3 ms·
I've been looking into alternative search workflows. Google (and near equivalents) feel a bit like Windows GUI when I want Linux cli for a) quickly going to a
by ZeroGravitas 5y ago
I've been looking into alternative search workflows.
Google (and near equivalents) feel a bit like Windows GUI when I want Linux cli for a) quickly going to a site I already know, b) answering a question when I have a bias about where good answers will be found, c) picking one thing from a long list
I feel there's already an equivalent with a few rough edges but a power tool I can be productive with.
Any product that aims at Google's entire market is only going to turn into Google, I need an ecosystem of tools so there will always be one that is aimed at my use case.
It feels like there's a common sentiment around this so there may be room for a new business model or open source approach.
One thing is bang searches, like duckduckgo has. So if you already know the site you want to search, and they have decent search then you can use their search directly.
There's an open standard for advertising your local search but it feels a bit forgotten like RSS Vs twitter. You can find lists of sites that advertise this search in the Mycroft project on mozdev.
I think Firefox has a non standard extension for search suggestions too.
If the sites don't have decent search, you can get a site specific search via Google or whatever, but it feels like site specific search for code related things would be a good niche. Like Google, you don't need to own the content, just index it better by adding intelligence from context.
You can even aggregate across a few different sites, like official docs plus official discussion, email archives, a Reddit sub and so on as appropriate. Yes you'll miss the long tail, but you'll have limited the scope so much that there would be a lot for value in exact searches for errors and other similar things.
Possibly language communities could build their own index?
The same for discussion sites seems possible and other general groups of search tasks. Wikipedia is a good target too.
If you built a good one you could then offer it to the sites to use as their internal search too, in a virtuous cycle.
On that note I'm going to go look up what Algolia does in this space.