acatton
Joined 10 years ago
Joined 10 years ago
The small web and indie web audience doesn't use google. They use RSS readers:
https://shkspr.mobi/blog/2026/05/rss-feeds-send-me-more-traf...
https://matduggan.com/you-can-absolutely-have-an-rss-depende...
Small web is there to stay, big tech can't do much about it:
https://www.citationneeded.news/we-can-have-a-different-web/
The only threats to small web are:
* the lack of net neutrality
* lack of competition in the the PC component industry which is the backbone of cheap VPSes and hosting services
* browser monopoly, as any monopolistic browser could impose their small-web-unfriendly version of the web
Airbus currently has two factories finalizing the airplane assembly: one in Toulouse and one in Hamburg. You could copy this model and just open different fab in different countries to spread production.
Also, another model is one country making wafers, one country making EUV-lithography machines and parts, one country mining and refining silicon, etc.
Yes. And I think this is what most tech-literate people fail to understand. The issue is scale.
It takes a lot of effort to find the right clip, cut it to remove its context, and even more effort to doctor a clip. Yes, you're still facing Brandolini's law[1], you can see that with the amount of effort Captain Disillusion[2] put in his videos to debunk crap.
But AI makes it 100× times worse. First, generating a convincing entirely video only takes a little bit of prompting, and waiting, no skill is required. Second, you can do that on a massive scale. You can easily make 2 AI videos a day. If you want to doctor videos "the old way", you'll need a team of VFX artists to do it at this scale.
I genuinely think that tech-literate folks, like myself and other hackernews posters, don't understand that significantly lowering the barrier to entry to X doesn't make X equivalent to what it was before. Scale changes everything.
> [1] https://imgur.com/a/3E17Dts
This is generated on device with llama.cpp compiled to webassembly (aka wllama) and running SmolLM2-360M. [1] How is this different from the user clicking on the link? In the end, your local firefox will fetch the link in order to summarize it, the same way you would have followed the link and read through the document in reader mode.
[1] https://blog.mozilla.org/en/mozilla/ai/ai-tech/ai-link-previ...
The quote is attributed to Soichiro Honda, in the book Driving Honda: Inside the World's Most Innovative Car Company by Jeffrey Rothfeder
Why solve hard problems when you can just lobby your way out of it?
NotJustBikes on youtube has a video listing more of these features which don't exist in cars sold in the US: https://www.youtube.com/watch?v=--832LV9a3I
> Go 1.25 introduced a waitgroup.Go function that lets you add Go routines to a waitgroup more easily. It takes the place of using the go keyword, [...]
99% of the time, you don't want to use sync.WaitGroup, but rather errgroup.Group. This is basically sync.WaitGroup with error handling. It also has optional context/cancellation support. See https://pkg.go.dev/golang.org/x/sync/errgroup
I know it's not part of the standard library, but it's part of the http://golang.org/x/ packages. TBH, golang.org/x/ is stuff that should be in the standard library but isn't, for some reason.
It reminds me a professor who recently told me, regarding ChatGPT use in university: "we're receiving every week applications from foreign students written in perfect German, then when we schedule a call for an interview with the potential scholar, they're incapable to speak either German or English."
Run SSH behind some layer. Some people use Wireguard, and that's okay, I prefer spiped [1] because I can run it as an unprivileged user in a fully hardened systemd unit [2], and I can use ProxyCommand in my ssh_config, which makes it transparent: no need to be constantly on a VPN or to turn it on, I just ssh.
This guide recommends two-factor authentication, which IMHO is overkill and lowers your server reliability by using some random pam authentication modules. Also your spiped key (or your wireguard key) can be considered a second factor authentication.
And a second independent layer lowers the probability of being vulnerable to a 0-day vulnerability on SSH [3] or to Jia Tan [4]
fail2ban means you have a daemon running as root parsing random logs and modify your firewall rules... Yikes... [5] If you're concerned about bruteforce bots, they'll go away as soon as SSH behind something. Also with that layer, you don't need to make you firewall dynamic.
[1] https://www.tarsnap.com/spiped.html
[2] https://ruderich.org/simon/notes/systemd-service-hardening
[3] https://www.qualys.com/2024/07/01/cve-2024-6387/regresshion....
[4] https://en.wikipedia.org/wiki/XZ_Utils_backdoor
[5] Yes I know, you can use as a user, and modify the firewall rules with custom script with an SUID. But nobody does this, actually this guide doesn't do this at all, just everything as root!
From the last archived version of lobste.rs/about: https://web.archive.org/web/20250601002505/https://lobste.rs...
"Lobsters is hosted on three VPSs at DigitalOcean: a s-4vcpu-8gb for the web server, a s-4vcpu-8gb for the mariadb server, and a s-1vcpu-1gb for the IRC bot. DNS is hosted at DNSimple and we use restic for backups to b2. (Setup details are available in our ansible repo.) Lobsters is cheap to run, so we don't take donations."
Currently employed by Google, Inc. as a Site Reliability Engineer.
All my opinions are my own and NOT the views of my current or past employers.