[denysvitali]

Everyone here is complaining about the search results - but instead I think we should all take some time to appreciate that someone worked hard to create a search engine (including the scraper / crawler part) and making it open source (AGPL).

The results will be improved over time I guess, and for the few search queries I've done - I'm fairly happy with the results.

Kudos to the authors!

[vedranm]

I'm presently on leave at a university in Croatia and my research group's name is Group for Applications and Services on Exascale Research Infrastructure, so the acronym is GASERI. [1]

The first result for gaseri on Stract is our presentation of group's research work, and the second result is our landing page.

Can't complain.

Viva open source, viva AGPL search engine!

[1] https://group.miletic.net/

[nunobrito]

I've tried with an exotic keyword of mine that I use as testing for search engines.

Worked well.

[djbusby]

I searched for a few things that were of the class "you should have this and match DDG/Kagi/G/Y/B" and the top 2/3 were matching.

That's pretty good for a new-ish player.

[forgotmypw17]

I've been doing some extensive searching for a particular topic for months using primarily Google, and I just found a bunch of sites that I had not previously found just running one query on this.

I think that as with ChatGPT vs Bard, the result space is so huge, there are going to be many strength/weakness tradeoffs for any given query.

[O1111OOO]

> I think we should all take some time to appreciate...

There are lots of things in the settings that I really like too: The Manage Optics (Copycats removal, Hacker News, Fediverse, more...), Site Rankings and Explore (similar sites).

I like the overall thinking behind this search engine. It feels like they are creating the kind of search engine I would spec out myself. It starts off on a strong foundation.

I do hope they crawl many of the older pages. Google, Bing (DDG), etal... seem to only index pages for the last 10-15 years or so.

Quibble: "Allow usage statistics" is on by default which leads to this:

"We primarily store the text you used for the search, and which results (if any) you clicked on."

If you regularly clear cookies or have an extension that does this: it's something to keep in mind.

While I intend to keep it _on_ to help the author(s) to help me find stuff - I hope it isn't used to once again make the popular sites - popular. In the process, bury the unpopular older sites (that have nearly completely vanished).

[Gratuity1901]

It is super hard to match the 'big' players. But straight programming a proper index listing is hard.

[millzlane]

>Everyone here is complaining about the search results

It worked to find my obscure website for my company that has only been around for 3 years. I'd say that's pretty damn good.

[ramon156]

> type google > get anything but google

??

[viraptor]

It's an interesting case. The kneejerk reaction is "it should return google.com". But really... why? If I wanted google.com, I'd add the .com. If I wanted to search for something I wouldn't search for google first.

I guess my top candidates would be: wikipedia page about google, google stock chart, recent news about google. google.com would never be a result I want to click. The current results are not amazing, but also do we care what the results for that one are? It's like someone telling you "cow" and expecting you'll know the context of what they're thinking of at the moment. Maybe a heading like "I have no idea what you're on about, here are some clickable ideas: google news, google stock price, ..." would be the best solution?

[Brian_K_White]

It should absolutely return the actual thing you search for if it exists, and as the first result.

It would require an extraordinary circumstance to justify anything else, like if a thing exists, but overwhelmingly the entire net is full of some other reference that is probably what everyone is looking for.

The overwhelming majority of people typing "google" into a search are not looking for the Wikipedia page about google.

[tomw1808]

I was amazed looking at people entering the first part of the domain into the search bar, then searching and clicking the first result. That was before chrome had the integrated google search in the address bar.

I had multiple people (albeit small sample size) who went to google.com, entered the domain they wanted (e.g. news.ycomb) then clicked the first suggestion which ran the actual search and then clicked on the first hit.

[jsgo]

While if you wanted to do so you would probably just add the .com and skip the middle man, I know of quite a few people who use search to get to sites even though they know their URLs.

That isn’t to say I hate what this site is doing, I think it is quite neat, but I do think we have to consider that there are more ways used to get to Google than just entering Google.com

[tmcdos]

Those people should be educated how to create and then (re)use bookmarks/favorites. It's a better tool than searching the same domain every day or multiple times per day.