6 ms·

Honestly I would love to see it be the norm, maybe even law, that a recommendation is accompanied by a reason for the recommendation. Is this because I watched

Honestly I would love to see it be the norm, maybe even law, that a recommendation is accompanied by a reason for the recommendation. Is this because I watched a similar video? Or is it because you snooped on my microphone when I was at the local cafe?

thegeomaster 4y ago
You just put in a pile of all the data you know about the user and session into a pile of linear algebra (some ML-based recommendation system or whatever), and recommendations come out. In other words, I doubt they even know the reason for any particular one.
FredPret 4y ago
This might not be possible - there are many inputs into the model; it would be impossible to trace which one tipped the scale. Better to make listening on the mic illegal, punishable by prison time for the execs.
- NateEag 4y ago
  > This might not be possible - there are many inputs into the model; it would be impossible to trace which one tipped the scale. I am not a fan of problem-solving via legislation, but I wouldn't lose a lot of sleep if Big Tech was no longer allowed to burn billions of CPU cycles on making opaque, inexplicable guesses as to what ads are most likely to make them money if shoved in front of my eyeballs.
  - FredPret 4y ago
    I get the annoyance, but also, targeted ads means both a more efficient economy and information environment. Mass ads means more money goes into ads and ads are irrelevant to many people. Only big co’s can afford mass ads. Targeted ads means ad money is vastly more efficient, and your information environment is less cluttered. One-man shops can afford targeted ads. I like seeing relevant ads. I like supporting small businesses targeting me. Privacy isn’t a concern either because the three-letter agencies can find out whatever they want about me if they want anyway. FWIW, I’m not in advertising.
- Tao3300 4y ago
  All the more reason for transparency if no one can explain how it works. Ideally "my model is too opaque" isn't an excuse, but an admission of negligence.
- than3 4y ago
  They'll simply say, its not 'listening' if an algorithm does it.
  - FredPret 4y ago
    I love machine learning, data, and even targeted ads. But no marketing algorithm should have access to mic and camera data, period.
    - than3 4y ago
      Fun tangent, I recently went to Rally's (a really hard to find burger joint around my parts). I rarely go so its always a treat with their shake/glorious fries. Found they've completely replaced the order taker with a robot, with order signage saying "help train our robot". No opt out, and you can't get to an employee unless it recognizes the word employee. Recording in progress. The ADA non-compliant parts aside for people with voice related disabilities, its BS what companies are forcing people to do just to buy a burger. I'm not going back after that.
loa_in_ 4y ago
In systems complicated enough there could be no clear answer
- quasimodem 4y ago
  Then should we really be building systems that complicated?
  - euos 4y ago
    Because it is harmless. The worst that can happen is a bad recommendation.
    - Tao3300 4y ago
      Or the hijacking of attention by unwittingly flooding the streams with "engaging" content that is politically divisive and spreads misinformation. Maybe you left that out because it already happened.
  - tzs 4y ago
    Even fairly simple systems might not have a clear way to answer. Consider a movie recommendation system. One simple way you could build such a system is to come up with a list of things about movies that might affect whether or not someone would like them, where for each thing on the list we can assign each moved a number from -1 to 1 that says how much of that thing the movie has. Call this list the movie's vector. Some examples of things we might pick are how much comedy is in the movie, how much romance is in the movie, presence of A-list stars, how musical it is, and thing like that. We might also have items for specific stars or directors. Then we could go through our movie collection and have someone figure out each movies scores for all those things in our list. Then we could figure out for our users a list that lists for each of those things how important it is to that user, from -1 (I hate movies that have this!) to 1 (I love movies that have this!). Let's call this the user preferences vector. If we have a list for a given user of movies they have already watched and how they rated them on say a 0 to 5 scale then it is some straightforward math to figure out the user preferences vector for that user that does the best job classifying the movies they have already seen in a way that agrees well with that user's ratings. That user preferences vector can than be used to recommend new movies and should work pretty well if (1) we picked a good list of things to score movies on, and (2) when we manually assigned the scores we got it right. To predict how well a user would like a given move we just take their user preference vector and compute the dot product of it with the movie's vector. The more positive that result the more we think the user would like the movie. With this system, it would be easy to tell someone why you recommended a movie. We could look at their preference vector and compare it to the movie and tell them things they really like that the movie has and things they really hate that the movie does not have. But the system described above has a drawback. It is hard to figure out what factors to include in the movie classification. Should comedy for example be one item, or should it be broken down into several such as physical comedy, insult comedy, bodily function comedy, and so on? Also, if you have a large collection of movies it is a lot of work to go through them all and score them on each factor. And if you later find out you need to add or remove factors you have to do it again. It turns out that there is a way to sidestep both the "what should my factors include?" and "how do we get the factors scored?" problems. What you do is just decide on how many factors you will have. So let's say we decide we are going to have 50 factors. We don't have to decide what they mean. We'll just call then F1, F2, F3, ... for now. Initially we just assign each factor a random value from -1 to 1. We also do the same thing for the initial user preference vectors. Just assign each factor in the preference vector a random value. Then we can do a loop, consisting of these two steps: 1. Using the known 0 to 5 star ratings from users of films they have seen, adjust their preference vectors so that ordering movies by the dot products of the movie vectors with the preferences vector matches the ordering by the user's star ratings. 2. Same thing, except instead of adjusting the preference vector to better work with all the movies a user has seen, adjust the movie vector of each movie to better work with the preference vectors or all the users have have rated that movie. Keep looping until things aren't changing much. You then end up with a set of movie vectors for you movie catalog and preference vectors for your users that do a good job of ordering movies that user has seen that matches well with how the user rated those movies, and that likely does a good job predicting how well they will like new movies. This was with an arbitrary decision to have 50 factors in our movie vectors. Since the process described above can be entirely automated and is pretty quick, we can experiment with difference vector sizes. We end up with a recommendation system that is very likely much better than the one we would get if we picked the items that went into the movie vector. It is still a simple system, just like it was when we came up with the components ourselves and assigned them by watching the movies ourselves. But notice that we now have no idea what the heck the factors in the movie vector mean. We can no longer tell someone we recommended a movie because they like comedy and romance and hate kids movies and this movie fits that well. We could tell them they liked past movies that have a high score in components 12, 15, 23, and 102 and a low score in components 19, 77, 83, and 107, but they are probably not going to find that to be a useful answer.
notimetorelax 4y ago
Story time: I have disabled all types of tracking in Twitter, same for Amazon. Now, I bought a few e-books for my wife. Those are very girly books that definitely don’t fit my profile. Twitter is now showing me kindle ads for those books. Twitter has a menu offering to explain why I see it. The explanation says: Amazon wanted to advertise this to people in your region. I say they lie… We need Blackbox monitoring against those behaviors with some legal teeth.

