I was looking through 20x200 today, looking for my (roughly) monthly dose of photography purchasing. I bought three prints and had a hard time wanting anything else. From their “Our Story” page:

As we see it, there are a lot of people out there who want to sell their art and a lot of people who’d like to buy it. They just have a hard time finding each other. The internet is the perfect place to bring those people together, and we’re exactly the right people to make it happen.

I was with them until the last part. At two new editions per week you can be part of the solution but only a very small part. A single gallery hand-picking artists can’t scale, but that’s ok because it couldn’t in the physical art world either. What I can’t imagine is that a network of galleries, whose scale did seem to be enough for the physical art world, can curate something like Flickr. Searching through a billion photos is not easy and even if a bunch of self-anointed gatekeepers could do it it would still be their selection.

State of the Art

The web has achieved tremendous size and Google solved search for anything that is objectively relevant like looking for a programming reference or a photo of New York. What Google does not solve is “find me something I want to read/watch/look at”. I want to sort primarily by taste, my taste, not general relevance to the search terms.

Blogs and Twitter are a decent way to build your own web of curators, but that takes way too much effort for an imperfect result. I have to try and find people whose tastes are similar to mine and follow them. And even though I’m happy with my list of blog feeds, not all posts in all blogs are interesting, so I still have to manually skip quite a few. On Twitter it is even worse as not all of my follows have equal value to me but all are given equal weight. Because I only check it ever so often, 90% of it I don’t ever see.

Hacker News and Reddit have community voting, so instead of trying to read everything I’m subscribed to I am peeking into ever-changing list of “what people with tastes like yours think is interesting”. Except they’re not really tastes like mine, they just happen to be communities that are narrow enough in focus/demographics that I loosely identify with them (Hacker News as a whole and reddit thanks to subreddits, although increasingly less so). The result of that is that what’s on the frontpage at any given time is only a very rough approximation of the kind of thing I want to read.

Build it and They will Come

None of the tools available properly solve this problem. But as Netflix has shown the data exists and the information can be extracted from it. Reddit used to have a tab of articles recommended to me but has since lost that feature, probably due to scaling concerns. Google Reader does have a blog and post suggestion feature that is at least a start.

These days it seems everywhere you go on the web accepts votes, reviews, followers, etc. By aggregating all this it should be possible to implement a search that does indeed sort by the user’s taste. It doesn’t have to be perfect to be much better than the alternatives.

A search by taste would let you find things you like and find the people whose tastes you like, the curators. By continuously accepting or rejecting search results and curators the user’s profile on the web would be continuously improving. In a lot of ways this has already happened with the advertising on the web, and it’s about time it’s turned around and made to work for the user. Google, this sounds right up your alley.

(Written on 13 Nov 2010)