4 ms·

Welcome to the future. Everything is decided by AI based on data we don't really understand.

coding123 6y ago
If only there was a way to track the number of seconds they spent on pinterest until the back button was hit and... factor that in as a feature in ML. voila.
- wanderr 6y ago
  I don't think it's that simple. When a user goes back to image search results, is it because they are unsatisfied with the result they clicked on, or are they looking for even more, hoping to compare? And when a user closes out instead of doing back, is it because they got what they wanted or because the results were so garbage they gave up? Based on my own usage of image search results I don't think you would be able to tell when I was happy with the results or when I wasn't. But you can probably tell that my usage of image search went down when it started getting shitty due to this spam.
  - pfranz 6y ago
    What are your use-cases for image search? Mine are pretty straightforward; find a higher quality/uncropped version or to find the source (article, recipe, or photo set). Use case #1 I click through save and exit. Use case #2 I engage like I would for web (text) search. #1 might be confused for not finding the intended result, but could be discerned after looking a bit more. Also, if a lot of people add in "-site:pinterest.*" (or click on pinterest sites but come back, then find the intended result) that might be a hint it's not adding real value.
    - wanderr 6y ago
      I think my use cases are pretty varied, anything from looking for pictures of a place I'm considering going hiking, looking for different views of a product I'm considering purchasing, looking for other examples of an artists art, to trying to figure out where some image originated
ryanb 6y ago
This answer is super spot on, and this will be the story of our lives for a while.
- TeMPOraL 6y ago
  It could be better than what we have now, assuming somewhat smart AI. Right now, data gives plausible deniability, and is the universal scapegoat - you get to do whatever you want, backing this up with misapplied A/B tests and telemetry, with zero statistical rigor. And if someone complains about your decisions, you point to these tests and that telemetry, and cry "we're a data driven company, and the data proves we're right".
  - xg15 6y ago
    I mean, it would help to be honest about the metrics that the A/B tests are evaluated against. If you optimize purely for growth and engagement, then your test results will reflect that - even moreso if your tests are scientifically sound. I'm continually surprised how companies manage to sell "let's try to keep users glued to the screen as long as possible" as doing something for the general good.
  - adrianN 6y ago
    I have little hope that the people not understanding their data today and lacking any statistical rigor will be able to train an AI to do better. Figuring out whether an AI gives you garbage results seems to be a harder problem than figuring out whether your own results are garbage.
ehsankia 6y ago
No one said it had to do with AI. The point being made above is that while you and me may dislike Pinterest, the data may be showing that to most users, it's actually very valuable. If anything, ML would be able to detect my and your use cases and intelligently hide Pinterest for us while hiding it for others. So chances are they aren't using ML yet which is why it's just favoring the majority rather than us (the minority).

If only there was a way to track the number of seconds they spent on pinterest until the back button was hit and... factor that in as a feature in ML. voila.

wanderr 6y ago
I don't think it's that simple. When a user goes back to image search results, is it because they are unsatisfied with the result they clicked on, or are they looking for even more, hoping to compare? And when a user closes out instead of doing back, is it because they got what they wanted or because the results were so garbage they gave up? Based on my own usage of image search results I don't think you would be able to tell when I was happy with the results or when I wasn't. But you can probably tell that my usage of image search went down when it started getting shitty due to this spam.
- pfranz 6y ago
  What are your use-cases for image search? Mine are pretty straightforward; find a higher quality/uncropped version or to find the source (article, recipe, or photo set). Use case #1 I click through save and exit. Use case #2 I engage like I would for web (text) search. #1 might be confused for not finding the intended result, but could be discerned after looking a bit more. Also, if a lot of people add in "-site:pinterest.*" (or click on pinterest sites but come back, then find the intended result) that might be a hint it's not adding real value.
  - wanderr 6y ago
    I think my use cases are pretty varied, anything from looking for pictures of a place I'm considering going hiking, looking for different views of a product I'm considering purchasing, looking for other examples of an artists art, to trying to figure out where some image originated

ryanb 6y ago

This answer is super spot on, and this will be the story of our lives for a while.

TeMPOraL 6y ago
It could be better than what we have now, assuming somewhat smart AI. Right now, data gives plausible deniability, and is the universal scapegoat - you get to do whatever you want, backing this up with misapplied A/B tests and telemetry, with zero statistical rigor. And if someone complains about your decisions, you point to these tests and that telemetry, and cry "we're a data driven company, and the data proves we're right".
- xg15 6y ago
  I mean, it would help to be honest about the metrics that the A/B tests are evaluated against. If you optimize purely for growth and engagement, then your test results will reflect that - even moreso if your tests are scientifically sound. I'm continually surprised how companies manage to sell "let's try to keep users glued to the screen as long as possible" as doing something for the general good.
- adrianN 6y ago
  I have little hope that the people not understanding their data today and lacking any statistical rigor will be able to train an AI to do better. Figuring out whether an AI gives you garbage results seems to be a harder problem than figuring out whether your own results are garbage.

ehsankia 6y ago

No one said it had to do with AI. The point being made above is that while you and me may dislike Pinterest, the data may be showing that to most users, it's actually very valuable. If anything, ML would be able to detect my and your use cases and intelligently hide Pinterest for us while hiding it for others. So chances are they aren't using ML yet which is why it's just favoring the majority rather than us (the minority).