4 ms·

For the health of society, all recommender algorithms have to be daylighted. For a start. Yes, the gamification will be terrible. So what? Is maximal advertisi

For the health of society, all recommender algorithms have to be daylighted. For a start. Yes, the gamification will be terrible. So what? Is maximal advertising revenue for monopolies societally important? Every other efficient open market is heavily regulated. Accountability, transparency, fair play, information symmetry, prohibit self-dealing & conflicts of interest, tort, etc. Right now we can't even imagine applying casino level scrutiny to social media. What's so special about social media that it doesn't warrant some hygiene, some guard rails?

Grimm1 6y ago
I don't disagree with you but not all reccomenders are causing the political issues we see now. As an example, what Netflix movies are being recommended to you is, in my opinion, immaterial. I think there needs to be nuanced law here maybe just targeting and regulating recommendations on what we would consider news or communications platforms.
- specialist 6y ago
  IDK. I feel pretty siloed on Netflix today. And I wander how 3rd party content producers feel about Netflix's recommenders.
  - Grimm1 6y ago
    I think that's a product issue though and not something that requires regulation. I feel the exact opposite, I barely look at them and get my recs from friends or other sites I frequent.
esrauch 6y ago
> Yes, the gamification will be terrible. So what? Is maximal advertising revenue for monopolies societally important? I don't really see how gamification being terrible will only cause ad revenue to drop. Surely it will also make the results worse and less relevant to you? As in, if SEO spam managed to take over the top results of major queries, it doesn't just drop ad revenue but actually makes Search suck, right?
- specialist 6y ago
  How could future results be any worse?
  - wrkronmiller 6y ago
    In the worst case: search becomes completely unusable because the top pages of every major set of search terms is taken up by nonsense blog-spam. That is absolutely not currently the case.
  - esrauch 6y ago
    I actually don't see how "gamification" (adversarial websites exploiting loopholes in ranking) can even result in making ad revenue lower excepting the scenario where it makes Search itself less useful and therefore less used overall?
    - specialist 6y ago
      IDK. My prediction: Ad revenue peaked. COVID-19 is accelerating the decline. Google and others switched strategies to engagement and "rundles".
crabmusket 6y ago
It's interesting that "security through obscurity" is rightly shunned, but algorithms like search are kept closely guarded. It speaks to just how different a problem search is- it's not something you can achieve mathematically, it's an imperfect attempt to provide a subjectively good experience.
burtonator 6y ago
You can't just open source the algorithms - you also have to release the data and the platform on which the algorithm was executed. This is becoming a major problem because a lot of research papers publish their results and their methodology but not the data so it's not reproducible. A LOT of the data Google would have would be semi-private. It's generally OK if Google has it because (arguably) the user shared the data with Google but they might not want other people to have access. I'm not saying it's right - just that it's a problem
- specialist 6y ago
  Rhetorical: Why does Google have PII? Advertising, engagement. I'm ok with banning, sharply curtailing targeted advertising. If personalized search results get impaired, darn. I'd much prefer to curate my own filter bubble any way. Newsfeeds and other algorithmic outrage maximizers should be banned outright.
screye 6y ago
2 reasons. 1. the quality of the recommendations is the real competitive advantage. It's like saying that all the sports scouting agencies need to fully reveal their algorithm to finding top talent to ensure fairness, when that exact algorithm is the thing that makes them a top scouting agency. 2. The company itself doesn't really know how the recommender works.(source: I built explainable ML models for a living). The inherent black box nature of any non-linear function approximator makes it near impossible to enable Explain-ability and gauge fairness for any system. In fact, I'd go as far to say that a sufficiently complex system of any kind quickly goes beyond human understanding. There are some decent proxies for it (SHAP, attention maps, gradient maps), but many recent papers raise questions on whether they really capture what we think they capture. I think regulation of some form is certainly a good idea for tech ethics. However, it does appear that laws are being drafted by politicians, lawyers and social scientists. There is a glaring lack of qualified ML/CS researchers and practitioners on the implications of vaguely worded policy to their current processes and pipelines. I was at ACM FAccT (ACM's fairness,accountability and transparency conference) this year and too many of the attendees seemed to be more concerned with the 'what needs to be achieved' rather than the 'how we can go about achieving it'. Terms such as fairness and accountability do not have widely accepted or principled meanings, which makes it near impossible to make a structured discussion around how anyone would go about implementing these in practice. In conclusion, what you state is necessary, but we are very far from coming up with actually useful laws to implement it. Moving too quickly risks haphazard rules that do more damage than help.
- specialist 6y ago
  It's great that you're conscientious, seriously working on ML, big data, adjacent policy issues. Thank you. In the spirit of "yes, and...": Can't manage what we can't (or won't measure. So policy wise, we should not delegate judgement, and default to hippocratic until we know more. ML & big data is following the fad cycle, just like optimization, expert systems, etc. I'm 95% certain that >95% ML and big data is hooey, most value has been realized, the rest is imagined. But meanwhile, these foolsgold pathogens are wreaking the economy. Or worse.
  - 6y ago

