Cheap GPUs render strong passwords useless

[Asher]

Both of these are totally absurd. This reminds me of how Che was saying that he doesn't look up to KH anymore since he thinks quants are parasitic scum.

No, it's not absurd. Trust me.

I've worked at investment banks with PhDs in math writing code (like KH). Incredibly bright guys who wrote the worst software you would see in your life. Software development is a craft more than a field of math, they're not overly related. People who are strong in math tend not to be good at software development because they're largely relying on different strengths. A good software developer is imprecise and efficient, a mathy is anal and obsessive. A mathy will build an overly complicated solution with massive classes (or more realistically, procedural/functional programs existing inside a class...), while a good developer whips up a much simpler solution that solves the same problem in a cleaner design with less time.

[Hauldren Collider]

Not true at all.

Kuci is very strong at math and math theory. I've not seen ANY indication at all that he's strong at computer science. He certainly doesn't make a case for being strong in CS in this thread. His usual contributions in CS/SENG threads are to just ask questions.

Dude, math and math theory IS computer science. And it isn't some sort of ivory tower, don't-get-your-hands-dirty academic interpretation of computer science, it's genuinely useful in the real world. Remember the RNG thread, where you were reseeding and didn't believe KH or Kuci until you actually coded it yourself and discovered that you actually didn't understand RNGs after all? And that your first piece of advice to someone writing an RNG was actually a huge mistake?

Understanding how these things work is important. It's also important to be able to keep up with new technologies but it's a totally different set of skills to be able to pick up a new programming language in a day, learn a new operating system within a week, and generally stay with the times. Both are very important if you want to be a software developer.

[Hauldren Collider]

As for physicists and PHDs writing crap code, that doesn't surprise me. Clever people can accomplish so much through their cleverness that they never need to fully learn something before they can figure out how to do what they want. But because they don't learn things fully they do it the wrong way and use bad practices. That doesn't mean they're stupid, it means they're too clever for their own good and need more formal education in programming.

[Asher]

Dude, math and math theory IS computer science.

Computer Science is far more diverse than you know. You must take lots of math to be in CS, but when you're done with school? It's almost completely irrelevant and not at all useful. Real world CS is more practical, less iv

And it isn't some sort of ivory tower, don't-get-your-hands-dirty academic interpretation of computer science, it's genuinely useful in the real world. Remember the RNG thread, where you were reseeding and didn't believe KH or Kuci until you actually coded it yourself and discovered that you actually didn't understand RNGs after all? And that your first piece of advice to someone writing an RNG was actually a huge mistake?

It's not nearly as useful in the real world as you think. That kind of stuff is in math libraries you just use. 99% of your time you have no need for that kind of expertise or knowledge. Which is precisely why I never even thought about it before, because it just doesn't come up.

Understanding how these things work is important. It's also important to be able to keep up with new technologies but it's a totally different set of skills to be able to pick up a new programming language in a day, learn a new operating system within a week, and generally stay with the times. Both are very important if you want to be a software developer.

Yes, it's important to understand how it works. It's more important to know how to do it.

Kuci obsesses over the former, not the latter.

[Kuciwalker]

Except that's not true at all. The huge decreases in run time is significant, because now those 150M passwords stolen from a site can all be used pretty readily whereas before that'd be cost and technologically prohibitive. The scale differential is immense.

Bull****. Processing power is cheap. If you have 150M passwords and the operation to take advantage of them you've got a botnet or, hell, grab cloud computing time from Amazon or some ****. Even a hundredfold decrease in the cost of computing power (roughly what this is) doesn't significantly change the cost of getting those 150M passwords. And the actual three-hundredfold decrease in price represents only eight bits of extra password complexity you can handle, which isn't even two characters!

If the practical upshot of a development is "everyone should probably add a couple letters to their password, maybe" then it's not "immense".

You dismiss the progress of GPU computation outright because you either cannot comprehend the present applications, cannot envision future applications, or genuinely have no appreciation for just how much faster those computations can be. They've fundamentally changed how supercomputers are even architected in recent years. It is a big deal, whether you're capable of understanding that or not.

