5 ms·

A modest proposal to fix web search

21 points by yanmaani 5y ago

gkrishna 5y ago
Today, I came across a Tweet from Paul related to this issue - https://twitter.com/paulg/status/1477760548787920901 https://twitter.com/paulg/status/1477760548787920901. Looks like this is a growing problem.
wolpoli 5y ago
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?
robbedpeter 5y ago
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.
emptyparadise 5y ago
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.
- jpl56 5y ago
  I found (here [0] on Hacker News) the uBlacklist extension for Firefox, it does exactly that. There's a "block this website" on each Google result. Good bye w3schools . [0] https://news.ycombinator.com/item?id=29546433 https://news.ycombinator.com/item?id=29546433
- Nasreddin_Hodja 5y ago
  > I want a search engine where I can upvote and downvote sites, making them go up and down in the results. It will not work well, popular opinion is not always best. Also search engines already use come kind of upvoting: it is most likely page is being upvoted when you visit it from search results (probably this is one of reasons why many bullshit results on top).
feb 5y ago
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).
- yanmaani 5y ago
  It would be even smarter to replace the paywalled sites with an unpaywalled copy.
danShumway 5y ago
Slightly unrelated, but this proposal is my first time running into tldr pages (https://github.com/tldr-pages/tldr 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?

21 points by yanmaani 5y ago

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

wolpoli 5y ago

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?

robbedpeter 5y ago

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.

emptyparadise 5y ago

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.

jpl56 5y ago
I found (here [0] on Hacker News) the uBlacklist extension for Firefox, it does exactly that. There's a "block this website" on each Google result. Good bye w3schools . [0] https://news.ycombinator.com/item?id=29546433 https://news.ycombinator.com/item?id=29546433
Nasreddin_Hodja 5y ago
> I want a search engine where I can upvote and downvote sites, making them go up and down in the results. It will not work well, popular opinion is not always best. Also search engines already use come kind of upvoting: it is most likely page is being upvoted when you visit it from search results (probably this is one of reasons why many bullshit results on top).

feb 5y ago

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).

yanmaani 5y ago
It would be even smarter to replace the paywalled sites with an unpaywalled copy.

danShumway 5y ago

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