thegeomaster 4y ago

You just put in a pile of all the data you know about the user and session into a pile of linear algebra (some ML-based recommendation system or whatever), and recommendations come out. In other words, I doubt they even know the reason for any particular one.

FredPret 4y ago

This might not be possible - there are many inputs into the model; it would be impossible to trace which one tipped the scale. Better to make listening on the mic illegal, punishable by prison time for the execs.

NateEag 4y ago
> This might not be possible - there are many inputs into the model; it would be impossible to trace which one tipped the scale. I am not a fan of problem-solving via legislation, but I wouldn't lose a lot of sleep if Big Tech was no longer allowed to burn billions of CPU cycles on making opaque, inexplicable guesses as to what ads are most likely to make them money if shoved in front of my eyeballs.
- FredPret 4y ago
  I get the annoyance, but also, targeted ads means both a more efficient economy and information environment. Mass ads means more money goes into ads and ads are irrelevant to many people. Only big co’s can afford mass ads. Targeted ads means ad money is vastly more efficient, and your information environment is less cluttered. One-man shops can afford targeted ads. I like seeing relevant ads. I like supporting small businesses targeting me. Privacy isn’t a concern either because the three-letter agencies can find out whatever they want about me if they want anyway. FWIW, I’m not in advertising.
Tao3300 4y ago
All the more reason for transparency if no one can explain how it works. Ideally "my model is too opaque" isn't an excuse, but an admission of negligence.
than3 4y ago
They'll simply say, its not 'listening' if an algorithm does it.
- FredPret 4y ago
  I love machine learning, data, and even targeted ads. But no marketing algorithm should have access to mic and camera data, period.
  - than3 4y ago
    Fun tangent, I recently went to Rally's (a really hard to find burger joint around my parts). I rarely go so its always a treat with their shake/glorious fries. Found they've completely replaced the order taker with a robot, with order signage saying "help train our robot". No opt out, and you can't get to an employee unless it recognizes the word employee. Recording in progress. The ADA non-compliant parts aside for people with voice related disabilities, its BS what companies are forcing people to do just to buy a burger. I'm not going back after that.