Grimm1 6y ago

I don't disagree with you but not all reccomenders are causing the political issues we see now. As an example, what Netflix movies are being recommended to you is, in my opinion, immaterial. I think there needs to be nuanced law here maybe just targeting and regulating recommendations on what we would consider news or communications platforms.

specialist 6y ago
IDK. I feel pretty siloed on Netflix today. And I wander how 3rd party content producers feel about Netflix's recommenders.
- Grimm1 6y ago
  I think that's a product issue though and not something that requires regulation. I feel the exact opposite, I barely look at them and get my recs from friends or other sites I frequent.

esrauch 6y ago

> Yes, the gamification will be terrible. So what? Is maximal advertising revenue for monopolies societally important? I don't really see how gamification being terrible will only cause ad revenue to drop. Surely it will also make the results worse and less relevant to you? As in, if SEO spam managed to take over the top results of major queries, it doesn't just drop ad revenue but actually makes Search suck, right?

specialist 6y ago
How could future results be any worse?
- wrkronmiller 6y ago
  In the worst case: search becomes completely unusable because the top pages of every major set of search terms is taken up by nonsense blog-spam. That is absolutely not currently the case.
- esrauch 6y ago
  I actually don't see how "gamification" (adversarial websites exploiting loopholes in ranking) can even result in making ad revenue lower excepting the scenario where it makes Search itself less useful and therefore less used overall?
  - specialist 6y ago
    IDK. My prediction: Ad revenue peaked. COVID-19 is accelerating the decline. Google and others switched strategies to engagement and "rundles".

crabmusket 6y ago

It's interesting that "security through obscurity" is rightly shunned, but algorithms like search are kept closely guarded. It speaks to just how different a problem search is- it's not something you can achieve mathematically, it's an imperfect attempt to provide a subjectively good experience.

burtonator 6y ago

You can't just open source the algorithms - you also have to release the data and the platform on which the algorithm was executed. This is becoming a major problem because a lot of research papers publish their results and their methodology but not the data so it's not reproducible. A LOT of the data Google would have would be semi-private. It's generally OK if Google has it because (arguably) the user shared the data with Google but they might not want other people to have access. I'm not saying it's right - just that it's a problem

specialist 6y ago
Rhetorical: Why does Google have PII? Advertising, engagement. I'm ok with banning, sharply curtailing targeted advertising. If personalized search results get impaired, darn. I'd much prefer to curate my own filter bubble any way. Newsfeeds and other algorithmic outrage maximizers should be banned outright.

screye 6y ago

2 reasons. 1. the quality of the recommendations is the real competitive advantage. It's like saying that all the sports scouting agencies need to fully reveal their algorithm to finding top talent to ensure fairness, when that exact algorithm is the thing that makes them a top scouting agency. 2. The company itself doesn't really know how the recommender works.(source: I built explainable ML models for a living). The inherent black box nature of any non-linear function approximator makes it near impossible to enable Explain-ability and gauge fairness for any system. In fact, I'd go as far to say that a sufficiently complex system of any kind quickly goes beyond human understanding. There are some decent proxies for it (SHAP, attention maps, gradient maps), but many recent papers raise questions on whether they really capture what we think they capture. I think regulation of some form is certainly a good idea for tech ethics. However, it does appear that laws are being drafted by politicians, lawyers and social scientists. There is a glaring lack of qualified ML/CS researchers and practitioners on the implications of vaguely worded policy to their current processes and pipelines. I was at ACM FAccT (ACM's fairness,accountability and transparency conference) this year and too many of the attendees seemed to be more concerned with the 'what needs to be achieved' rather than the 'how we can go about achieving it'. Terms such as fairness and accountability do not have widely accepted or principled meanings, which makes it near impossible to make a structured discussion around how anyone would go about implementing these in practice. In conclusion, what you state is necessary, but we are very far from coming up with actually useful laws to implement it. Moving too quickly risks haphazard rules that do more damage than help.

specialist 6y ago
It's great that you're conscientious, seriously working on ML, big data, adjacent policy issues. Thank you. In the spirit of "yes, and...": Can't manage what we can't (or won't measure. So policy wise, we should not delegate judgement, and default to hippocratic until we know more. ML & big data is following the fad cycle, just like optimization, expert systems, etc. I'm 95% certain that >95% ML and big data is hooey, most value has been realized, the rest is imagined. But meanwhile, these foolsgold pathogens are wreaking the economy. Or worse.
- 6y ago