0664: "Academia vs. Business"

[randall_out_of_my_head]

http://www.xkcd.com/664/

Title-Text: Some engineer out there has solved P=NP and it's locked up in an electric eggbeater calibration routine. For every 0x5f375a86 we learn about, there are thousands we never see.

This is a theme we haven't seen in a while...

[rho421]

Interesting but true. It's hard for one person's expertise to be appreciated - truly appreciated - by someone else.

[irishnut]

still remember when i was first learning to code, was taking a class at the time, halfway through the course i wrote a piece of code in like a third of the lines that wed been shown how to...teacher kept staring at it and running my program and never figured out how i got it to work...still one of my prouder nerd moments lol

[Omegaton]

Haha, so totally true. Gotta get them pubs!

Though unfortunately I don't know enough comp sci to get the title-text, I'm sure it's amusing for those who do.

[phlip]

Becuase I know it's going to be asked [edit]and, indeed, I was ninjaed by someone asking[/edit], the magic number in the title text is a reference to this.

[gmalivuk]

You apparently also don't know enough about the intricacies of using Google?

[hbsmiley7]

Mmm.. looking forward to the frustration.

[masher]

Though unfortunately I don't know enough comp sci to get the title-text, I'm sure it's amusing for those who do.

It's to do with the fast inverse square root finding algorithm in the Quake code...

[Felona]

I believe the "money" issue is poorly represented in the two outcomes

[tim314]

Randall wrote "0x5f375a86", not "0x5f3759df" . . . Is this a mistake?

[gmalivuk]

Randall wrote "0x5f375a86", not "0x5f3759df" . . . Is this a mistake?

No. The number in the title text isn't the one used in Quake (which was ...9df), but is actually an improvement discussed in this paper.

[masher]

Randall wrote "0x5f375a86", not "0x5f3759df" . . . Is this a mistake?

See http://www.lomont.org/Math/Papers/2003/InvSqrt.pdf . Chris Lomont finds that 0x5f375a86 is better than 0x5f3759df.

So, sounds like my original post was wrongish...

[swordswinger12]

I love this comic, especially since i'm considering grad school in compsci. Nice to know what the future will hold haha.

[Peripatetic]

The crucial detail missing from this comic is that only three of the papers will ever get read. Those papers will be read by a total of two people. One of them was just looking for papers to fill out the citations section of his dissertation.

[gmalivuk]

So, sounds like my original post was wrongish...

But only ish. It's still to do with a fast inverse square root algorithm. It's just an even better one than what's discussed there.

(Honestly, to be truly faithful to the theme of the comic, Randy probably should have used the original number. That's the one whose author is now unknown, and who may have just been some programmer working for a game company who came up with it and threw it into the code without any particular fanfare or recognition at the time. The a86 number happens to be better, but this was discovered and discussed in an academic setting, and we know who did it.)

[Terminalmancer]

Ah, innocence. In reality, it would end up something like this:

"You got the program to stop jamming up? Great. Did you write up a ticket for that so we can track it? No? Well, go do that first. When you're done with that, take it down the hall and run it past Nick. When Nick's done signing off on it, run it past Sue and get her sign-off. After you've incorporated their changes to your code, run it past all of us again and we'll have one last round of review."

"But... but... none of you even know how to program!"

Seriously, stay in academia if you can. Business isn't all terrible, but it often is.

[wsdenker]

You have beautifully summed up in one short comic exactly why it is that I don't program professionally anymore. Thank you for that.

[virati]

Funny, but these days its probably the opposite. The only thing Academian comic-strip-dude will have is a lifetime of headaches. Business comic-strip-dude will probably get a nice raise for fixing outlook.

[JustMe]

Am I the only one having trouble finding the comic? xkcd.com still goes to "sagan man". I have to put in xkcd.com/664 to see it.

[Q_<]

I found a more space-efficient integer knapsack algorithm while doing my homework the other week (undergraduate level). I told my professor what it could do and he told me I must have made a mistake (he's written papers on the subject himself). I ran it by the TA and he casually acknowledged that it was perfectly valid. No one seems to care though. Academia: where your brilliance will be recognized, but only if you're already well-known.

[DesertPilot]

Ho ho. But not even close to true. In academia he'd either 1) be too
busy dealing with administrativia to do any research, 2) the referees
would rip his paper to shreds and it would never get published, his
grant proposals would all get rejected as 'too innovative, but doesn't
contain any new ideas', so he'd be forced to move to Europe or China
to get funding

[Max2009]

Hehe.
I'm an undergraduate student in CS, and I could never understand my friends who want to go academia all the way.
Now I do.

