Oddbean new post about | logout
 Apparently I was able to pay for a https://kagi.com/ professional plan using only money from zaps today. I didn't know I was getting so many zaps. Thank you. 
 Is Kagi delivering results on par to Google or are premium customers buying into the promise that they will get there once enough people pay? 
 For most things it's better than Google. I have no idea of how they do it nor I expect it to get better, I'm already very happy.

It's much faster than Google too, and that counts a lot. 
 That is quite crazy indeed. So just using it, can you run a test for me? Find me the original Bitcoin chain or Bitcoin torch experiment on bitcointalk. I know I (giszmo) participated and the guy "pirate" who was behind the "Bitcoin Savings & Trust" scam. I would remember more details but with only these details I could not find it in half an hour. 
 Is it this? https://bitcointalk.org/index.php?topic=92658.0 
 Looks like a "bitcoin torch" although there is no "torch" in the name. I found it with the query "site:bitcointalk.org torch pirate giszmo". On Google I get nothing. 
 "Circle of Trust" it was! Thank you fiatjaf! I'll give https://kagi.com a try. 
 Ok, this convinced me. I just created an account. Let's see how it goes ... 
 Hm… you‘re good advertisers. Trying as well. 
 Paid with Lightning, by the way. 
 Kagi is amazing and I also instantly bought myself a subscription when they added lightning support. Maybe should convince them to use btcpayserver over open sats. 
 They seem like they may have a similar ethos to nostr. I will write them to add nostr key support in lieu of email. https://image.nostr.build/6f82b87c398a3f9376c9545d8e748c58dac7650c177e23360c68258f90a30c7a.jpg  
 Maybe they could index Nostr content natively. 
 Tbh it sounds like something that’d be up their street 
 Kagi is a good search engine 
 Do you also get access to the source code or do you pay to use a proprietary non-FOSS program? 
 Do you get access to the source code of your washing machine? Do you get access to the recipes of the places you eat in? Do you have the models of the clothes you're wearing? 
 Lol I don't have any "smart" devices so none of those things are connected to the internet, pinging remote servers, or even worse executing code in personal devices that can be used to fingerprint & track all my other online activities, so yes I use a washing machine & clothes with peace of mind, while I'd never execute non-free programs on my laptop or send all my data to a closed source website. Hopefully that's clear 
 To be fair, even DuckDuckGo is closed-source.  The only fully open-source search engines I know of are wiby.me and YaCy. 
 There is also https://stract.com/ 
 I took a brief look at it and it looks good.  Do you know if it does its own indexing or if it uses something else like Bing in the background? 
 It says on their GitHub page that they have their own completely independent crawler. 
 There is https://html.duckduckgo.com and they have onion addresses as well, which is a good solution for the client-side but of course the engine itself is not free. Just heard about stract from this thread, there is also https://metager.org/ 
 The neat thing about YaCy is that the index is distributed peer-to-peer. 
 Not bad, although your question was about source code, not about sending data to third-party services.

Do you also not use Google or any other search engine, browse the web with JavaScript disabled and use https://github.com/mikedilger/gossip as your Nostr client? 
 I actually disable javascript yes unless I trust the source or can't avoid it, it's an easy toggle, I click many links during my browsing that I can't trust everything.

I do use "things I can't avoid" like google accounts, banking apps, microsoft products at work, slack, my question was with the intent that if you did get access to the source code as a paid user then that could be a good deal, since the "free" means "libre" not "gratis" and there can be FOSS paid apps, but with many good FOSS options already I don't see a reason to give away your freedoms even if a product is too good 
 No, I did not get access to the source, but there are no good FOSS options for search engines (what do you have in mind?) -- although there are for Google accounts and Slack, but I don't blame you for using these. 
 I don't blame you either for looking for the best or most convenient options

nostr:note1vgm0ve3hrudkxgeq2c8n3dz4plujp9vayxvwqwd2ycdye4vc5mmqpqcrq7 
 I have a hard time building this. I mean it builds but doesn’t launch. It’s either some complaints about Wayland libs or not using a very specific glibc version. I suspect egui is the culprit 
 egui is always the culprit. Did you try running the compiled distributed binary?

nostr:nprofile1qyfhwumn8ghj7mmxve3ksctfdch8qatz9uqsuamnwvaz7tmwdaejumr0dshszynhwden5te0dehhxarjw4jjucm0d5hsqg8wzxjalaqvrxj4taqlus453uqwvxxfzgjky2hr0dkzhdnmwmzwfyq630g5  nostr:nprofile1qyvhwumn8ghj7un9d3shjtnndehhyapwwdhkx6tpdshszymhwden5te0wp6hyurvv4cxzeewv4ej7qgewaehxw309aex2mrp0yhxummnw3exjcmp9e3k7mf0qqstlszcex4my59z7nc0ysppptn4qgsmv983nwv8965vmav6d8tgj9qlhqxdu  please help 
 Don't build with the video-ffmpeg feature and most problems will be resolved.   Without that feature it doesn't need any libraries that aren't part of core linux:

$ cargo build --features=lang-cjk --release

$ ldd target/release/gossip
	linux-vdso.so.1 (0x00007fff6f063000)
	libgcc_s.so.1 => /usr/lib/libgcc_s.so.1 (0x00007a64360fe000)
	libm.so.6 => /usr/lib/libm.so.6 (0x00007a6432914000)
	libc.so.6 => /usr/lib/libc.so.6 (0x00007a6432732000)
	/lib64/ld-linux-x86-64.so.2 => /usr/lib64/ld-linux-x86-64.so.2 (0x00007a6436164000)

 
 I want to build with video-ffmpeg, how do I do it? 
 $ cargo build --features=lang-cjk,video-ffmpeg --release

You don't strictly have to use '--release' anymore, but it does optimize the whole thing if you do.  The dependent crates are now always built in release/optimized mode (since those don't get recompiled every time you edit a line and compile again, the cost of doing it isn't borne as badly as if the main crate were optimized every time you just changed a few lines and recompiled). 
 Hi thanks for the help. It builds fine but then it complains about:

- on Nixos not finding libwayland-client or something
- on Fedora 38 it complains about the GlibC version

I think there is some gotcha with the NixOS build that can be worked around but the weirdest thing is Fedora complaining about Glibc version even though it built. Both issues seem egui related.

Not at home right now but will reproduce first chance I get. 
 Best search engine by a lot. Really makes quitting the progressively enshittyfied Google Search easy to quit! 
 But how do they do it? 
 Reading their github they focus on the “small web”, whereas google weighs heavily cerified psy-op and fact-checked “authoritative” websites. 

Seems kagi is going for more long tail organic info. 

Also noticed they like to surface new content <7 days old. 

Sounds like a nostr feed. 
 - They have their own Index, and don't just rebrand Google or Bing search (but they also consider other indexes for some results)
- They have Stephen Wolfram on their board and are integrating Wolfram Alpha in their search results
- They have no incentives to show you enshittyfied content (ads, sponsored content, affiliate links, SEO bullcrap) as they have a classical subcription model instead of data based one.
- They filter out sites with a lot of trackers and ads (20+)
- They boost pages from the "small web" 
 Their search customization options are a plus as well