I dismiss it because it's massively overhyped and only gets us where we were already going a few years earlier.

Soon GPUs will be treated like FPUs are, as extension of the core CPU. I'm not even sure if you're old enough to remember the discrete FPU days.

Dude, that already happened two or three years ago. The result is that most people just have crappy Intel GPUs because they don't need better, not that everyone is now using their Radeon to run Folding@Home.

If you don't like it, don't read it.

I work in the industry and I'm seeing the change and capability GPUs unlock. You make ****ing spreadsheets. Know your role at the bottom of the totem pole, embrace it, and shut the **** up.

This coming from the guy who's had, what, four jobs in five years and is always the first one let go when the company starts losing business?

Not to mention none of the jobs you've worked at have seemed to require any real skill beyond generic codemonkey training. I remember the last time you posted asking for help with a cs "theory" problem for your job - when you were writing that crappy GPS app - it was something an undergrad should have been able to do.

[Hauldren Collider]

Understanding how things work and how to get the job done with off-the-shelf tools is the difference between an architect and a carpenter.

xpost

[Asher]

As for physicists and PHDs writing crap code, that doesn't surprise me. Clever people can accomplish so much through their cleverness that they never need to fully learn something before they can figure out how to do what they want. But because they don't learn things fully they do it the wrong way and use bad practices. That doesn't mean they're stupid, it means they're too clever for their own good and need more formal education in programming.

No, you don't get it. It's a different mindset and braintype. People think in different ways. Good programmers think in a certain way. Good mathys do not.

You'll find in CS academia a Software Engineering professor is the complete opposite of a Theory of Computer Science professor. There's a reason for that. It's not because the Theory guy is too clever for his own good, it's because he's ****ing smart in one area and ****ing stupid in another.

[Kuciwalker]

It's a good thing that real world CS genuinely doesn't need that much math talent, because otherwise Asher would be unemployable.

[Asher]

Understanding how things work and how to get the job done with off-the-shelf tools is the difference between an architect and a carpenter.

You're discussing an industry you don't understand.

[Hauldren Collider]

That's a little harsh, Kuci....

Asher: I won't pretend to understand the industry, all I've had are a couple internships, but I've observed enough people to realize that understanding is the difference between leaders and followers.

[Asher]

It's a good thing that real world CS genuinely doesn't need that much math talent, because otherwise Asher would be unemployable.

I'm not a math guy. Never have been.

And all of the math people I know make a fraction of what I do.

I'm certain I make far more than you do, too, with the same degree from a far less prestigious university. You're still figuring out the real world.

[Asher]

That's a little harsh, Kuci....

Asher: I won't pretend to understand the industry, all I've had are a couple internships, but I've observed enough people to realize that understanding is the difference between leaders and followers.

Yep.

I've got 30 people working under me in three different cities and two continents at age 27.

How many does Kuci have?

[Hauldren Collider]

Yep.

I've got 30 people working under me in three different cities and two continents at age 27.

How many does Kuci have?

Kuci graduated a year ago. Why don't you give him a little time?

My dad's an actuary and after 10 years or so led a group at an actuarial firm. He started doing law 10 years ago and now he's a practice group leader at a law firm. He understands what he's doing. He doesn't apply tools, he creates them.

I won't deny you have a lot of understanding about what you do, but in different respects. I'm sure you know C++ backwards and forwards. I'm guessing Kuci is, compared to you, a C++ padawan. I'm not even that much. But knowing the theory of computer science is just as important as experience.

[Kuciwalker]

Asher, by personal experience you don't want to try comparing actuary and programmer salaries on any kind of reasonable basis (e.g. lifetime earnings).

[Hauldren Collider]

Salaries aren't the be-all and end-all of what career you should pick, but in the career you DO pick, you have some choices you can make that will determine how far up the ladder you can go. If you limit yourself to applying tools, then you can only go up so far. There's nothing wrong with that, of course. Depending on what tools you're talking about, that can be quite high.