/sci/ - Science & Math

[Anonymous No.6524600]

All trolling aside, is Computer Science worth one's time to study anymore?
I'm not necessarily talking about the tech bubble and money to be made, although that's one of my preoccupations, I'm also not sure if CS will be central to technological innovation in the near future. Will it, in your opinion?

[Anonymous No.6524606]

>>6524600
There is no tech bubble when one programmer can put 10 other people out of work

[Anonymous No.6524610]

>>6524600
>All trolling aside, is Computer Science worth one's time to study anymore?

On <span class="math">your~own[/spoiler] or only in graduate school. It's completely retarded to go to university to major in it. (Think of it as majoring in Spanish, sure it's useful but why learn it in a university)

>CS will be central to technological innovation in the near future.

You've been listening to far too major CS majors if that sounds at all plausible.

[Anonymous No.6524622]

>>6524600
>is Computer Science worth one's time to study anymore?

4 years of university spending tens of thousands, hell no. 1 year of free/cheap self-study, yes if you're interested in theory. Just follow a curriculum and read about everything you're interested in (see http://www.scotthyoung.com/blog/mit-challenge/ )

CS is shockingly easy to learn on your own.

[Anonymous No.6524633]

>>6524622
>>6524610
By "worth one's time to study", I essentially meant getting into the field itself, regardless of the cost of the curriculum, that isn't important.

I want to know if CS has a bright future, in terms of technological advancement and innovation. I'm aware CS is broad though, I'm guessing only specific subsets are going to truly be important in the coming years.

Although I'm curious to know if the "concrete" part of CS (aka. software engineering, networking, all the Silicon Valley development shit) will die out soon.

[Anonymous No.6524645]

:(

CS isn't about technological advancement and innovation ... CS is math :(

The best use of CS imo is for use in computational science, which is mostly an engineering field. It mostly differs from CS in that it is based on calculus, whereas CS is based on Discrete Mathematics.

[Anonymous No.6524648]

>>6524633
It has a VERY low barrier of entry. Pretty much anyone with an IQ of about 70 can learn and do it.

>technological advancement and innovation

Unless you classify "angry birds" and "metro" as "technological advancement and innovation" you're thinking of the wrong field.

[Anonymous No.6524650]

>>6524645
>CS is math

Stop spreading that lie.

[Anonymous No.6524653]

>>6524648
>very low barrier of entry
Sorry if my question sounds stupid but how does that correlates with what a field can bring to the table in terms of innovation?
To you, the fields with the highest barrier of entry are the most likely to contribute to humanity's advancements? What are they, by the way?

>you're thinking of the wrong field
I wasn't necessarily thinking about coding apps and websites.

[Anonymous No.6524655]

>>6524650

pure math troll is back! you always post the same idiotic community college-tier discrete math book.

Can you name a single university who uses that textbook? No , no you can't.

PS: Look up Global Rule 3 about posting trolls outside /b/ http://www.4chan.org/rules

[Anonymous No.6524674]

>>6524653
Why the study of algorithms and theory of CS <span class="math">is[/spoiler] interesting in itself. The majority of software development is mindless coding with basic algebra being the most advanced topic required. More technical fields aren't accessible with just a mere bachelors study with no experience so CS undergrads are 'mostly' limited to code monkey tasks.

Just because they are using computer doesn't make them responsible for innovation brought by computers.

>I wasn't necessarily thinking about coding apps and websites.

That's pretty much all you're qualified for with just a BS in CS. Go look at some job posting and note the education and experience requirements.

>the fields with the highest barrier of entry are the most likely to contribute to humanity's advancements? What are they, by the way?

If high/middle schoolers can easily do it, then someone would have already done it and advanced humanity. If you really really want to advance humanity then you have to study material science, solid state physics, engineering, or mathematics at the graduate level.

[Anonymous No.6524677]

>>6524674
I wouldn't want to just get a BSc anyway.
What kind of things would I be able to do with an MSc in CS? Would I get much more opportunities with a PhD than with a master's?

>engineering
That's very broad. What do you think are the engineering fields that are going to be the most interesting in the coming years?
>mathematics at the graduate level
I might be talking shit here, but at the graduate level, doesn't CS also imply a good understanding of pure math?
What would a mathematics MSc/PhD bring to the table additionally?

[Anonymous No.6524680]

>>6524674
Not the guy you answered to, but it'd be nice if you could explain what kind of groundbreaking research in materials science is being done right now?

[Anonymous No.6524682]

>>6524655
>Can you name a single university who uses that textbook? No , no you can't.

