3 ms·

I think you miss the fact that to figure out what scientific code does, you also have to know the science. Yeah, you can look for dumb bugs like the one they sh

by lutorm 17y ago

I think you miss the fact that to figure out what scientific code does, you also have to know the science. Yeah, you can look for dumb bugs like the one they show in that clip, but you have no way of knowing whether calculations are correct without knowing what's going on. I found it disingenuous how they flashed a line of code that says "fudge factor" as if that makes the code incorrect. Unless you know what that fudge factor is for, you can't make that judgement.

andrewcooke 17y ago
especially when it's numerical and certain approximations make scientific sense while others don't.
riffer 17y ago
Climatologists should articulate what adjustments are necessary and why. Popularizers can then make it accessible to laypersons if need be - there is certainly enough interest. The whole point is that up until now we've been trusting the details to climatologists, and that trust has been broken. Layer on top the stakes, and also consider that this issue really lies at the intersection of many fields: climatology, computer science, economic, politics; opening this up is the only sensible next step.
- lutorm 17y ago
  Oh, I totally agree that it all should be public. I'm just saying that the OP's take that "let's get some professional programmers in there and clean it up" won't work unless they work in close collaboration with the scientists themselves.
mixmax 17y ago
While I agree with your comment I think it should be noted that it is a lot easier to reproduce results than to make original research. It would probably not be viable to come up with new climate models, etc. but it might be viable to verify and eventually correct the current models and algorithms. After all the science has been done, and is known, so checking assumptions behind the code should be doable.

3 ms·

I think you miss the fact that to figure out what scientific code does, you also have to know the science. Yeah, you can look for dumb bugs like the one they sh

by lutorm 17y ago

andrewcooke 17y ago
especially when it's numerical and certain approximations make scientific sense while others don't.
riffer 17y ago
Climatologists should articulate what adjustments are necessary and why. Popularizers can then make it accessible to laypersons if need be - there is certainly enough interest. The whole point is that up until now we've been trusting the details to climatologists, and that trust has been broken. Layer on top the stakes, and also consider that this issue really lies at the intersection of many fields: climatology, computer science, economic, politics; opening this up is the only sensible next step.
- lutorm 17y ago
  Oh, I totally agree that it all should be public. I'm just saying that the OP's take that "let's get some professional programmers in there and clean it up" won't work unless they work in close collaboration with the scientists themselves.
mixmax 17y ago
While I agree with your comment I think it should be noted that it is a lot easier to reproduce results than to make original research. It would probably not be viable to come up with new climate models, etc. but it might be viable to verify and eventually correct the current models and algorithms. After all the science has been done, and is known, so checking assumptions behind the code should be doable.