3 ms·

Not to mention that "research" can be further split into myriad activities that take up substantial amounts of time. Personally, I'd contend that writing grant

Not to mention that "research" can be further split into myriad activities that take up substantial amounts of time. Personally, I'd contend that writing grants and writing scientific code could also constitute full-time jobs (this might explain why a lot of scientific code is spaghetti; most researchers have to balance coding with publishing- and coding loses in this battle). Regardless of how many full-time jobs the modern researcher is expected to perform, the end result is the same: lower quality science furnished by 80 hour weeks and emotionally wrecked postdocs.

killjoywashere 10y ago
> writing grants this is the kind of crap that administrators (as a class in society) force on faculty and then use it as justification for "our role is necessary" and "see, they couldn't be that busy".
mathperson 10y ago
I think the reason scientific code is spaghetti has more to do with different incentives than time management. you don't need scalable production quality code. you need a minimal working example of something even its a kluge. correct and ugly and novel will win over optimized and pretty and commented and non-novel...
- neuromantik8086 10y ago
  I don't know whether or not the incentives or the time management contribute more towards the messiness, but I will nitpick one implicit assumption you're making here, which is that scientific code is "correct". There has been at least one well-publicized incident where scientific code is both messy and very much incorrect. I also think that, while optimization isn't necessarily something that researchers should focus on, pretty code / documentation / comments / systematic testing are elements that researchers really need to emphasize, because these practices ensure that the likelihood that their simulations / analysis pipelines aren't spewing out nonsense is substantially decreased. Furthermore, these practices would make code reviews a lot easier- it's much harder to make sense of your code when every other variable is named "i", "j" or "k" and this seems to be a common practice in some fields at least.
  - mathperson 9y ago
    You know that is a very good point actually. I have talked with some bioinformaticians who have trouble reproducing others work because the dataset was unavailable or the code was not provided. Personally, all the academic code I've used has been written by EE people or Applied Math people and had nice enough style/documentation/accessibility for me to have minimal issues.

killjoywashere 10y ago

> writing grants this is the kind of crap that administrators (as a class in society) force on faculty and then use it as justification for "our role is necessary" and "see, they couldn't be that busy".

mathperson 10y ago

I think the reason scientific code is spaghetti has more to do with different incentives than time management. you don't need scalable production quality code. you need a minimal working example of something even its a kluge. correct and ugly and novel will win over optimized and pretty and commented and non-novel...

neuromantik8086 10y ago
I don't know whether or not the incentives or the time management contribute more towards the messiness, but I will nitpick one implicit assumption you're making here, which is that scientific code is "correct". There has been at least one well-publicized incident where scientific code is both messy and very much incorrect. I also think that, while optimization isn't necessarily something that researchers should focus on, pretty code / documentation / comments / systematic testing are elements that researchers really need to emphasize, because these practices ensure that the likelihood that their simulations / analysis pipelines aren't spewing out nonsense is substantially decreased. Furthermore, these practices would make code reviews a lot easier- it's much harder to make sense of your code when every other variable is named "i", "j" or "k" and this seems to be a common practice in some fields at least.
- mathperson 9y ago
  You know that is a very good point actually. I have talked with some bioinformaticians who have trouble reproducing others work because the dataset was unavailable or the code was not provided. Personally, all the academic code I've used has been written by EE people or Applied Math people and had nice enough style/documentation/accessibility for me to have minimal issues.