5 ms·

They're called "Likes" because they represent what people like. You are setting a limit on the number of things people like? Your problem isn't with the conce

They're called "Likes" because they represent what people like. You are setting a limit on the number of things people like? Your problem isn't with the concept of likes, it's that you wish they solved a problem which they don't solve. You wish they were "recommendations" instead. But this is already a solved problem. When I like something enough that I want to recommend it to someone else, I use email. Or I call. Or I mention it in the hallway. Or I comment about it on Google+. This type of recommendation existed long before we were born. It doesn't really need any help from you, but if you want to add a "recommendation" feature, go for it. But this concept doesn't replace the completely different concept you might call "like-based recommendations". This is a new concept that wasn't possible before, and now you're talking about crippling it by setting artificial limits!

marchra 13y ago
Google+? Do you work in SEO?
Cushman 13y ago
The claim is that likes are a flawed currency to represent what people like because the cost of clicking a button on a web site trades for an infinitesimal amount of liking, so your representation winds being nearly useless to predict reality. Placing a limit on the "like" action to force people to choose is one potential way to increase the real value of the representation, hopefully extracting a stronger, more usable signal.
- yajoe 13y ago
  Right, but this solution tries to fit what is otherwise continuous into discrete so his calculations are easier. It is naive and unrealistic. I've never seen that succeed, and the 'approximations' people do with those kind of numbers are scary. Are all likes equal in weight? Does my affinity remain constant over time? Am I capable of liking a countable number of concepts? Do I naturally revisit all items I've liked in my past life and re-evaluate if they make my cut? The solution isn't in limiting likes, it's from finding a way to measure 'likes' using actual human behavior passively. For example, on newegg, sorting by 'most popular harddrive' is usually a better proxy than 'highest rated' for the quality I'm looking for. People vote with their wallets, and newegg was among the first to offer that data to inform recommendations. The tradeoff is this behavior helps entrench the winners where markets follow power laws, and that is why Amazon, for example, doesn't offer a similar feature (Amazon's is 'popular and new'). There will be a startup (if there isn't already), who will be acquired by a bank, designed to mine mint data with a POS receipt log and turn that into marketing data (customers who buy these bananas also buy gym membership at ...). JPMorgan (I think?) bought a groupon competitor for this reason. Facebook is tackling this from the other direction by combining browsing history with experian.
  - Cushman 13y ago
    Don't people "voting with their wallets" send a very strong signal precisely because people's wallets are limited? That seems to argue in the direction of the article to me. It also doesn't seem intuitively obvious to me that there shouldn't be more active signals than that with nonzero value.
    - lmm 13y ago
      Indeed, but cash is (almost) continuous rather than discrete. Some kind of scarce recommendation system is probably a good idea, but 10/day feels like a very arbitrary cutoff.
      - robotmay 13y ago
        I must admit that it is pretty arbitrary; it's something I'll be monitoring and revising in future depending on how it's used.
potatolicious 13y ago
So don't call it a "like". Call it something else, say "props" or "kudos". The whole point here is to establish a currency for acknowledging quality, it need not represent actual "liking" of anything. There is a deeper point here which is: what is a like anyways? Clicking a button to indicate you enjoy something has no use for the user. On Facebook users do it as a form of social signaling. On some other sites "liking" is essentially bookmarking for later use. It's important to separate the actual utility of clicking a button vs. the concept notion of liking something. > "You wish they were "recommendations" instead. But this is already a solved problem." You're conflating two things. What you're talking about is sharing - you share a restaurant with a friend, or you share a funny cat picture with your cousin. "Recommendation" in this context means generating, programmatically, relevant things for you to look at. The power of "likes" (or any kind of quality score, really) is that you don't need to know the user's social graph to locate for them content they will be interested in. Even in a social networking context, a quality score allows you to expose the user to good content outside of their immediate social sphere, which on a photography website is certainly a desired feature. The website in question is not about viewing the photo feeds of your internet friends, but about discovering the best content across the entire community, even from users who are completely disconnected from you in the social graph. Unless I'm reading this blog post horribly wrong, when you click on "Recommend" you don't direct this recommendation towards anyone at all. It's simply "I mark this with my stamp of approval, which in turn will help surface this good content to other users of this website". There is not a social angle to this at all, nor any sort of directed sharing. Anyways, I think this is a great idea. There was a website I used a long time ago called thesixtyone. It was an indie music site where discoverability was the core feature. You got "hearts" which you can give to specific songs. Your hearts replenish daily (use them or lose them), making them a currency, and preventing you from just heart'ing everything willy-nilly. There's an elegance to this system that photographer.io can borrow: if you heart something (and therefore expose it to more users) and they in turn also heart the song, you get more hearts to spend. Essentially, users with taste that the community agrees with will get a louder voice, and the community becomes more directed as a result. It also allows you very clear visibility into influencers in the community. There was also a neat gamification angle to it that gave achievements for tasks, with the overall goal of making sure you spend your hearts instead of doing nothing with them.
jeffasinger 13y ago
YouNeedMyGuy is trying to do exactly this. They allow each user to recommend one, and only service/guy in each category.

marchra 13y ago

Google+? Do you work in SEO?

