12 ms·
Infinity: open-source search engine
- softwaredoug 6y agoI wish there was a way to differentiate a “search engine” as in one that crawls the web and a “search engine” like Elastic or Solr. Maybe the latter should be a search server?
- nkristoffersen 6y agowouldn't a crawler be a "web crawler"? It doesn't actually handle the search engine aspects.
- somurzakov 6y agosearch engine vs full-text search engine?
- searchableguy 6y agoGoogle would count as a full text search engine then.
- penagwin 6y agoWeb search engines generally are also full text search engines. This project is useless for people who have their own private text to search.
- ralusek 6y agoWeb search vs application search?
- tekkk 6y agoI think the correct terms would be "web search engine" or "web search" and "search engine" or "full-text search engine". Wikipedia says "web search engine". But I agree, the ubiquity of Google as the "search engine" makes it a bit muddy to acknowledge that a search engine is any program or system that is used to search data. It's an interesting information retrieval concept that can be applied to many things.
- 1_person 6y agoI think it's a lot easier to maintain clarity when we identify this as what it actually is: yet another low effort front end for the Bing search engine's API, which aims to attract and monetize privacy-conscious consumers with a big lie.
- tomphoolery 6y agoI like the term "search database", since it's a database that has some APIs and logic purpose-built for general search engines built on top of it. Elasticsearch, for example, is frequently used as a document-oriented data store rather than a search engine, like in the case of Kibana or APM.
- tgb 6y agoIf the devs are reading this, I found a bug: if you use the wikipedia bang and search, say: "!w knot theory" the actual search page it brings up is for "knottheory" without the space.
- grey_earthling 6y agoAlso if you search for just "!w", you don't arrive at the English Wikipedia front page, which is the behaviour I'm used to from DuckDuckGo. "!bbc" should also lead to the BBC front page, etc.
- adsjhdashkj 6y agoNeat, their ads are powered by: https://www.ethicalads.io/ https://www.ethicalads.io/ This makes me wonder - what ads would be acceptable in the eyes of privacy focused developers? Any? And i don't mean UX - i mean, what ads would you, a privacy focused developer, recommend i install on my site. Assuming i want to fully respect by users privacy, while still attempting to pay for the server costs. Maybe this warrants a AskHN, but it seems relevant here as they're implementing this exact concept. Thoughts?
- utf_8x 6y agoPersonally, I think ethicalads.io and contextcue.com are just fine. Maybe even carbonads.net but good luck getting them to even talk to you without a very large audience already in place.
- nicoburns 6y ago> This makes me wonder - what ads would be acceptable in the eyes of privacy focused developers? Any? Ads that: - don't track you. - Are either textual or images (non-animated) - Are clearly marked as adverts and don't try to blend into the page
- paul_f 6y agoNot denying this is desirable to consumers. Be aware this list of requirements would make any ads run low value to marketers. Retargeting is important to many campaigns, as are ads that grab attention.
- ziddoap 6y ago>what ads would be acceptable in the eyes of privacy focused developers I think it's a pretty straight-forward and easy answer. Don't put ads that breach user privacy. Just use context-based advertisements. If your ads are asking about my canvas size, mouse position, etc. or otherwise are attempting to track me from website to website, I'm not cool with that and will block it whatever way possible. Other dark patterns (blending in to website e.g. Reddit's sponsored posts) are also terrible, but not necessarily related to your question re: privacy. If I know a website is using context-based advertisements with no tracking, I allow the ads and sometimes even follow them if I need whatever is being advertised.
- numpad0 6y ago> When you search for something on our site, we take the results from other search engines and our own indexes,
- searchableguy 6y agoHere are the results: https://ibb.co/album/Mfc6br https://ibb.co/album/Mfc6br Obviously one data point but I searched few more things that I didn't include. Google is better for the same query. Now that I think about it, google's best strength is their crawler and cache infrastructure. I never find latest and up-to date content with the same speed on other search engines. They don't seem to modify results from bing.
- lisper 6y agoYes, AFAICT this is just DDG with a less silly name.
- searchableguy 6y agoI like duckduckgo's name though. It's playful. They also own duck.com Hey, duck this query
- ffpip 6y agoPS: Google was the company that sold duck.com to DDG. DDG also owns ddg.gg
- thinkmassive 6y agoAlso ddg.co
- qihqi 6y agoCurrently its just Bing: https://gitlab.com/infinitysearch/infinity-search/-/blob/master/MainApplication/Aggregator/CombineResults.py#L46 https://gitlab.com/infinitysearch/infinity-search/-/blob/mas...
- faitswulff 6y agoIf GPT-3 is trained on a corpus of the entire internet, could you just use it as a search engine by posing questions to it?
- freediver 6y agoNo, but it can look deceivingly able to: https://twitter.com/paraschopra/status/1284801028676653060 https://twitter.com/paraschopra/status/1284801028676653060 (and this is how to prime it: https://twitter.com/vybhavram/status/1284803750146617344 https://twitter.com/vybhavram/status/1284803750146617344) The main problem is that GPT3 has a free will of a creative but delusional person and will distort facts as it pleases to maximize its output function score. So, similar to evening news.
