Hacker News Reader

Comment by ravenstine | Hacker News Reader

I've been coming to a similar conclusion.

Increasingly I've had to use `site:<domain name>` syntax with DDG, but DDG has clearly become significantly worse in recent months. This was confirmed by others in another HN thread recently. Even if I use the site-syntax or use quotes to specify exact match (which DDG used to honor), I'll often get nonsense results whereas Yandex somehow manages to find results that aren't on DDG or Google.

And yeah, I've been having to use search engines on specific sites like Wikipedia because I know there's gonna be less shenanigans.

Sometimes I wonder if I'm living in the past. Do normies actually use search engines anymore?

View on HN

In the past Google appeared to better honour advanced searches using boolean operators. You could add operators to narrow a search quite effectively.

From my experience, they have diluted that functionality in favour of keyword and context matching against popular searches. Often the thing I want to find is not the most popular thing. Breaking out of that and the search bubbles you get associated with is becoming increasingly difficult. I find myself more and more clicking past the first page if results hoping what I want is there just with less SEO. Rarely is there anything relevant beyond page 3. I definitely remember sometimes digging multiple pages deep into the search results and still finding useful information.

In the early days of Google the speed at which you got the search results was a major factor, I remember being surprised at the speed of response and the quality of the results the first time I was introduced to Google. They still include the search time at the top of the results.

I would prefer to regain strict boolean logic in search at the cost of increased search time. I often care more about actually finding what I want than how fast the result can be returned.

Maybe something like the default search works as it does today as it gives a fast result that has a high likelihood will contain what most people want to see. Then allow for advanced search that strictly adheres to the search terms but could take a few minutes to return the results.

I think there's a market for sites that have optimized search around a specific category of information. For example, imagine you had a hybrid directory+search model, where you could drill down into the directory: Top -> Programming -> Lisp -> Scheme

And from there, search a repertoire of Scheme related articles, tips, blogs, etc.

This model already exists, somewhat, in the form of scholarly articles, medical journals, etc. But there seems to be some innovation needed.

If there were such a "site" that focused on and aggregated Scheme related documentation, you'd think that being listed on that search-directory would be totally acceptable to the website owner. After all, a web article or blog generally wants to be read, especially by an interested reader.

Somehow we need to get back to the web being like a library, with a librarian and a well organized categorical index for references.

Getting humans back into the search loop I think is where we're going to circle to. Like asking your local librarian for information about a topic that is in their special collection.

this is known as faceted search https://en.wikipedia.org/wiki/Faceted_search

on edit: I remember Carrot2 had the ability to derive categories from an index dynamically at one point, so when we put all the Danish laws into it and you searched for murder it generated categories such as Knife Murder etc. anyway here is the online Carrot2 with a search on Scheme https://search.carrot2.org/#/search/web/scheme/folders

on edit2: obviously this Carrot2 online thing is just a demo, they don't have a particularly big or up to date index. I'm also not sure how well Carrot2 would scale as it was, IIRC, an academic project to investigate different ways of searching / organizing search results. Check out the treemap.

Hey - thanks for giving a name for this and the link to Wikipedia. I know it's not a "new" idea, but a faceted search organized and usable broadly across the www would be really useful.

Yandex searches the long tail better and doesn't have really awful politicized censorship like Google does for some searches.

I've been using a browser extension aptly named "Search the current site". Just click the icon and type a search term, while being on that website. It will append site:domain to your search.

Reply on news.ycombinator.com