Anybody can code

Anybody can code but what just anybody codes can be crap.

There's a real problem in my place of work and probably in other places where you have scientists and engineers who haven't been formally trained in CS and haven't taken the time or effort to learn about proper techniques but who are writing a lot of code.  It's one thing to whip something up quickly to answer a question, but some of this code becomes pieces of larger things or gets delivered to a customer. And someone has to maintain it, but it's got memory leaks, isn't documented, it is unreliable, it's not secure, and no one knows what all licenses were on the random bits and pieces.* See a brief discussion of this sparked by a post by Deepak Singh.

I attribute some of this to carelessness, but a lot comes from other sciences disrespecting CS.**  I remember even as a physics undergrad the hubris - physicists could do anything that engineers or computer scientists could do, just give us a few minutes with an envelope or chalk board to derive the equations! Or the only reason to hire a computer scientist is so you can spend more time with the real science and leave the grunt work to them.  Even in these times of "computational thinking," it's amazing how this mentality hangs on.  Computer scientists going for tenure or promotion or a raise have to explain that their product is software OR that it's actually harder to get papers accepted in certain conferences than many journals in CS OR that they had to undo a bunch of stuff non-professionals did so didn't get a chance to write up their work in a memo, technical report, or the holy journal article.

CS did grow out of math, to an extent and programming (a part but not all of CS) is used by every flavor of scientist including social scientists and, more recently, a lot of humanities scholars. But there is a difference between professional software engineering or even software development and random people (like me) coding.

 

* FWIW we have addressed this at my place of work within our quality management program and the specific part of that program that deals with software.

** Of course most computer scientists have ZERO respect for information scientists. ZERO. But you get enough of that in my other posts.

(this was supposed to be a much longer post on professionalization, but, sigh, not very prolix. Well, comparatively speaking :)  )

Tags

More like this

The normally sensible Bart went slightly gooey over Citizen Science recently, although to be fair he linked to his earlier article wherein he was distinctly more dubious about the benefits, and indeed about the good intentions about some of the wannabe scientists. And in the comments he is more…
I'm on my annual summer hiatus for the month of July so I'll be only publishing my weekly Friday Fun posts as well as re-posting some of the interviews I did a few years ago on the old blog with people from the publishing, library and science worlds. Not that my posting of late has been…
I'm currently away on a family vacation, and as soon as vacation is over, I'm off on a business trip for a week. And along the way, I've got some deadlines for my book. So to fill in, I'm recycling some old posts. I decided that it's been entirely too long since there was any pathological…
My last posts on why I don't like the open source metaphor for science have generated a lot of good comments, here and in my email, twitter, and in person. They've forced me to think about what exactly it is about the meme that makes me so uncomfortable, and raised some good objections and points…

There's a real problem in my place of work and probably in other places where you have scientists and engineers who haven't been formally trained in CS and haven't taken the time or effort to learn about proper techniques but who are writing a lot of code. It's one thing to whip something up quickly to answer a question, but some of this code becomes pieces of larger things or gets delivered to a customer. And someone has to maintain it, but it's got memory leaks, isn't documented, it is unreliable, it's not secure, and no one knows what all licenses were on the random bits and pieces.*

Now for the sad reality: My 10 years as a professional software developer has taught me that people who have CS degrees, and cart around the title "software engineer" seldom bother to use the proper techniques they damn well know, and just like the scientists and engineers you're complaining about, produce buggy, undocumented, unreliable, insecure code, attached to pieces with who knows what licenses.

Sort of why I went for Information Science as opposed to a pure Computer Science degree.

I can code, no doubt about that. Last contract I had I think they were stunned that I could figure out the UPC check digit algorithm.

Simple really, sum up the digits you have and then round up to next tens unit. Divide that by 10 and you've got your check digit.

It has to be simple, hardware manufacturers aren't going to waste CPU cycles to do a complex hash or calculation.