It sells the point. Graham-Knuth-Patashnik, Rosen, Epp, Biggs, or whatever other book you look at is are trivial as fuck and waters down the topics they cover to the point were they just provide definitions and nothing else. You don't learn any math in a discrete math book.

PS: this is /sci/ - "Science and MATH", not "CS and career/code advice". gbt >>>/g/ and never come back.

[Anonymous No.6524685]

>>6524677
You're better off doing anything other than a BS in CS before doing a MS/PhD in CS.

[Anonymous No.6524687]

>>6524685
Why is that?
Wouldn't it be easier?

[Anonymous No.6524688]

>>6524680
Every time there is a die shrink, there's a break through in material science/engineer.

[Anonymous No.6524696]

You guys need to distinguish between

>>6524648
>It has a VERY low barrier of entry. Pretty much anyone with an IQ of about 70 can learn and do it.
Programming

and
>>6524645
>CS isn't about technological advancement and innovation ... CS is math :(
Actual computer science theory.

Any tool can be a codemonkey. You honestly need to know math for theoretical computer science: http://en.wikipedia.org/wiki/Computer_science#Theoretical_computer_science

[Anonymous No.6524708]

>>6524696
>You honestly need to know math for theoretical computer science

>Implying you can do that with just an undergrad in CS

top kek

[Anonymous No.6524709]

okay so let's be clear about something before i get started:
/sci/ has extremely fucked-up memes regarding the relationship between software engineering, cs, and math that i have *never* seen elsewhere, in academia or industry. i'm a 6th-year csphd at a decent institution who has both taught and published, so i have at least some shitty experience; and i think that most of what /sci/ says about cs has no grounding in reality

particularly funny bullshit includes:
>graduate level cs research is all "cs theory." this is "actual" computer science
>corollary: a math degree prepares you better for a csphd than a csbs
>everything that is not "algorithms and theory" is software engineering
>corollary: because a csbs doesn't teach "pure mathematics", it at best prepares you to be a software engineer and at worst prepares you to be nothing
>of course computing technology has bought about great advances (to argue otherwise would be inane), but all these advances can be traced back to mathematicians and computational scientists pushing the boundaries of numeric computing or w/e
this shit is outer space lunar and while i'd normally be happy to explain why, it's fucking exhausting given how divorced from reality /sci/ is about these things. we'll see how much endurance i have. but i first want to establish as a premise that these threads are about as goofy as climate change denial threads and i take them about as seriously

[Anonymous No.6524736]

>>6524709
so
>graduate level cs research is all "cs theory." this is "actual" computer science
>corollary: a math degree prepares you better for a csphd than a csbs
computing is a natural phenomenon and can be studied empirically. this is what you are doing when you benchmark, whether that's time, space, or precision and recall.

this is easily 50% or more of what is done in typical CS research. there are many reasons for this. one reason is that algorithm theories have proven to often be very bad at predicting the behavior of algorithms on real machines (my goto example is triangle counting on terabyte+ graphs). this has driven advancements in algorithm theory (e.g. streaming algorithms) but the gold standard for measuring computing behavior is still empirical. even a theoretician will almost certainly benchmark at some point because it's a requirement for publishing in numerous venues; in particular conferences, which are favored over journals for most cs research

another reason is that the problems we often want to solve in cs are not theoretically tidy. big data analysis is a good example of this. often the properties of an algorithm depend on statistical properties of the input (e.g. dimension) that are not necessarily well-characterized, either because the input or the space of typical input is not well-understood. this can lead to algorithms that behave differently than predicted for *the kinds of input they typically run on in-field*. ultimately input-data gathering and analysis actually informs the design of systems and so data measuring, as in biology, has its own inherent value to the science

[Anonymous No.6524740]

>>6524736
if you're designing/hypothesizing and benchmarking different triangle counting strategies for different NUMA architectures then what you are doing doesn't resemble "pure CS theory" in the slightest. same goes if you are trying to develop a low-power distributed network of things, same goes if you are trying to develop a next-gen non-relational DB model. this is not applied engineering because presumably no one has ever tried anything like what you are doing and you don't have any expectations; so you have to find out, you run experiments. maybe for the DB you drag in the language people and the compiler people, but at the end of the day the PL theory has to produce queries that perform on a real machine and if they don't benchmark as expected then the hypothesis is falsified.

this is most of what is done in actual CS research. statistics will probably be useful no matter what you do, but what you really need (as in any scientific discipline) is familiarity with the literature and 90% of said literature is *not* mathematical in character. if you're limiting yourself to "pure math CS" you're limiting yourself to a subset that's at best half the published literature, and I think that's generous

[Anonymous No.6524756]

>>6524648
>Unless you classify "reusable space rockets" and "robot quadrupeds that won't fall over even if you kick them" as "technological advancement and innovation" you're thinking of the wrong field.
Your failure to understand what programmers are doing in the world doesn't invalidate it.

