3 ms·

> Getting denied a bank account/job/housing rental/... for no clear reason and with no way to appeal is becoming increasingly common, and seems to have all the

> Getting denied a bank account/job/housing rental/... for no clear reason and with no way to appeal is becoming increasingly common, and seems to have all the same negatives as the social credit program - if anything worse, since at least you can see your "social credit" score and have some idea of how to improve it. You seem to be confused about the salient differences between the social credit system and Western credit scores. Focusing on the fact that you can be "denied...for no clear reason or way to appeal" to draw an equivalence entirely misses the point. What's novel and scary about the social credit system is that political actions like criticizing the government or associating with activists could hurt your score. It's not scoped to financial activity like western systems. If I have a low FICO score, I'm 100% certain that it's not because I didn't vote for Trump.

lmm 8y ago
> What's novel and scary about the social credit system is that political actions like criticizing the government or associating with activists could hurt your score. It's not scoped to financial activity like western systems. If I have a low FICO score, I'm 100% certain that it's not because I didn't vote for Trump. What gives you that certainty? I worked on this stuff in the finance industry a few years back. Even at that stage we were taking any available data and throwing it into opaque machine learning models, and using those to make lending decisions. There was no nefarious intent, but it's not like we had (or cared to have) any way to exclude political affiliation effects either.
- monocasa 8y ago
  If I had an arrest at a protest that was expunged officially but still recorded by third parties, would that probably affect lending decisions?
  - lmm 8y ago
    Would it explicitly be taken into account? No. Would being the kind of person who'd get arrested at a protest be something our models could infer? Yes, though not at a high level of confidence. Would the specific fact of having been arrested have an effect? Probably not, unless they've got much better in the last few years. Would being the protesting sort affect lending decisions? If there was a correlation with non-repayment, yes, if not, no. We genuinely didn't care one way or another about that kind of thing, all we cared about was that the black box model could accurately predict how likely you were to default.

lmm 8y ago

> What's novel and scary about the social credit system is that political actions like criticizing the government or associating with activists could hurt your score. It's not scoped to financial activity like western systems. If I have a low FICO score, I'm 100% certain that it's not because I didn't vote for Trump. What gives you that certainty? I worked on this stuff in the finance industry a few years back. Even at that stage we were taking any available data and throwing it into opaque machine learning models, and using those to make lending decisions. There was no nefarious intent, but it's not like we had (or cared to have) any way to exclude political affiliation effects either.

monocasa 8y ago
If I had an arrest at a protest that was expunged officially but still recorded by third parties, would that probably affect lending decisions?
- lmm 8y ago
  Would it explicitly be taken into account? No. Would being the kind of person who'd get arrested at a protest be something our models could infer? Yes, though not at a high level of confidence. Would the specific fact of having been arrested have an effect? Probably not, unless they've got much better in the last few years. Would being the protesting sort affect lending decisions? If there was a correlation with non-repayment, yes, if not, no. We genuinely didn't care one way or another about that kind of thing, all we cared about was that the black box model could accurately predict how likely you were to default.