← Back to context

Comment by keiferski

7 months ago

I am increasingly convinced that the successor/replacement to Google will not be a more clever, AI-powered search engine, but a hyper-curated collection of links, selected by people who understand what good content is. Sort of like Yahoo in the pre-Google days.

I think this is only going to become more apparent once AI-generated content takes over the web.

This assumes the breadth of "things I will ever want to search in the future" is contained in whatever these "people" consider to be useful knowledge. Should we create such a group and have them thoughtfully consider every present and past field of knowledge, language, place on earth, political/religious viewpoint, and so on.

  • Well I imagine that searches probably follow a Pareto 80/20 rule to some extent. In other words, 80% of searches are on 20% of total topics. So the hypothetical group would focus on those high volume search topics first.

    • I also imagine the things I search most rarely may be the most pressing compared to the ones I do all the time (as an individual, the things I search most often, I probably know where to go to find the info directly -- API docs and such).

Very tangential but your description is exactly why I've dropped most streaming services except for the Criterion Channel.

I’m working on a pinboard competitor (that is, essentially, just reviving development of it)

One thing I want to look into is ranking algorithms based on individual engagement. So, if you save lots of stories from a site, it ranks higher. If lots of people save stories from a site, it also ranks higher, ect

  • And... when the bots come and "engage", what will you do?

    • 1. Voting rings are one of the easiest types of spam to detect. Of course, the bigger the service the better the bots, but that problem specifically is a later issue

      2. Zero tolerance to bots

      3. The service is not free

      4. Individuals have very little impact on the rankings of other users, so you need to pay for a lot of bots

      I believe that the true problem with bots on, say, twitter, is that they have perverse incentives to 'boost engagement' and whatever by allowing the bots to run rampant.

      1 reply →

I agree but we need to wait the next AI Winter, right now everybody is on the LLM hype train.

I am convinced it won’t be human curated but AI curated.

All information well organized without ads and junk. Like a super glorified Wikipedia with excellent search to dig exactly what you want.