I prefer to draw a distinction between Computer Science and Computer Engineering. CS pretty much *is* math. CE is programming, algorithms, design, ect.

CE folks need to know basic CS, but not much more typically. CS folks could care less about proper coding practices and project organization/design beyond the little needed to implement the odd proof of concept or tool.

That isn't to suggest the CE isn't very worthwhile, it is just quite different. Kind of like the difference between chemical engineering and chemistry.

BTW: I'm a bit odd... not CS or CE, but I dabble in both a bit. I'm quite interested in making the tools I've been developing actually usable since I think that is actually contribution to science. Sadly, the incentives are actually against me spending the time and effort involved in coding more carefully, much less actually putting together a decent package for distribution.

The saying while I was going to school was, "An engineer can do CS, but a CS can't do engineering."

I've seen loads of horrible programmers who have full-fledged CS degrees - the issue is competence, not training. At most institutions, CS is a "lighter" degree than something like engineering or physics. This attracts more of the people who couldn't cut a harder workload, and of course their lower level of performance continues throughout their life.

Obviously speaking in generalities - some programs have blistering CS routes, and there's quite a few CS people who could've easily handled the more math-intensive programs. But as a rule, CS < Eng/Phy.

By EngineerElf (not verified) on 15 Sep 2009 #permalink

'scientists and engineers who haven't been formally trained in CS and haven't taken the time or effort to learn about proper techniques but who are writing a lot of code.'

A quick comment on this - i'm a phd student, who needs to do a fair bit of coding - i've found it difficult to gather support to go on courses teaching basic good coding practice. My lack of experience limits the size of program i can write, and its efficiency.There are some course around, but they can take a bit of finding - i wonder if this is also a part of the problem?

I've known plenty of people who have a CS degree but still can't and/or don't code worth a damn. I've known plenty of people who were not CS and could code better than CS people.

Then again, computer science has little to nothing to do with writing good code.

"I can write programs that control air traffic, intercept ballistic missiles, reconcile bank accounts, control production lines."

"So can I, and so can any man, but do they work when you do write them?"

Fred Brooks, after Shakespeare, in "The Mythical Man-Month", quite possibly the best book on software engineering yet produced.

'To a man with a hammer, everything looks like a nail,' - Mark Twain.

The worst I've seen is a self-styled programming expert open a file, incorporate 200 lines of boilerplate C (to get his favourite include files, as well as asserting his identity and copyright), CVS-ing the thing and only then engaging the brain to understand it wasn't really that sort of problem.

Eric Raymond's Art of UNIX programming has plenty of wisdom on this - that programming should be appropriate - and that includes the fact that military-grade exception-handling is not usually required ...

Yes, I've seen this phenomenon, as well as its corollary that one's own code is excellent and everyone else's is junk. So I sometimes have to decipher code with unexplained magic numbers, mysterious loops, and the like.

I have the second edition of Numerical Recipes in C on my bookshelf. It's a useful reference for a number of things, one of which (a feature the authors presumably did not intend to include) is a tutorial in how not to write C code. The authors are disdainful of comments and white space, they make no real effort to have their variable names reflect their purpose, and they impose a 1 to N structure on a language that prefers to go from 0 to N-1, even when it makes no substantive difference to the algorithm. Which makes it difficult to figure out what they are doing when their algorithms get a little too clever.

Speaking of comments: In homegrown code, their primary purpose is not to make your code readable to others (although that is a secondary purpose). Their primary purpose is to remind you what your thought process was, so that you can debug or modify the code months or years later.

By Eric Lund (not verified) on 17 Sep 2009 #permalink

"I've known plenty of people who have a CS degree but still can't and/or don't code worth a damn. I've known plenty of people who were not CS and could code better than CS people."

I second that, Phil. I have come across a few CS types who have not got a clue about HCI or interface. They really should be kept away from doing HCI/design-related projects.

It's about imagination and drive, really, and how much you enjoy your job. I must say coding takes patience. It's the 'patience' bit that is tough, not the codes.