loa_in_ 4y ago

In systems complicated enough there could be no clear answer

quasimodem 4y ago
Then should we really be building systems that complicated?
- euos 4y ago
  Because it is harmless. The worst that can happen is a bad recommendation.
  - Tao3300 4y ago
    Or the hijacking of attention by unwittingly flooding the streams with "engaging" content that is politically divisive and spreads misinformation. Maybe you left that out because it already happened.
- tzs 4y ago
  Even fairly simple systems might not have a clear way to answer. Consider a movie recommendation system. One simple way you could build such a system is to come up with a list of things about movies that might affect whether or not someone would like them, where for each thing on the list we can assign each moved a number from -1 to 1 that says how much of that thing the movie has. Call this list the movie's vector. Some examples of things we might pick are how much comedy is in the movie, how much romance is in the movie, presence of A-list stars, how musical it is, and thing like that. We might also have items for specific stars or directors. Then we could go through our movie collection and have someone figure out each movies scores for all those things in our list. Then we could figure out for our users a list that lists for each of those things how important it is to that user, from -1 (I hate movies that have this!) to 1 (I love movies that have this!). Let's call this the user preferences vector. If we have a list for a given user of movies they have already watched and how they rated them on say a 0 to 5 scale then it is some straightforward math to figure out the user preferences vector for that user that does the best job classifying the movies they have already seen in a way that agrees well with that user's ratings. That user preferences vector can than be used to recommend new movies and should work pretty well if (1) we picked a good list of things to score movies on, and (2) when we manually assigned the scores we got it right. To predict how well a user would like a given move we just take their user preference vector and compute the dot product of it with the movie's vector. The more positive that result the more we think the user would like the movie. With this system, it would be easy to tell someone why you recommended a movie. We could look at their preference vector and compare it to the movie and tell them things they really like that the movie has and things they really hate that the movie does not have. But the system described above has a drawback. It is hard to figure out what factors to include in the movie classification. Should comedy for example be one item, or should it be broken down into several such as physical comedy, insult comedy, bodily function comedy, and so on? Also, if you have a large collection of movies it is a lot of work to go through them all and score them on each factor. And if you later find out you need to add or remove factors you have to do it again. It turns out that there is a way to sidestep both the "what should my factors include?" and "how do we get the factors scored?" problems. What you do is just decide on how many factors you will have. So let's say we decide we are going to have 50 factors. We don't have to decide what they mean. We'll just call then F1, F2, F3, ... for now. Initially we just assign each factor a random value from -1 to 1. We also do the same thing for the initial user preference vectors. Just assign each factor in the preference vector a random value. Then we can do a loop, consisting of these two steps: 1. Using the known 0 to 5 star ratings from users of films they have seen, adjust their preference vectors so that ordering movies by the dot products of the movie vectors with the preferences vector matches the ordering by the user's star ratings. 2. Same thing, except instead of adjusting the preference vector to better work with all the movies a user has seen, adjust the movie vector of each movie to better work with the preference vectors or all the users have have rated that movie. Keep looping until things aren't changing much. You then end up with a set of movie vectors for you movie catalog and preference vectors for your users that do a good job of ordering movies that user has seen that matches well with how the user rated those movies, and that likely does a good job predicting how well they will like new movies. This was with an arbitrary decision to have 50 factors in our movie vectors. Since the process described above can be entirely automated and is pretty quick, we can experiment with difference vector sizes. We end up with a recommendation system that is very likely much better than the one we would get if we picked the items that went into the movie vector. It is still a simple system, just like it was when we came up with the components ourselves and assigned them by watching the movies ourselves. But notice that we now have no idea what the heck the factors in the movie vector mean. We can no longer tell someone we recommended a movie because they like comedy and romance and hate kids movies and this movie fits that well. We could tell them they liked past movies that have a high score in components 12, 15, 23, and 102 and a low score in components 19, 77, 83, and 107, but they are probably not going to find that to be a useful answer.

notimetorelax 4y ago

Story time: I have disabled all types of tracking in Twitter, same for Amazon. Now, I bought a few e-books for my wife. Those are very girly books that definitely don’t fit my profile. Twitter is now showing me kindle ads for those books. Twitter has a menu offering to explain why I see it. The explanation says: Amazon wanted to advertise this to people in your region. I say they lie… We need Blackbox monitoring against those behaviors with some legal teeth.