Programming is like writing, drawing, music, or math: anyone can learn to do something with it, but few can do it for a living.

[Anonymous No.6524763]

>>6524736
>>6524740
k, I like your elaboration and the point you make.
PS: Coming from a science theory background, with an interest in philosophy and what you probably call "pure CS theory", what you describe is pretty off-putting. I don't see a way no not consider "benchmarking" the boring part. That's like if you force a scientist to do engineering work - every monkey can do it if he puts in enough man-hours, no?

[Anonymous No.6524784]

>>6524740
i'm going to skip the middle two because they seem somewhat redundant and i'm bored
>of course computing technology has bought about great advances (to argue otherwise would be inane), but all these advances can be traced back to mathematicians and computational scientists pushing the boundaries of numeric computing or w/e
i think this misconception exists because the field has only been academically distinct for a couple decades, so when you look at founding heroes such as dijkstra you see someone who "began with physics" even though he researched cs most of his life. i think also that /sci/'s hero narratives are tailored towards theoreticians and that they don't necessarily talk about guys like ken thompson

at any rate, i know very few graduate students who don't have a cs undergraduate degree, and when we're looking for undergraduate cs researchers we're looking in the cs degree program. such a degree qualifies you to start reading papers in your areas of greatest experience, which is what most of being a researcher entails. this is one thing you can do with / one reason we teach all that "30 year old" stuff in cs undergrad; it's so you can read a contemporary paper and know what the fuck the terms mean. this is how science is taught, it's not weird

maybe "pure math cs" terms are easy for a mathematician to grasp (i would assume so) and/or maybe a mathematician can make a focused study to grasp the terms of any particular cs topic area (i would assume so). but for many of those topics you won't be using that math degree, except maybe to crunch statistics for experiments. so i don't know why you would waste your time making a study of math unless you were specifically interested in a mathematical research topic, which as i've hopefully established is only a subset of cs

[Anonymous No.6524789]

>>6524763
benchmarking is excruciatingly boring, but running experiments in general is excruciatingly boring in many fields. my neurobiologist friends run experiments in which they torture animals for 36 hours, do you think any of them want to do this and consider it the fun part of science?

the fun part of science is research study and hypothesis formation. you read a ton of things, you get all the facts inside of your head, and at some point if you understand them fully you will begin to produce lines of inquiry. you do informal preliminary experiments for your own edification, but at some point if you want to publish you have to worry about statistics and reproducability and yeah, that's pretty fucking boring. but i'll laugh my ass off if anyone on /sci/ wants to argue that this method is unimportant

[Anonymous No.6524792]

>>6524682
>PS: this is /sci/ - "Science and MATH", not "CS

computer SCIENCE

you're trolling and hope you get banned again… just like last time.

[Anonymous No.6524798]

>>6524789
like here's a typical timeline for a paper
>1 day: oh, that's a neat problem. that problem seems important. that problem seems unsolved.
>1 month: read 20 papers on the problem. make careful study of what they've done, and what is left for future work
>1 week: realize a collective flaw, blind spot, or unresearched future work in all of the papers you have read. possibly you do this by making connections to another topic area in which you have expertise, which the paper authors may have not. interdisciplinary research is extremely common.
>2 months: implement your crazy idea for your own edification. realize it's harder to do than you initially imagined; solve a bunch of sub-problems on the way. maybe you do some novel math here w/r/t algorithms or whatever.
>1 month: scramble like a maniac to turn your shitty code into something that can produce reproducible and inter-comparable results. usually you have to meet some gold standard for your field, such as precision/recall on a specific data set (e.g. face recognition) in ML.
i have fun with pretty much every part but the last one, and the methodology of the last part is of course heavily debated and a subset of much serious interest. pure theory has failed to replace empiricism, however, which suggests something about the maturity of the theory

[Anonymous No.6524801]

>>6524798
*subject of much serious interest. the standards for publishable results are one of the most hot topics in CS as they have been in most sciences in perpetuity

[Anonymous No.6524813]

What an idiotic question, OP.


>That's right: According to BLS, 62% of the growth in jobs in science-related fields will be in CS. That's more than 750,000 new jobs, and a total of more than 1,350,000 job openings. How does that compare with the number of degrees likely to be granted by U.S. institutions? In 2009, those institutions granted something like 95,000 CS degrees, including about 32,000 associate's degrees and fewer than 1600 PhDs. Over 10 years, that adds up to more than the number of new jobs that BLS expects will be added, but significantly fewer than the number of job openings.

