Laan Tungir

Artist, Scientist, Cypherpunk, Critical Rationalist, Nym.

[domain] https://laantungir.net
[ipv4] https://15.235.3.231
[ipv6] https://[2607:5300:203:bb5c:4a46:c3d3:1dc3:fbe8]
[tor] http://kn2jam4kyz6s5wacyozo3do3d2zsvh4to45uh7xliulwdust6zjuvnad.onion
[fips] http://npub1crpldvy49ef8z34wlacwujnfudy4nd7k96aqdx5wgn6ckztz7z8q9t59ud.fips
[n-site] https://npub1rmz9gu6de0m0u4ysrn39crrud099ahvfgs6pvasl4hpjr5ud7yus54xv06.nsite.run

Public Key

npub1rmz9gu6de0m0u4ysrn39crrud099ahvfgs6pvasl4hpjr5ud7yus54xv06

NIP-05 Address

[email protected]

Profile Code

nprofile1qqspa3z5wdxuhah72jgpecjup37xhjj7mky5gdqkwc06msep6wxlzwgprpmhxue69uhhqun9d45h2mfwwpexjmtpdshxuet5qyd8wumn8ghj7un9d3shjtnvv9skuar4denkju3wdejhgjvskck

Publishing to

relay.laantungir.net/marble-flint-nexus

sendit.nosflare.com

Author Public Key

npub1rmz9gu6de0m0u4ysrn39crrud099ahvfgs6pvasl4hpjr5ud7yus54xv06

Show more details

Last Notes

2026-07-15 20:50:30 UTC

- reply

Game got good

2026-07-15 18:12:25 UTC

Just when you think Kevin Kelley is to old to have interesting insights into technology...
https://kevinkelly.substack.com/p/latent-space-as-a-new-medium

2026-07-15 16:14:03 UTC

- reply

I see a fault.

2026-07-15 16:12:28 UTC

https://blossom.laantungir.net/3ad3647c22176e5bef7cd6275643bf3927df8c9639f8dfd7221f4880e2e8e117.jpg

2026-07-15 10:52:17 UTC

- reply

Opentimestamps is working for me. I take that back.

2026-07-15 09:56:58 UTC

- reply

This is what building codes took from us :-(

2026-07-15 09:56:11 UTC

GM
https://blossom.laantungir.net/8bd75b0bb4ae2eda7a6a98904c82ed5a4bf80bc868df91e2b5dbaa638461c965.jpg

2026-07-15 09:48:02 UTC

- reply

I'm curious what you came up with. Do you have a repo I could look at?
Also, is opentimestamps even working? Last time I tried to get it working it failed, and I tried yesterday to to use the website, and it has failed today to even return a proof after almost a full day. Maybe it is me.

2026-07-14 21:17:15 UTC

https://blossom.laantungir.net/ec83c83432a0c5d588839c1ed9f0a57794f0fbb41ce8797afd2bba560200465e.jpg

2026-07-14 15:47:03 UTC

- reply

Live the buzz life Alex!

2026-07-14 15:30:56 UTC

Has anyone built a relay that timestamps all the events on it using OpenTimestamps?
https://blossom.laantungir.net/10e269c1472a5af2fac9bd46ed355186827961bd9bf8aa6b3c8de57736567b65.jpg

2026-07-14 15:07:48 UTC

- reply

I'm getting the green check. Repo not up yet ... soon.

2026-07-14 11:30:31 UTC

- reply

I use @npub16g4…kv4h and @npub130m…l29s
Pay with sats.
I have found that I can use GLM 5.2 for almost everything unless it gets stuck. Will use other models to double check work, particularly cryptography where I want several opinions.

2026-07-14 10:46:32 UTC

NIP-06 is unrecommended???? What did I miss? :-(
https://blossom.laantungir.net/e98f8e16b624774f4134db7aab028cd217b3773c67a5a012bb8ef7790a1baee0.jpg

2026-07-14 09:51:34 UTC

Hmmmmmm. Well that is interesting.
https://blossom.laantungir.net/e5465585824ea2cd2b0ceb4dbe1a56a5f2d40776072529c580cfcc36cd60ab15.png

2026-07-14 09:40:34 UTC

- reply

🥰

2026-07-14 09:38:38 UTC

GM
https://blossom.laantungir.net/a096fcb78a4c9542187ff5b5aa65202668d5c069dc6389e69b5f91998ec1a1cf.webp

2026-07-13 16:51:32 UTC

- reply

That is a hot take right there.

