undefined | Better HN

0 pointsJensson4y ago0 comments

> That's obvious? Computers don't understand anything. They're just machines. What did you expect?

They do arithmetic's perfectly.

> You're going to tell me that if I want to use math, I have to let go of every preconception I had and crawl into Plato's cave, to see the light in the darkness.

No, math is a huge with an extreme amount of flexibility and power. You have lists of solutions like this one, you can add those in a program and then have the program look them up and spit out answers, with maybe some simple algebra testing:

https://en.wikipedia.org/wiki/List_of_mathematical_series

But nobody will be able to just program all solutions to all math problems like that, or give you a bible of everything to look up, as nobody knows what math is useful to you specifically. This is why every field that applies math tend to create their own subfield in math, like mathematical physics or chemistry or statistics or computer science. They create their own compendiums of useful notation and results that you can use. Nobody understands even a fraction of all of that and where it is useful, instead you will have to understand what discipline the paper you are reading is coming from, usually noted somewhere on the paper, and then look that up. For someone doing computer science you mostly look at the subfields statistics, probability, combinatorics and algorithmic complexity. Most of those papers will be written by people who do math mostly on the side though, not pure mathematicians. Pure mathematicians mostly works on problems that are too abstract for programming. (Pure vs applied, probability statistics and combinatorics are examples of applied math fields).

Edit: So from my perspective the main problem you have is that CS professors have yet fleshed out and formalized the subfield Computer Science math. When I studied higher level physics all the math was taught by physics professors, since mathematicians doesn't understand that math, it is mostly developed by physicists. Why doesn't CS do the same? You can't expect mathematicians to do that work for your field, as there are too many fields depending on math and too few mathematicians to handle everything.

0 comments

3 comments · 1 top-level

klibertp4y ago· 2 in thread

> They do arithmetic's perfectly.

So does abacus. Even something like this: https://www.youtube.com/watch?v=GcDshWmhF4A is capable of doing arithmetic. Does that mean that the marbles somehow understand what addition is?

> look them up and spit out answers,

Wolfram Alpha does this, and let me tell you: things I couldn't understand listening to lectures and trying to work through handbooks (well, most of them were hand-me-downs from the '60s, so take that as you will), became clear in just a few days of work. I didn't cancel my subscription for 2 years after that, out of gratitude.

But, most of the papers out there don't use interactive platforms for publishing. Even though they could. But as you said, that would be too much work, so they simply don't. I accepted it, it's all good now.

> Pure mathematicians mostly works on problems that are too abstract for programming.

I was pretty sure we're not talking about "pure" math. I mean, we both agree that that particular branch is of interest to a tiny minority. Of course the pure mathematicians can stay inside their ivory tower forever, nobody cares. Notice that I repeated the word "use" multiple times. I was trying to imply having a goal that's not mathematical in nature.

EDIT:

> Why doesn't CS do the same? You can't expect mathematicians to do that work for your field, as there are too many fields depending on math and too few mathematicians to handle everything.

I... don't know. That's a very good question. Yes, teaching math in CS depts is generally outsourced to mathematicians. It was like that for me. You may be right, that could be the reason to all my problems.

So, you mean that if we want to have math that's understandable for programmers, we need to make it ourselves? Fair enough if so, I think.

JenssonOP4y ago

> Wolfram Alpha does this, and let me tell you: things I couldn't understand listening to lectures and trying to work through handbooks (well, most of them from the '60s, so take as you will), became clear in just a few days of work. I didn't cancel my subscription for 2 years after that, out of gratitude.

Wolfram Alpha didn't program everything, but it includes most things you will see in an undergrad. In my undergrad we used a 500 page reference book containing most of the formulas and notations of STEM undergrad programs, wolfram alpha is basically that book written as a program.

> I was pretty sure we're not talking about "pure" math. I mean, we both agree that that particular branch is of interest to a tiny minority. Of course the pure mathematicians can stay inside their ivory tower forever, nobody cares. Notice that I repeated the word "use" multiple times. I was trying to imply having a goal that's not mathematical in nature.

But then your main problem are mostly with computer scientists or statisticians math papers, not mathematicians. The problem with papers written by people who apply math is that usually they don't really understand the math, they just apply it, so they can't explain it well. There isn't much you can do about that. And papers written by pure mathematicians, they sure do understand what they do, but it uses mountains of abstractions in order to make the problem possible for a human to reason about so it wont be easy to understand either.

For example, most who study statistics views the math as a tool to solve problems. They wont put in the effort to fully understand it, and can therefore just give you the second hand explanations they learned themselves, and the result is that those explanations likely will be missing a lot of things. Pure mathematicians however wont understand the way you want to use the math as a tool, so they might have a full explanation or the math but they wont understand how to apply it in your field, so their explanations will also be bad. What you need is someone who did the work to study the pure math side, and then did a lot of work applying math, and then explain it. Such people are extremely rare though. I went that route, I knew a few others who did, but most of those I know didn't. They either went pure or applied, not both.

klibertp4y ago

> wolfram alpha is basically that book written as a program.

Exactly! Which is why I love it, and why I think it would be good to extend it to encompass more than just that one book, too.

> The problem with papers written by peo I went that route, I knew a few others who did, but most of those I know didn't. ple who apply math is that usually they don't really understand the math, they just apply it, so they can't explain it well.

This is already a third time you said something that made me really think again about the issue...

It sound very plausible. I think I've probably never even seen a proper maths paper. I've been struggling with math, but it didn't occur to me that a part of the reason (other than me being dumb) was that the authors struggled with it too! Were I in their place, I would definitely also try to reduce the amount of explanations to the absolute minimum, so that there's less of a chance I'll be called out for it.

EDIT:

> I went that route, I knew a few others who did, but most of those I know didn't.

I'm not joking, this is very serious: is it possible to learn from you somewhere/somehow? Because you're incredible, honestly, your patience is not running out even now, this deep into comments thread; and you really can speak in programming - this much I can see - and I think you're at least as competent on the other side (though I obviously can't judge it myself, so I'm deferring to (your) authority that I came to believe in). I'd be incredibly happy if I could somehow get a bit of your help.

(Seriously, I'm not sarcastic or anything, this is 100% honest. EDIT2: well, I still disagree with you on some points :P but I think if it's from you, I wouldn't have a problem with actually listening to you)

1 more reply

j / k navigate · click thread line to collapse

0 comments

3 comments · 1 top-level

klibertp4y ago· 2 in thread

> They do arithmetic's perfectly.

So does abacus. Even something like this: https://www.youtube.com/watch?v=GcDshWmhF4A is capable of doing arithmetic. Does that mean that the marbles somehow understand what addition is?

> look them up and spit out answers,

> Pure mathematicians mostly works on problems that are too abstract for programming.

EDIT:

> Why doesn't CS do the same? You can't expect mathematicians to do that work for your field, as there are too many fields depending on math and too few mathematicians to handle everything.

So, you mean that if we want to have math that's understandable for programmers, we need to make it ourselves? Fair enough if so, I think.

JenssonOP4y ago

klibertp4y ago

> wolfram alpha is basically that book written as a program.

Exactly! Which is why I love it, and why I think it would be good to extend it to encompass more than just that one book, too.

This is already a third time you said something that made me really think again about the issue...

EDIT:

> I went that route, I knew a few others who did, but most of those I know didn't.

1 more reply

j / k navigate · click thread line to collapse