>It's when you compare these numbers to other fields that you realize how much better prospects are likely to be in CS. Consider that in the physical sciences, for example, BLS predicts that fewer than 36,000 new jobs will be created—a 3% increase. BLS predicts 121,900 openings in all fields of physical science, at all degree levels, before 2020. According to the National Science Foundation, about 27,000 people were granted degrees in the physical sciences in 2009, at all degree levels, including associate's degrees. Multiply that by 10 and compare. You'll quickly see that supply is predicted to surpass demand.

>Finding: All indicators—all historical data, and all projections—argue that CS is the dominant factor in America's science and technology employment, and that the gap between the demand for CS talent and the supply of that talent is and will remain large.

>While there will be inevitable variations in demand for every field, the long-term prospects for employment in CS occupations in the US are exceedingly strong. All other S&T fields pale by comparison.

http://sciencecareers.sciencemag.org/career_magazine/previous_issues/articles/2013_03_25/caredit.a1300053

[Anonymous No.6524833]

>>6524813
>significantly fewer than the number of job openings
yeah i was avoiding the topic of jobs and industry since that sort of thing is sometimes derided as pleb or temporal but CS has been a seller's job market for as long as i've been here. /sci/ will talk as if software engineering is the only vocation and as if a csbs in no way prepares you to do SE and
>implying you're screwing yourself with a csbs
but hire-ability has *never* been a concern for my undergrads or my co-researchers. most people i know received multiple job offers when their thesis defense date was posted. this is anecdotal + probably not true of all instutitions so take it with salt but then again there are statistics up above

[Anonymous No.6524834]

>>6524792
>It's has science in it's name
>must be technical and mathematical

[Anonymous No.6524839]

>>6524756
>implying all code must come from CS majors

Yes, who needs deep knowledge of engineering when you know thousands of emac keybinds. That's the most important part of making system.

[Anonymous No.6524840]

>>6524600
You decide what's worth your time, you fucking faggot.

[Anonymous No.6525458]

To what extent does CS involve math and to what extent does it involve concrete principles?

[Anonymous No.6525630]

>>6525458
I truly don't know how much is actually involved in the career field but at my uni, by the time you finish, you minor in math/physics

[Anonymous No.6525635]

>>6525458
I've been told CS is pretty much a math major.

[Anonymous No.6525641]

>>6525635
I can feel the heads of the math purists boiling

[Anonymous No.6526140]

>>6525458
>>6525635
this is totally wrong and i discussed it a lot above. it's computer SCIENCE. like most sciences there is a theory branch with its own particular brand of mathematical modeling, but most computer science research papers are empirical studies of computing phenomenon

a good example is
1. here is a wacky new idea we had for machine learning
2. intuitively and from preliminary experiments, the principles we're trying to exploit are...
3. EXPERIMENT: how does this method compare to the best existing methods on established problems, e.g. facial recognition?
4. discussion of the results of the experiment

this kind of paper is standard in all natural sciences and computer science is no different

[Anonymous No.6526143]

>>6525641
why

[Anonymous No.6526144]

>>6526140

[Anonymous No.6526145]

>>6526140
now to be clear you won't be writing papers like this in undergrad. you'll be learning the terms of the science and its research history, so that you'll be equipped to understand research papers later should you choose to go into research. alternatively (and obviously this is what most people do), you can opt for something more vocational and apply what you've learned in industry, but if so it's going to be applied science and it's less likely you'll be doing novel work. you'll be applying well-establlshed patterns, hence software engineering. this is just like any other science

misunderstanding about the role of *science* in computer science and what computer scientists actually do is at the heart of /sci/'s whole goofy shtick where "real CS is math" and "undergrad CS is a waste unless you learn SE."

[Anonymous No.6526147]

>>6526144
lol
fwiw i kind of think SICP is totally wrongheaded

[Anonymous No.6526149]

>>6526147
like if you wanted to have a giant epistemological argument about the relationship between math and computing that's p. fucking juicy but i'm not sure that 4chan is equipped for it

[Anonymous No.6526439]

>>6524648
ITT arrogant narcissists and autists talk in a condescending tone about something they know nothing about to veil their feelings of inferiority, self-loathing and incipient homosexuality

[Anonymous No.6526659]

>>6526439
>incipient homosexuality
hey im gay and im the only person not shitposting in this thread so

[Anonymous No.6526689]

>>6526659
>not shitposting