Hacker News Reader

Comment by orangepurple | Hacker News Reader

Improving your own brain with memories and knowledge is like greatly increasing the size of a CPU L1 cache and the improvement in execution performance is similar

View on HN

CPUs don't have logic or reason, so it's not a good analogy. E.g. you can replace an infinite set of memories by distilling them into a few bits of knowledge that underpin that given set of experiences.

I don't think an analogy is meant to be factually equivalent, it is meant to convey meaning. I perfectly understood the meaning of the parent comment, even if it is technically inaccurate.

We aren't trying to understand minds or caches, we're trying to encourage a practice that has little to do with the inner workings of either.

Good point. Reminds me "All models are wrong, but some are useful"[1].

[1]: https://en.wikipedia.org/wiki/All_models_are_wrong

CPUs have logic, and programs can do reasoning. And that's now how the brain works (if you mean generalization).

Analogies are almost always flawed but they are powerful tools for forming strong associative memories.

They are a starting point for encoding new knowledge in a custom symbolic language known only to yourself. It is the basis for all learning.

Some other major tools for learning are mnemonics and spaced repetition.

Fair point, but on the other hand: I need to format strings and use basic regular expressions at least a couple of times per week and the execution performance increase and convenience from finally having learnt those by heart (simply by using them enough) is definitely worth it for me and not at all similar to having to look them up each time.

We are saying the same thing with different words

Sure but if you use like five languages then the differences between s.length len(s) and length(s) will blur when at the beginning of the day you need to refresh the cache for todays language.

i could memorize them all but it is less interesting than reading papers about optimizing algorithms for cloud economics or other more valuable (to me) ideas.

Why are you using five languages? I mix them up when I switch, but I rarely work in more than two languages at once, and keeping two separate isn't that hard.

Context switching between Java, SQL, Javascript and a build scripting language isn't _that_ unusual an experience. Throw in a templating language there and you're at 5.

When I'm working in my preferred environment, I need to work in its language, but the language/VM is built in something (usually C), and the kernel I'm running on is built in something (usually also C), and I might be making a webpage where I need a bit of Javascript. And sometimes I need to poke at a shell script or something that's overgrown shell and is now Perl. And sometimes my preferred environment isn't really the right fit for the problem at hand, so I've got to use something else. Oh, and maybe I need to compile stuff, so here comes Make.

For sure I always have to Google 'make' the first two hours of using it again.

> Improving your own brain with memories and knowledge is like greatly increasing the size of a CPU L1 cache

Yup... and as the size of L1 cache grows so does its latency. Eventually you reach the latency of main RAM and then... wait I forgot why you increased L1 cache size. Can you just reset it to "fast" please?

Love the analogy.

Reply on news.ycombinator.com