1/ Hot take:
I'm at the "stick to convention and swallow your pride to get it #phdone" stage of my #phdlife. But, I'm going to vent on here and say that I don't think ODD, ODD+D, and things like beautifully-rendered UML are useful.
And, often, they are harmful.
-
-
2/ There are many bad dev practices in computational social science. When I started I rarely saw shared model code, let alone unit-tests! It's gotten better. But, it's still bad. I think part of the problem is a focus on latex-friendly artifacts to satisfy institutional concerns.
1 reply 1 retweet 3 likesShow this thread -
3/ The result is papers / books that have elegantly written and illustrated VV&T sections... ...on incorrect models. Since I started, I've decompiled two Java political models published in monographs because the code wasn't shared. The models were wrong -- grossly so.
1 reply 1 retweet 1 likeShow this thread -
4/ Code communicates a model. Well-written code clearly communicates a model. You can't understand a model without reading the code. Everything else communicates intention.
1 reply 1 retweet 14 likesShow this thread -
5/ Intention is a good thing to convey. But more often than not, these things just facilitate quick skimming -- another institutional demand. The result is a perception of understanding without understanding over an ever-growing set of invalid models that have pretty façades.
1 reply 0 retweets 3 likesShow this thread
6/ Again, it can be a useful adjuvant. But when you delude yourself into thinking diagrams are capable of fully-communicating a model -- and invest your time in them accordingly -- your model usually that suffers. (And, your docs and code probably end up temporally mismatched.)
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.