A modest proposal to fix web search

I want a search engine where I can upvote and downvote sites, making them go up and down in the results. Unlike HN, this should not primarily impact that website's global ranking, but specifically the results I get. There should be a way to give -10 points to w3schools and the like, sending them to page 2, or -∞ to Pinterest to get rid of it forever.

If search engine could group results by their domain names, it would make the result a lot more readable. Most searchers would learn that pages from developer.mozilla.org are a lot more useful than pages from blog.hubspot.com. I mean why is Hubspot even blogging about 11 ways to centering css element to begin with?

Web search can be trivially fixed, at the cost of lost ad revenue and surveillance data. Google and Bing could likely unleash phenomenal search capabilities with advanced search modes and regular expressions, but then they would lose what makes their services generate revenue. That's also why nobody else can start a good search engine - you'll get acquired and/or litigated into non-existence.

The results given by the big search engines are the results that earn the most, not the results that provide the most value to users.

Today, I came across a Tweet from Paul related to this issue - https://twitter.com/paulg/status/1477760548787920901. Looks like this is a growing problem.

Slightly unrelated, but this proposal is my first time running into tldr pages (https://github.com/tldr-pages/tldr).

Has anyone used them before (or the variants that the repo mentions)? Are they good? Their live demo seems to be producing decent-ish results from playing around with it. Tried a random sampling of non-default programs on my computer and it's got some of them, but not most of them.

Is this a space I should be looking at more, and is there a dominant project that's getting more attention/coverage of obscure programs?

The proposal consists in filtering out crappy or scoring sites based on how much affiliate links and ads they use. In addition, it could also penalize:

  * GDPR popups
  * excessive JavaScript
  * presence of modal dialogs requesting email

It's an idea I've been thinking about since a long time. It should also include filtering out "scraping" websites which republish content from other websites (like pinterest) and websites which make block content behind pay- or login-walls (like again pinterest).