Cushman 13y ago

The claim is that likes are a flawed currency to represent what people like because the cost of clicking a button on a web site trades for an infinitesimal amount of liking, so your representation winds being nearly useless to predict reality. Placing a limit on the "like" action to force people to choose is one potential way to increase the real value of the representation, hopefully extracting a stronger, more usable signal.

yajoe 13y ago
Right, but this solution tries to fit what is otherwise continuous into discrete so his calculations are easier. It is naive and unrealistic. I've never seen that succeed, and the 'approximations' people do with those kind of numbers are scary. Are all likes equal in weight? Does my affinity remain constant over time? Am I capable of liking a countable number of concepts? Do I naturally revisit all items I've liked in my past life and re-evaluate if they make my cut? The solution isn't in limiting likes, it's from finding a way to measure 'likes' using actual human behavior passively. For example, on newegg, sorting by 'most popular harddrive' is usually a better proxy than 'highest rated' for the quality I'm looking for. People vote with their wallets, and newegg was among the first to offer that data to inform recommendations. The tradeoff is this behavior helps entrench the winners where markets follow power laws, and that is why Amazon, for example, doesn't offer a similar feature (Amazon's is 'popular and new'). There will be a startup (if there isn't already), who will be acquired by a bank, designed to mine mint data with a POS receipt log and turn that into marketing data (customers who buy these bananas also buy gym membership at ...). JPMorgan (I think?) bought a groupon competitor for this reason. Facebook is tackling this from the other direction by combining browsing history with experian.
- Cushman 13y ago
  Don't people "voting with their wallets" send a very strong signal precisely because people's wallets are limited? That seems to argue in the direction of the article to me. It also doesn't seem intuitively obvious to me that there shouldn't be more active signals than that with nonzero value.
  - lmm 13y ago
    Indeed, but cash is (almost) continuous rather than discrete. Some kind of scarce recommendation system is probably a good idea, but 10/day feels like a very arbitrary cutoff.
    - robotmay 13y ago
      I must admit that it is pretty arbitrary; it's something I'll be monitoring and revising in future depending on how it's used.

potatolicious 13y ago

So don't call it a "like". Call it something else, say "props" or "kudos". The whole point here is to establish a currency for acknowledging quality, it need not represent actual "liking" of anything. There is a deeper point here which is: what is a like anyways? Clicking a button to indicate you enjoy something has no use for the user. On Facebook users do it as a form of social signaling. On some other sites "liking" is essentially bookmarking for later use. It's important to separate the actual utility of clicking a button vs. the concept notion of liking something. > "You wish they were "recommendations" instead. But this is already a solved problem." You're conflating two things. What you're talking about is sharing - you share a restaurant with a friend, or you share a funny cat picture with your cousin. "Recommendation" in this context means generating, programmatically, relevant things for you to look at. The power of "likes" (or any kind of quality score, really) is that you don't need to know the user's social graph to locate for them content they will be interested in. Even in a social networking context, a quality score allows you to expose the user to good content outside of their immediate social sphere, which on a photography website is certainly a desired feature. The website in question is not about viewing the photo feeds of your internet friends, but about discovering the best content across the entire community, even from users who are completely disconnected from you in the social graph. Unless I'm reading this blog post horribly wrong, when you click on "Recommend" you don't direct this recommendation towards anyone at all. It's simply "I mark this with my stamp of approval, which in turn will help surface this good content to other users of this website". There is not a social angle to this at all, nor any sort of directed sharing. Anyways, I think this is a great idea. There was a website I used a long time ago called thesixtyone. It was an indie music site where discoverability was the core feature. You got "hearts" which you can give to specific songs. Your hearts replenish daily (use them or lose them), making them a currency, and preventing you from just heart'ing everything willy-nilly. There's an elegance to this system that photographer.io can borrow: if you heart something (and therefore expose it to more users) and they in turn also heart the song, you get more hearts to spend. Essentially, users with taste that the community agrees with will get a louder voice, and the community becomes more directed as a result. It also allows you very clear visibility into influencers in the community. There was also a neat gamification angle to it that gave achievements for tasks, with the overall goal of making sure you spend your hearts instead of doing nothing with them.

jeffasinger 13y ago

YouNeedMyGuy is trying to do exactly this. They allow each user to recommend one, and only service/guy in each category.