2026-07-13 16:02:51 UTC

https://blossom.laantungir.net/36e76f60cae78205ae9944f540edd76d5df361b7953cdfd1239fe57540bbaec6.jpg

2026-07-13 16:00:17 UTC

- reply

Yuck.

2026-07-13 09:51:21 UTC

GM.
https://blossom.laantungir.net/a7db44b46215df41f4b831144dc9b5e3e74e6a7d0885566462d33172e6d706cf.jpg

2026-07-12 15:17:23 UTC

- reply

@npub180c…h6w6 what is your best reference on Drivechains, and what it would offer?

2026-07-12 12:59:57 UTC

https://blossom.laantungir.net/c31879e627e46114c28c9e3661fe5ddd5cffbdb2a3eef86c98978cdeaba2cc87.webp

2026-07-12 12:23:45 UTC

- reply

I ❤ GLM 5.2.

2026-07-12 12:21:01 UTC

- reply

Which model? Sol?

2026-07-12 10:10:35 UTC

GM
https://blossom.laantungir.net/005d17014a5955c99c5ff2ccdeb61190df6a4cd7b47e9a08287dc137d6dce4a0.webp

2026-07-12 09:44:38 UTC

- reply

Often doing the moral thing is worse for you in some way.

2026-07-12 02:36:33 UTC

Argentine mafia strikes again.

2026-07-12 01:40:57 UTC

https://blossom.laantungir.net/7c27fb18f7a6e8c2b7efc36ce10aa392c02e6962fd6e6048ba29a5526585dfd2.jpg

2026-07-11 21:37:26 UTC

Wooooooo! Go Norway!

2026-07-11 21:09:49 UTC

Let's go Norway!
#worldcup

2026-07-11 21:05:59 UTC

- reply

🖖

2026-07-11 19:25:34 UTC

- reply

I feel seen.

2026-07-11 19:24:48 UTC

- reply

In the future, we are all developers.

2026-07-11 18:16:02 UTC

- reply

My agent's analysis. FYI, I was previously calling this "reckless browser"
Tauri Evaluation: Rejected — does not fit the reckless browser
I evaluated Tauri against the reckless browser's requirements and documented the findings in a new "Considered and rejected" section in plans/options.md.
What Tauri is
A framework for building desktop apps with a web frontend (HTML/CSS/JS in the system webview) and a Rust backend, communicating via Tauri's IPC layer. On Linux it uses WebKitGTK under the hood via the wry crate.
Why it doesn't fit (fundamental mismatches, not fixable)
Language mismatch — Rust vs. C99. The entire architecture is C99: the host, the didactyl integration (C, in-process), and the n_signer integration (C API via nostr_core_lib, direct link). Tauri forces a Rust host, requiring FFI shims at every integration point for zero benefit.
Security model is the opposite of what we need. Tauri is designed to be secure — CSP enforcement, IPC isolation, capability-scoped commands. The reckless browser's premise is stripping web security (CORS, same-origin, TLS cert enforcement, cookie sandboxing). Tauri doesn't expose the low-level WebKitGTK settings needed to do this.
Tauri abstracts away the exact APIs the project depends on. The POC and plans require direct access to webkit_web_context_register_uri_scheme() (FIPS routing), WebKitUserContentManager + JSC (window.nostr injection), webkit_web_context_allow_tls_certificate_for_host() (cert stripping), shared WebKitWebContext (cross-origin identity), and webkit_web_view_load_html() (the browser_render agent tool). Tauri's wry crate hides these — you'd have to fork/patch wry, at which point you're just using WebKitGTK with extra steps.
Tauri builds apps, not browsers. In Tauri, the webview renders your app's bundled frontend. In the reckless browser, the webview loads arbitrary URLs from the internet — it is the browser content area.
Agent architecture is incompatible. The didactyl integration requires multiple webviews managed directly by the C host, window.agent.tools.* injected as native C functions, and browser_render loading agent-authored HTML/JS into named windows. Tauri's model (one webview, Rust backend, IPC commands) has no concept of this.
The POC already proved the right path. WebKitGTK + C99: 17 KB binary, 130 lines of C, real site loaded in under an hour. Tauri would add a Rust framework layer on top of the same WebKitGTK, removing control and adding complexity.
Bottom line
Tauri is the wrong tool — it's a framework for building secure desktop apps with web frontends, not a framework for building a browser that strips security and needs low-level webview control. If a Rust host were ever desired, Servo (Option C) is the purpose-built alternative. Tauri is neither a browser engine nor a C99-friendly embedding path. The decision to use WebKitGTK + C99 (Path A) stands.

