4 ms·

> What I'm saying is that this is a heuristic, not an algorithm, and that you should blame Facebook, not "the algorithm" Anyone who ever had any direct contact

> What I'm saying is that this is a heuristic, not an algorithm, and that you should blame Facebook, not "the algorithm" Anyone who ever had any direct contact with implementing heuristics would call bullshit on this take. Some heuristics don't even return the same result twice. Now the guy that presses the button is supposed to be held responsible for what comes out of that crapshoot?

rndgermandude 4y ago
Yes, that guy pressed the button even though he should have known there are consequences to that. If you implement heuristics, it's your damn job to test your implementation and monitor it for adverse effects. In the context of this article, somebody made the decision to specifically change the thing (no matter if you want to call it a heuristic or an algorithm or whatever) to bury a certain story. This is not even an unintended outcome, it's an entirely intended outcome.
- arinlen 4y ago
  > Yes, that guy pressed the button even though he should have known there are consequences to that. You are only complaining that an heuristic outputted a result you don't like. That's not a bug nor a problem. It's just that you feel you're entitled to an outcome that suits your fancy, regardless of the process involved. Heuristics don't even ensure they'll output the same result twice, either because their input changes with time or because they already integrate a stochastic component. Demanding that an operator polices the output of said heuristics especifically to suit your personal taste makes no sense at all, let alone demanding people be "held responsible" for disappointing you.
mgraczyk 4y ago
I implemented heuristics like this one at Facebook. It's not bullshit, I even know how this particular one was implemented. In this case there is some non-determinism but the false positive rate is very low and the false negative rate also very low.

rndgermandude 4y ago

Yes, that guy pressed the button even though he should have known there are consequences to that. If you implement heuristics, it's your damn job to test your implementation and monitor it for adverse effects. In the context of this article, somebody made the decision to specifically change the thing (no matter if you want to call it a heuristic or an algorithm or whatever) to bury a certain story. This is not even an unintended outcome, it's an entirely intended outcome.

arinlen 4y ago
> Yes, that guy pressed the button even though he should have known there are consequences to that. You are only complaining that an heuristic outputted a result you don't like. That's not a bug nor a problem. It's just that you feel you're entitled to an outcome that suits your fancy, regardless of the process involved. Heuristics don't even ensure they'll output the same result twice, either because their input changes with time or because they already integrate a stochastic component. Demanding that an operator polices the output of said heuristics especifically to suit your personal taste makes no sense at all, let alone demanding people be "held responsible" for disappointing you.

mgraczyk 4y ago

I implemented heuristics like this one at Facebook. It's not bullshit, I even know how this particular one was implemented. In this case there is some non-determinism but the false positive rate is very low and the false negative rate also very low.