Scraping is hard. Very good scraping is even harder. And today, being a scraping business is veeery difficult; there are some "open"/public indices, but none of these other indices ever took off
Search indexes are hard, surely, but if you were to strip it to just a good index on the browser, made it free, kept it fresh, it cannot be 100 billion dollars to build. Then you use this DoJ decision and fight against google to not deny a free index to have equal rights on chrome you can have a massive shot at a win for a LOT less money.
I mean... Duckduckgo uses bing api iirc and I use duckduckgo and many people use duckduckgo.
I also used bing once because bing used to cache websites which weren't available in wayback archive, I don't know how but It was pretty cool solution for a problem.
I hate bing too and I am kind of interested in ecosia/qwant's future as well (yes there's kagi too and good luck to kagi as well! but I am currently still staying on duckduckgo)
The small distributed team grinding it out against the goliath. They are awesome and perhaps the right example of what a path like this would look like. Maybe someone from their team can chime in on the difficulties of building a search engine that works in the face of tremendous odds.
It's available in all major browsers (Here in zen browser, it doesn't even have a default browser but rather on the start page it asks between the three options, google duckduckgo and bing but yes if you press next it starts from google but zen can even start from ddg, its not such a big deal)
Duckduckgo is super amazing. I mean they are so amazing and their duck.ai or ai actually provides concise data instead of Google's AI
DDG is leaps ahead of Google in terms of everything. I found Kagi to be pleasant too but with PPP it might make sense in Europe and America but privacy isn't/ shouldn't be the only who only pays. So DDG is great for me personally and I can't recommend it enough for most cases.
Brave/Startpage is a second but DDG is so good :)
It just works (for most cases, the only use case I use google is for uploading images to then get more images like this or use an image as a search query and I just do !gi and open images.google.com but I only use this function very rarely, bangs are amazing feature by ddg)
Same here. It may be 'not very good' for highly specialized or complex technical questions ... but I do research across a broad range of (non-specialized) topics daily. I often need to find 2nd and 3rd points of view on a topic ... or detailed facts about singular events ... and I rarely need to go to the 2nd page. And all ad-free!
It's a remarkable education tool. A curious, explorative kid these days could easily sail WAY beyond their age group using DDG. I can only wish I'd had it.
Their recently added 'Search assistant' consistently provides a couple of CITATIONS to backup its (multi-leveled) responses (Ask for more, get more.) I've seen nothing like it elsewhere. It is even quite good at diggin up useful ... and working ... example code for some languages. Also with citations.
Google Search sucks not because Google is incapable of filtering out spam and SEO slop (though they very much love that people believe they can't), but that spam/slop makes the ads on the SERP page more enticing, and some of the spam itself includes Google Ads/analytics and benefits them there too.
There is no incentive for these companies to build a good search engine by themselves to begin with, let alone provide data to allow others to build one.
No, the customer isn't 'always' right, but these guys like to get big and once big, fuck you, we don't have to listen to you, we're big; what are you going to do, leave?
People complain that user-agent need to be filled. Boo-hoo, are we on hacker news, or what? Can't we just provide cookies, and user-agent? Not a big deal, right?
I myself have implemented a simple solution that is able to go through many hoops, and provide JSON response. Simple and easy [0].
On the other hand it was always an arms race. It will be. Eventually every content will be protected via walled gardens, there is no going around it.
Search engines affect me less, and less every day. I have my own small "index" / "bookmarks" with many domains, github projects, youtube channels [1].
Since the database is so big, the most used by me places is extracted into simple and fast web page using SQLite table [2]. Scraping done right is not a problem.
[0] https://github.com/rumca-js/crawler-buddy
Building a metasearch engine is not hard to do (especially with AI now). It's so liberating when you control the ranking algorithm, and can supplement what the big engines provide as results with your own index of sites and pages that are important to you. I admit, my results & speed aren't as good as Kagi, but still good enough that my personal search engine has been my sole search engine for a year now.
If a site doesn't want me to crawl them, that's fine. I probably don't need them. In practice it hasn't gotten in the way as much as I might have thought it would. But I do still rely on Brave / Mojeek / Marginalia to do much of the heavy lifting for me.
I especially appreciate Marginalia for publicly documenting as much about building a search engine as they have: https://www.marginalia.nu/log/
I started off with a meta-search calling out to Brave / Mojeek / Marginalia, and the basics of that are something that you can ask an AI to make for you as a one-file PHP script. I still think this is a good place to start, because you'll quickly find "okay, I can replace my everyday search engine with this". Once you're dogfooding your engine every day, you'll notice all the rough points you want to improve.
Once you've got an array of objects with Title, URL, Description, and splitting the URL into domain, TLD, subdomain, path, file extension... there's a lot of ranking you can apply just to those. Honestly, a lot of my "ranking" has just been applying increased rankings to domains that I visit most often. I have an array of about 600 domains that it applies ranking boosts to. You can try experimenting with your re-ranking there, before even starting to build your own index.
As for building your own (small, personal) index, the technical details are not as difficult as you'd think. An SQLite database file, that your PHP file reads, will take you a long way... especially if you enable FTS5 indexing. I only did that last week, and I should have done that at the beginning. Search times are 10ms, and not just on my personally curated index of 80,000 pages... I just added a 2nd database with 1.3 Million entries from DMOZ (the old Mozilla Directory), and it's still only about 10ms. My search engine now feels super fast when it gets results from my database. And when it finds zero results, it automatically falls back to the metasearch.
At 1.3 Million entries, the two databases are only about 550MB total. It's running on a shared hosting account and apparently they're not worried - but it's only available to me, so I'm only hitting it maybe 50 times a day maximum. I'll move it onto a VPS eventually, but every time I think "this must be using up too many resources", I find I'm thinking too small by at least a factor of 10x.
For getting started with PHP & SQLite, I found this blog post helpful - but at this point, your AI can vibe code the entire thing for you:
https://davidejones.com/blog/simple-site-search-engine-php-s...
It's amazing how far you can get with just SQLite and FTS5 and a little PHP. Read the Marginalia blog too, there's so much good information in there.
Don't hold yourself back, don't think it's impossible.
Exactly, why can't we just hoard our bookmarks and a list of curated sources, say 1M or 10M small search stubs, and have a LLM direct the scraping operation?
The idea is to have starting points for a scraper, such as blogs, awesome lists, specialized search engines, news sites, docs, etc. On a given query the model only needs a few starting points to find fresh information. Hosting a few GB of compact search stubs could go a long way towards search independence.
This could mean replacing Google. You can even go fully local with local LLM + code sandbox + search stub index + scraper.