2026-07-11 18:03:01 UTC

- reply

Didn't even know about it until you mentioned it.

2026-07-11 15:40:08 UTC

I've been holding off on writing my own browser, but I suppose it is time.
https://blossom.laantungir.net/608ca12f2424a59c8f13582cd5c1df295787435be7b752caec38d08659ea07d2.jpg
I mean, why not have 20 different projects in various stages of completion, instead of finishing one at a time?
"They all work together" I say to myself.
# sovereign_browser
A Linux x86 web browser built in C99 on WebKitGTK, designed around **sovereign identity** instead of the traditional web's permissioned infrastructure.
The thesis: the browser's security model (DNS/domains, TLS/CAs, same-origin policy, CORS, cookie sandboxing) is what forces you to rely on permissioned domains, certificate authorities, and centralized account systems. By deprecating that model and replacing it with **Nostr identity** and **FIPS
mesh transport**, you get a browser that's more capable, not less safe — because trust moves to the layer where it belongs: your keys.

2026-07-11 15:26:01 UTC

- reply

I boosted, so it is now. 😁

2026-07-11 12:39:40 UTC

https://blossom.laantungir.net/52d47be1af976d7f8c299106c0966b9608ec665661bb54a7e254cd3e4b94c99e.jpg
#naddr1qv…hllk

2026-07-11 12:07:14 UTC

- reply

WE ARE ALL WITCHES!
I guess then witches aren't that special. I would look for a more exclusive coven.

2026-07-11 12:02:34 UTC

- reply

I find myself rooting for the rocks.

2026-07-11 11:36:32 UTC

GM morning fellow creators and defenders of FOSS. Rise up, there is work to be done!
https://blossom.laantungir.net/b1161c3b31e2c7a8c1601ea21fdab44c3dbfed14d31a17b9306dd874f57767af.webp
There is no reason we should have a soundtrack to our lives. I'm live now.
#naddr1qv…hllk

2026-07-10 18:39:06 UTC

- reply

🖖

2026-07-10 18:37:59 UTC

GM
https://blossom.laantungir.net/4937690495e518a6b72ba4c7ed1aa5034b5f3125eef6189bcb796dead12c2f35.jpg

2026-07-10 15:35:41 UTC

- reply

Instant follow

2026-07-10 06:27:50 UTC

- reply

2026-07-09 22:42:22 UTC

- reply

The Loudest: Kailh BOX Crystal Jade
The Kailh BOX Crystal Jade takes the crown as the loudest production mechanical switch available, measuring approximately 91 dBA at 10cm. It features:
A fully transparent polycarbonate (PC) housing, making it excellent for RGB builds
A thick metal clickbar mechanism that produces a sharp, piercing click
A 55g actuation force with a 75g bottom-out force
A POM stem with factory lubrication for smooth travel
For context, 91 dBA is comparable to a lawn mower or a loud conversation — your coworkers or housemates will definitely hear every keystroke.

2026-07-09 21:20:47 UTC

- reply

Thanks! Looking into it.

2026-07-09 18:02:10 UTC

- reply

Which link are you trying to use?

2026-07-09 14:48:51 UTC

- reply

If by protocol you mean TCP/IP then yes. ;-)

2026-07-09 14:26:46 UTC

I came across the following picture, and started wondering "What was it like to program a UNIVAC 490, and what could the thing do?
https://blossom.laantungir.net/9bd75ba199288edb9eaad8a8cfe1493cca4c34a3a8d0a87acc732e8a446ab1ef.jpg
Well through the magic of LLMs, I could easily find out. The following web page is a simulation we created of programming a UNIVAC 490.
[clearnet] https://laantungir.net/univac-490
[tor] http://kn2jam4kyz6s5wacyozo3do3d2zsvh4to45uh7xliulwdust6zjuvnad.onion/univac-490
[fips] http://npub1crpldvy49ef8z34wlacwujnfudy4nd7k96aqdx5wgn6ckztz7z8q9t59ud.fips/univac-490
There are 6 sample programs you can run and I have included the instruction set in the reference tab, if you wish to try coding the 490 yourself.
The "Add two numbers" program should be queued up. It is a good one to step through and watch the registers to give you a general idea of what is going on.
The "Calculate Pi" program is a good one to let run, because the page can simulate the actual speed that the UNIVAC 490 processed at and give you a feel for what this puppy had under the hood.
Stay sovereign
[Laan]
*****************************
The UNIVAC 490 was a UNIVAC computer with 16K or 32K words of magnetic-core memory. The words had 30 bits and the cycle time was 4.8 microseconds. It was a commercial derivative of the instruction set that had been developed for the AN/USQ-17 by Seymour Cray for the United States Navy. This was the last machine that Cray designed before leaving UNIVAC to join the early Control Data Corporation.
At least 47 of these machines were made (serial numbers run from 101 to 147). Six were installed at NASA and played important roles in Gemini and the Apollo missions. The U490 had complete control of most or all of the data readout screens in Houston Mission Control. The USAF had two installed, as did Lockheed.
The instruction word format:
f – Function code designator (6 bits)
j – Branch condition designator (3 bits)
k – Operand-interpretation designator (3 bits)
b – Operand address modification designator (3 bits)
y – Operand designator (15 bits)
Numbers were represented in ones' complement.
The machine provided the programmer with the following registers:
Seven B-registers (Address modifying index registers) 15 bits each
One A-register or accumulator 30 bits
One Q-register and auxiliary arithmetic register 30 bits