[Izawwlgood]

Uh, rather jaded of Randall... A programmer friend of mine in business has been handed increasing amounts of responsibility for every accomplishment he's undertaken...

I think the difference could just as well read more like:
In Academia: Good work grad student, now as your PI I'll get credit and increased funding for this! Keep at it, gotta make sure you keep bringing in the papers to hold onto that 20k annual stipend! No, the student medical center still sucks.
In Business: You're contributions are valuable and to prevent losing you, we're giving you a raise. And a company car. And a department.

[Eketek]

Oh, my, another excellent comic -- I was finally convinced to make an account to post here!

Anyway, it mirrors my opions perfectly:

Excellent programming in concrete business -> more interesting assignments
Excellent programming in the abstract academic -> more documentation (blechhh!)


I'd rather program a phone than explain a program.

[markchd]

I love reading a comic where none of the readers even have to ask about P=NP. Gentlemen (and lady), we are nerds.

[Lewton]

Chances are the socalled beautiful code he wrote would be completely inappropriate in a business setting, Filled with obscure tricks, and hacks, being completely incomprehensible, and eventually ending up costing the company more in the long run, than was saved, due to maintenance costs

[Gerino]

Ah, the famous 0x5f375a86. It is probably the number, that pushed me to stick with CS. Such a crazy solution to a mundane problem, and it is the correct one - if it's not art, I don't know what is

[Magic Molly]

I love reading a comic where none of the readers even have to ask about P=NP. Gentlemen (and lady), we are nerds.

Actually, I only have a vague idea, and am too embarrassed to ask what it is exactly, as i'm too tired to understand the wiki page on it right now.

[westrim]

I love reading a comic where none of the readers even have to ask about P=NP. Gentlemen (and lady), we are nerds.

Well, I don't know what P=NP is, but I didn't ask because I don't feel a need to know and I doubt it will be terribly relevant to Asian History. The story behind 0x5f375a86 was interesting though, as I do play video games.

[jroelofs]

I found a more space-efficient integer knapsack algorithm

Does it trade space for time?

[phlip]

Well, I don't know what P=NP is, but I didn't ask because I don't feel a need to know and I doubt it will be terribly relevant to Asian History. The story behind 0x5f375a86 was interesting though, as I do play video games.

Yeah, the P vs NP thing is a matter of significant importance to computer scientists, reasonable importance to programmers in general, and potentially interesting to people who find maths interesting, but pretty much irrelevant to anyone else.

Suffice it to say (spoilered for long and irrelevant):

Spoiler:

P is the set of all puzzles which are easy to solve (for a very specific and broad definition of "easy")... NP is the set of all puzzles that, when you're given an answer, it's easy to check whether the answer is correct. Like Sudoku for instance... if I give you the answer to a Sudoku puzzle, it's pretty simple to check if it's right... just make sure it follows all the rules, of each number appearing once, etc. But if I give you a partially-filled in grid and ask you "is it possible to solve this?", that's a harder question to answer. The question is whether it's hard enough, given the aforementioned specific-but-broad definition of "easy". That is, is P=NP... is every problem which is easy to check also easy to solve? Or are there some problems which are easy to check but hard to solve? It's generally presumed that P ≠ NP, but it's never been proven.

[lulzfish]

I love reading a comic where none of the readers even have to ask about P=NP. Gentlemen (and lady), we are nerds.

I know what it is, but I hate calculus-and-above math, so I'd rather not understand the details.

[phlip]

I know what it is, but I hate calculus-and-above math, so I'd rather not understand the details.

There's nothing calculus-and-above in computational complexity... there's some stuff that's similar to limits, but that's as complicated as it gets.

[Kaijyuu]

Programming is plus, minus, and some bitwise operations. That's it.
From a higher language programmer's point of view, though, it's a lot of trig and arithmetic.

[thetrivialstuff]

I also have to say that in my experience, this comic is backwards, at least for undergrads (and at this point I'm getting so disillusioned with my university that I probably won't continue into graduate stuff).

At university, I slave away to get things done in unreasonable timeframes, and sometimes my solutions are great. Then, a TA takes three times as long to mark it, but that code never actually does anything useful in the world. They assign it a score and it gets thrown out because it's only an exercise. I slaved into the small hours, sacrificed sleep and sanity, and for nothing.

At my job, when I write some piece of code it's because we need a piece of code that does that. It may not be glamorous or brilliant most of the time, but it immediately becomes part of a *real* system that helps real people get real things done. And it sits there and functions, day after day, often for years afterwards. And I get paid to do it. And I usually have much more reasonable time frames, and I don't have to take the work home with me nearly as often. *That's* rewarding.

