. Just an anecdote to add here, there is a lot more background processing than you would imagine here.

Most learner's are trying to parse the new language into their old, and discount context (understandably with English), and that sudden burst your noticing, is that realisation.

I could go on for quite a while about this, but my main point is: learning a new language is a whole bunch of background processes that most people don't realize.

be careful of the assumption.

Humans when they are born can understand 10,000 word sounds in multiple languages but as newborns approach the age of 3 their brains delete the un-sued neurons in the language area of the brain...it's why it's harder to learn languages as we age.

I don't think it's harder per se. But 1h a day (next to work and other stuff) for a new language doesn't compare to 8h+ as a child.

Anecdata: I've had this happen to me. In travel for an old job, I've had the pleasure of spending time in other countries with foreign languages. Any prep study in those languages absolutely paled in comparison with being forced to use the language all day long. It was especially hard when I was tired from a long day of work. However, I now know that was the most important time for me. Those tired moments when I just wanted wine or beer or a taxi ride and didn't want to speak the language anymore but had no other choice were when I really cemented the language in my head. It's incredibly difficult, and expensive, but I really feel that living in the language for a few weeks is worth the effort and time, much more so than any course. That said, take the course for a few months before you get on the ground, it really helps.

In that story, isn't the cognition time the same with a bottle neck at the audio output? That would be like saying your computer's CPU is slower because your internet upload speed is low.

You must be running better software than me, last I checked my tongue cannot be actuated independent of my CPU.

For that matter, you can certainly write software that has poor upload speed that improves with processing power, regardless of your internet connection.

We all use slightly different forks from the same software.

I must be running my grandmother's fork, as she taught us: "Turn your tongue seven times before speaking. This way you'll have time to think if you ought to say the things you want to say.".

I think first and then I produce a whole sentence or paragraph. My text or speech is not improvised word for word as I progress. There is a small delay when I convert my French thoughts into English, but that's it.

Reply on news.ycombinator.com