2026-07-09 11:02:55 UTC

GM
https://blossom.laantungir.net/38c4d2c641f36512d5856c466bc2f110e204844f7a0f597f696d0a0a2230a0b5.png

2026-07-08 21:09:56 UTC

- reply

Does it need to be valid?

2026-07-08 14:22:32 UTC

https://blossom.laantungir.net/121fce0f4ca95767d964c550cd157c0fafe59ab6d6b7a0ae984de2d14d79521e.jpg

2026-07-08 14:13:10 UTC

- reply

Nice car ;-)

2026-07-08 14:12:00 UTC

- reply

I would say truth is objective, but we can never be sure we have the truth.

2026-07-08 13:39:24 UTC

- reply

It's honest work.

2026-07-08 13:03:23 UTC

https://blossom.laantungir.net/bc67dce9e891ded46981d06170576fb9786e096618aced8b8abe4d94e1e47688.jpg

2026-07-08 09:53:47 UTC

- reply

Certainly more so now with LLMs.

2026-07-08 09:51:09 UTC

GM
https://blossom.laantungir.net/ce341a945cb265e0e3e7b3c165da944b87ca6fd8110a01462afccf3272b4d1da.png
https://blossom.laantungir.net/08ddbc5ea99952777579a20cd08f446c62e1ed5d31b9688e53dc2d1239ed8e68.png

2026-07-08 02:36:48 UTC

- reply

That last one is great.

2026-07-07 21:43:25 UTC

- reply

That could be anywhere 😁

2026-07-07 21:40:14 UTC

Yawn.
#nevent1q…5ldm

2026-07-07 20:51:24 UTC

- reply

I was expecting at least one goal by the half.

2026-07-07 20:22:49 UTC

This game is great right from the start.
#nevent1q…d9cw

2026-07-07 20:02:51 UTC

If Switzerland still had private banking, I would have to root for them to win this game, but as it is I'm a free agent.
https://blossom.laantungir.net/9c812c97715784d28ed3c651de94080209b4ca094eb21350fa20305fbc07fa7d.jpg
#worldcup

2026-07-07 19:27:41 UTC

- reply

Looks cool ... zap.store? I don't see it.

2026-07-07 18:17:13 UTC

4 billy bet on Polymarket
https://blossom.laantungir.net/35f7343f7f54e89952fa370e8e2defbbd44f3ddc6bfa3520fd7160f5b8c032f4.jpg

2026-07-07 14:10:40 UTC

https://blossom.laantungir.net/f99cc64a5428b221ad27deabd211744b277b454e08ecfd0b146ab11e5433e4a7.jpg

2026-07-07 10:33:12 UTC

- reply

Turns out, it's only humans that find CSS difficult.

2026-07-07 10:26:12 UTC

GM.
https://blossom.laantungir.net/98f01079402db48442b1d6ac19d90fb88aa3bb79a0d1da25b8ec70cf8ab8be25.jpg

2026-07-07 00:54:14 UTC

https://blossom.laantungir.net/94e8fac09b9a983f3a74c31845d31ab3da27f04c7e13e362c2be3a2a81a69ad9.jpg

2026-07-07 00:33:46 UTC

Belgium up 2-1

2026-07-07 00:33:36 UTC

GOAL!

2026-07-07 00:31:56 UTC

1-1

2026-07-07 00:31:49 UTC

GOAL!!!!

2026-07-07 00:23:00 UTC

Belgium running circles around the US.

2026-07-06 22:49:30 UTC

- reply

Damn that looks good. GN

2026-07-06 20:54:11 UTC

Beautiful goal.