- worldsayshi 6y agoHmm, can you turn GPT-3 api into a fact checker by priming it with tuples of (in)correct statements, true/false? And then maybe prime it to add explanations and references for the responses as well.
- wizzwizz4 6y agoYes, but it's only effective for things a few months before its creation; anything after that is just “sounds plausible to GPT-3”. Likewise, its explanations aren't very good unless it has good domain knowledge on the subject; if it doesn't know something, it won't tell you so unless you've primed it with not-knowing being an option. https://www.gwern.net/GPT-3#expressing-uncertainty https://www.gwern.net/GPT-3#expressing-uncertainty
- petra 6y agoAny info on search operators ? And do they work, unlike Google ? And where's the source ? could i create search plugins just for my own use ? because that would be a killer feature. So it's only technically open-source, it uses Bing's API probably for their "web" tab.[1] Their "infinity tab" doesn't give good results at all, currently. This is what may run on their own system. Maybe. [1]"How It Works For the web application, we use Flask. For the search results, we use the Microsoft Cognitive Services Search API and the DuckDuckGo Instant Answers API. We will have more detailed information about how our system works in the future." -- from their source : https://gitlab.com/infinitysearch/infinity-search https://gitlab.com/infinitysearch/infinity-search
- searchableguy 6y ago> If you have ever used DuckDuckGo before, we integrated the same bang feature as them so that you can redirect your searches to other websites directly from our site. Here is an example: https://infinitysearch.co/results?q=!ddg https://infinitysearch.co/results?q=!ddg privacy Right now, we have the same ones as DuckDuckGo so you can use the same ones found at https://duckduckgo.com/bang https://duckduckgo.com/bang for now. > The whole Infinity Search system is open source and the code can be found at https://gitlab.com/infinitysearch https://gitlab.com/infinitysearch. From the FAQ here: https://infinitysearch.co/why https://infinitysearch.co/why
- wirelessbrain 6y agoWhat's the difference between "web" and "infinity" search? It doesn't seem to be explained anywhere.
- fsflover 6y agoSee also: peer to peer free search engine https://yacy.net https://yacy.net
- DoctorNick 6y agois this just a searx instance that's been loaded with ads?
- Dowwie 6y agoVery cool to see innovation emerging from Oklahoma!
- didip 6y agoThe video search is broken for me, tested on Chrome with the keyword "golang". It's just showing blank.
- rmetzler 6y agoI find it really disturbing that you can't link to result pages. I think this is some kind of REST principles violation.
- Mandatum 6y agoGuessed the param on first try: https://infinitysearch.co/results?q=test https://infinitysearch.co/results?q=test
- kgraves 6y agoI saw everything this search engine has to offer, all looks good except their 'Ethical Ads'. What does that even mean? There should be NO ads. period. I would go so far to say that ALL ads are a distraction. You're supposed to be an open source privacy focused search engine, not a surveillance capitalist serving search ads like Google, this is NOT OK.
- hinkley 6y agoHow do you propose to pay for the infrastructure?
- oska 6y agoApart from you question not being in good faith (because there have always been business models to support publishing and providing services that do not rely on adverstising and these models are well known), the main point is that they do not have to provide the answer to this question. Push advertising is unethical and culturally toxic and thus is not justifiable using the simple pragmatic argument of "It pays the bills". You can't justify unethical behaviour using a simple pragmatic argument. (Quite obviously; this shouldn't even need to be said.)
- hinkley 6y agoTANSTAAFL shouldn't have to be said either, and yet here we are. This would be an interesting ethical discussion if it weren't predicated on a conversation every parent has with their 14 year old. Yes, I know you want this thing, but things don't just magically appear. It all takes effort. Are you going to contribute? Or do you just expect other people to do stuff for you or you'll throw a tantrum? You are getting older now and this is not acceptable behavior. The problem is not ads. The problem is that people want things for free. If you get things for free you're not a customer. If you're neither the customer or the vendor in the transaction, then you are the product. Ads are by far the most obvious form of this but there are others actors who would be much, much worse if they were the customer. If 'somebody worse than ads' are the only customers, then they will just get more brazen. Like loan shark brazen, or possibly police state brazen. We need a way to pay for things that's not a paywall, and not a surveillance state. Given the other conversations in this thread, this company is trying to be neither. Whether they succeed or have ulterior motives I have no way of knowing. If you have the name recognition of Wikipedia or PBS, you can to a fund raising campaign every year. You can seek funding from a foundation of some sort. If you're just a few people with an MVP? Neither of those will work for you. (I suppose maybe GoFundMe might, but I wouldn't count on it.)
- Tanath 6y ago> You will need to enable JavaScript to view the results. If you do not, the loading bar below will keep staying that way. Really? Or you could fix your site.
- Animats 6y agoHm. Another company in this space, like Neeva. Are people seeing opportunities going up against Google?
- esperent 6y agoThis is the first one I've seen that's open source, however.
- surround 6y agoI’d like to mention that DuckDuckGo’s backend is not open source.
- declank 6y agoNot sure why this post is getting scathing criticism, but this would be a nice project to fork for a metasearch engine.