[ParticleBoy]

Spot on Randall

I've been there done that... three ways.

I started using IT skills in business. "Power user" = spreadsheet pivot tables. I had a nice suit, help generate great commercial success (profit) and received great monetary rewards... but it felt hollow and miserable.

I then did a stint as a student (MSc) then finally into research science. Power user = sim'ing transport of particles through matter

The first two were a good match for the comic. The third is where I truly felt rewarded for my work. Where coding guides avenues of investigation and furthers our body of knowledge.

My advice to young'uns: Happiness != Money

PB

[westrim]

Suffice it to say P is the set of all puzzles which are easy to solve (for a very specific and broad definition of "easy")... NP is the set of all puzzles that, when you're given an answer, it's easy to check whether the answer is correct. Like Sudoku for instance... if I give you the answer to a Sudoku puzzle, it's pretty simple to check if it's right... just make sure it follows all the rules, of each number appearing once, etc. But if I give you a partially-filled in grid and ask you "is it possible to solve this?", that's a harder question to answer. The question is whether it's hard enough, given the aforementioned specific-but-broad definition of "easy". That is, is P=NP... is every problem which is easy to check also easy to solve? Or are there some problems which are easy to check but hard to solve? It's generally presumed that P ≠ NP, but it's never been proven.

Oh. Hmm. Well, going from this line in the Wikipedia entry "In a 2002 poll of 100 researchers, 61 believed the answer is no, 9 believed the answer is yes, 22 were unsure, and 8 believed the question may be independent of the currently accepted axioms, and so impossible to prove or disprove", I'd go with the last 8, but simply because such a simple equation can't possibly account for all the variables.

But then, this is why I stay as far as possible from math with letters in it; after one too many exchanges of "why does it equal that?" "because it does, you are intended to know that type of equation and assume that value," I realized that math was as logical and straightforward for me as economics or sociology.

Aherm. Back to the point, P and NP both have nearly infinite variables, so formulating a rock solid equation is impossible. You mathy people will just have to accept that there are just some things that you aren't going to get, because they just aren't there.

[jonassimple]

Well, I may be wrong. But I think Heuristics in Business are sometimes considered real Solutions. I agree that an engineer out there has a polynomial heuristic for a NP-Complete problem that provides the right answer for 89% of its possible instances.

You mathy people will just have to accept that there are just some things that you aren't going to get, because they just aren't there.

Gödel already did!

[Plasma Man]

Good and true comic. In my experience (in business), the only reward for being competent at your job is that your job becomes to fix other people's incompetence.

[Waffles]

Interesting that this has thread has splintered into a 'hardcore maths' (Only hardcore maths people spell it with an 's' at the end) discussion and a 'which is less favorable: business or academia'. I'll add to the latter discussion.

I tend to find pure academia for its own sake trite and self aggrandizing, but business is a mistress who likes it very rough. I don't know how academia would treat this, but in business it tends to go something like:

1) "Oh, you want to save us 6 months and $500,000? We'll have to run it by some people who have no idea how either process works, what either process costs, and have no contribution to the process either way, but are in charge and have all the say."
2) "Well, we're not going to implement your drastic improvements to our product because I don't want to piss off a more senior member of staff than you who happens to be my chum. In a year you'll be swearing through your teeth and quitting because none of the ten great ideas you had within your first two weeks here were ever implemented, even though you fully coded and debugged half of them."
3) "We've discarded your redesign for a new and drastically improved product in favor of continuing to code the equivalent of a Woolly Mammoth that has been retrofitted as a wagon, then a sailboat, train, truck, and airplane, now into a space station."
4) "Oh, you materialized $200,000 in equipment out of thin air just by running around like a maniac and using a couple hours of other peoples' time? We'll wait until you install it before we lay you off, so we don't have to spend your salary of $30,000 per year. Thanks for saving us $500,000 in the two years you've been here."
5) "Great job team. Because of the drastic improvements in automation that Bob made, we can lay off half of you, and the rest of you can handle their work! Feel free to thank Bob on the way out for the 2 cent rise in your stock options!"

... All four of these are real examples that I have witnessed first hand - whether it happened to me, a coworker in my department, or a close relation. I'm not bitter at #4 at all. Of course not. And I'm not frustrated at all that they didn't give me four hours to polish a hack-job process improvement into something that any schlub could use, thus saving four invaluable guys three days about four times per quarter.