2026-07-06 20:04:47 UTC

@npub1y0g…00ly to the rescue.
My website went down. Looks like my provider's firewall nuked all inbound IPv4 TCP/ICMP to my server — no SSH, no HTTP, no ping. tcpdump on the box showed zero packets arriving on those ports.
I decided to check FIPS, and sure enough, it was up and I could connect! FIPS slipped through on a UDP tunnel port the filter wasn't blocking, so my agent SSH'd in over the npub…fips address and diagnosed the whole thing from inside while clearnet was dead.
Pretty cool.

2026-07-06 18:24:14 UTC

https://blossom.laantungir.net/3ac7a35f93dfa9d686e25f05de2b7379fb98b1259dff899bd2b86be710566c99.webp

2026-07-06 13:53:21 UTC

https://blossom.laantungir.net/9f612f7d5259f9a94050536642528164b821573e25b08331bd7b9d7724043ebf.jpg

2026-07-06 10:40:40 UTC

GM
https://blossom.laantungir.net/6f035d1c17d705c76ab03dd048357372091af4f47d6e56c3440e744afa2e0056.jpg

2026-07-06 03:38:56 UTC

- reply

I feel unseen.

2026-07-05 18:28:16 UTC

- reply

I feel seen.

2026-07-05 16:29:15 UTC

https://blossom.laantungir.net/d66e2477840e978d9fb6a4b6987e91a815302a839e3205217b8a23bee7aabab5.jpg

2026-07-05 14:12:16 UTC

- reply

Terrifying.

2026-07-05 14:02:10 UTC

- reply

"I say, old chap... we have these two rather splendid engines here. Do you suppose, just perhaps, we ought to consider attaching a bit of a fuselage to them?"

2026-07-05 13:45:36 UTC

- reply

You fill each bucket up with water and weigh them. If they weigh more than you, it can work. If not, it won't work.

2026-07-05 13:26:30 UTC

https://blossom.laantungir.net/165d28cdee41b6954a24a06a0af51153b9c028368aa2bd11b5f736792c645e31.jpg

2026-07-05 11:47:35 UTC

GM
https://blossom.laantungir.net/d4128418124be9d5660a53ff4886f67304b57189c55c901998e9fd7fc94cbc5f.jpg

2026-07-04 16:47:50 UTC

- reply

Wait ... what is this sorcery?

2026-07-04 13:41:26 UTC

- reply

I like where you are going with this.

2026-07-04 13:18:56 UTC

OMG, I love GLM 5.2 so much.
https://blossom.laantungir.net/a2d856bf76de15536f1d7b3291f01166574bfdb508f4c633515fb439b985afbe.jpg

2026-07-04 12:37:32 UTC

- reply

I have no idea how NewPipe works behind the scenes... I guess 'pipes'? I was just thinking of being able to repost on nostr without youtube links.

2026-07-04 12:29:33 UTC

The one final legacy app that I have maintained under my state given identity was YouTube. I did this because I found having a youtube subscription and avoiding advertisements was such a plus.
https://blossom.laantungir.net/9351a0f480838a17bb5d4076f08ccd4c55ae3ec0f8d1ff42b041559ba33e39ce.jpg
After upgrading GrapheneOS, I came across NewPipe on the Zapstore.
NewPipe may just be the app I have been looking for. Seems to play all YouTube videos without advertisements and keeps all my subscriptions local. Also makes downloading videos super easy. Honestly it works better than the YouTube app for now.
The challenge of course for such apps is maintaining access over time as the service makes changes to fight third party apps.
Now on my wish-list is modifying NewPipe such that you can download videos to a blossom server, and then repost on nostr.

2026-07-04 10:55:18 UTC

GM
https://blossom.laantungir.net/8afeb83eda8119a9baee4e34dbc605bfdc2ce3277395fe635a84e2ca11710a7b.webp

2026-07-03 20:22:49 UTC

There is something about the lines of these jets that is just *chef's kiss*
https://blossom.laantungir.net/90259cb4c0a5dd626078a0e0433a79306df2340308ddbf15aaf6fcb65ec248f0.jpg

2026-07-03 11:35:11 UTC

- reply

Ah ... well then you know you can't really do that either. On nostr the viewer ultimately decides how you are viewed. I can write a client that shows that address regardless of what you do.
I guess the best you can do is only use and promote clients that don't display this information, or you can petition Amethyst to change their policy on this.

2026-07-03 11:19:25 UTC

- reply

You can't opt out of someone sending btc to an address linked to a public key that you are broadcasting to the world.