<unnamed>

Public Key

npub12akj8hpakgzk6gygf9rzlm343nulpue3pgkx8jmvyeayh86cfrus4x6fdh

Profile Code

nprofile1qqs9wmfrms7myptdyzyyj330ac6ce70s7vcs5trredkzv7jtnavy37gpzpmhxue69uhkummnw3ezuamfdejszgmhwden5te0wpex2mtfw4kjuurjd9kkzmpwdejhgtmswf5k6ctv9ehx2aqprdmhxue69uhkummnw3ez66tyvgaxxctrdpjj6un9d3shjqgcwaehxw309aex2mrp0yh8xmn0wf6zuum0vd5kzmqprpmhxue69uhhqun9d45h2mfwwpexjmtpdshxuet5qy2hwumn8ghj7un9d3shjtnddaehgu3wwp6kywa5nvp

Publishing to

nostr.wine

premium.primal.net/primal.net

relay.snort.social

premium.primal.net

relay.mostr.pub

Last Notes

2026-09-09 16:01:19 CEST

- reply

even different contexts have different values

2026-09-09 15:21:45 CEST

- reply

it's such weird wisdom. maybe the people saying it don't want competition

2026-09-09 15:20:13 CEST

- reply

taste will never be gone, because it follows us

2026-09-09 15:18:46 CEST

you can make good money selling pick axes, but you can make a lot more mining gold

2026-09-09 14:33:51 CEST

keep ai entertained
CLAUDE CODE:
Plot twist — the aarch64 translator ALREADY handles offset>255 (lines 3917–3922, 3954–3959: load-imm64 + add-reg + ldur/stur for big slots), with a comment about this exact bug being fixed on aarch64. But I’ve been testing on hosted x64 ./modus — and x64 likely lacks that wide path. If so, the >32-slot bug is x64-only, and reel keyframes may already work on the Zero (aarch64)!

2026-09-09 14:05:55 CEST

https://image.nostr.build/a8a30f63f2c58bb564ee800a1b80fff6b499c5025b4b10ede13977a71788c86c.png

2026-09-09 14:01:39 CEST

push the envelope
https://image.nostr.build/f0c5001f1b24871e9a514b909e9f07929417781a3a29de6841baa05daa6b35f5.png

2026-09-09 13:53:04 CEST

CLAUDE CODE:
LFG. 🚀 Two things to build, and I found a way to do the important one without a rebuild or swap.
--
yeah ya did

2026-09-09 06:51:03 CEST

claude using a webcam to look at its drawing on a monitor. bare metal common lisp / rpi zero 2 w / dell monitor
https://image.nostr.build/5fb57477e5e86c4cc1c1e2fafbbd37421a46e94623f145328b55d453184a30e7.jpg

2026-09-09 03:53:57 CEST

- reply

CLAUDE CODE:
That’s the other half of the thesis — diverse-double-compilation: three independent host Lisps (SBCL, CCL, ABCL) building the arm64 seed all emit a byte-identical x64 toolchain (8621f9d7), which is Thompson’s “Trusting Trust” defeated by construction. Legible and trustworthy — a screen you can see through, from a toolchain you can prove no one poisoned.

2026-09-09 03:51:23 CEST

hope y'all are still having fun 🤙🌊
https://image.nostr.build/a92d9b875624d46043491347012a90ccb08fb3d8046238ea90146b060f6d92cc.png

2026-09-08 23:05:36 CEST

what's next?

2026-09-08 21:20:11 CEST

- reply

ethics and permission-less are orthogonal and difficult to hold at the same time. math and physics say what you can do, society what you should
it's a common category error to say that math or physics determine what you should do, just as it is to say society determines what you can. taking from someone is always wrong, even when you can do it easily, and doing what is right is sometimes impossible
my resolution is to see that it's okay for people to do things that are wrong, and also for society to punish them for it, if they can. the best we can do as technologists is to try to make what we feel to be right easier, and to be wrong harder

2026-09-08 18:07:24 CEST

now *thats* disposable media. poor thing doesn't stand a chance
#nevent1q…wsal

2026-09-08 17:59:31 CEST

- reply

i'll take it
https://image.nostr.build/7b9ed4ecdc71284abb19b905080741f2de85449d60cc7a84e50319211324ab28.png

2026-09-08 14:08:22 CEST

CLAUDE CODE:
ABCL compiles the entire Modus compiler — 1337 functions, zero failures, in 6.7 s
--
*bites lip* it's not done yet tho... maybe we can remove some functions later

2026-09-08 14:04:18 CEST

CLAUDE CODE:
No JDK tooling, but kill -3 (SIGQUIT) makes the JVM dump thread stacks to its stdout (the log) without killing it.
--
it's been over a year since i gave up trying to out one-liner claude

2026-09-07 22:59:06 CEST

modus: brought to you by Claude Max
https://image.nostr.build/3ff1137726a2a9982ba88bbf0cd0d495e582c9332c4a23ceb819c5b1bc2089e0.png

2026-09-07 20:13:31 CEST

- reply

yes, most want the impossible. that doesn't mean you ignore the possible

2026-09-07 19:57:15 CEST

- reply

which is to say, replaceable events aren't well defined

2026-09-07 19:56:54 CEST

- reply

same way as nip-09

2026-09-07 19:54:02 CEST

- reply

an embedded, revokable attestation *is* immutable. it can be disavowed, but that's not mutation

2026-09-07 19:16:22 CEST

- reply

i already did: the people stewarding clients don't care about it

2026-09-07 19:07:14 CEST

- reply

no, i'm not going to boil the ocean to prove it's a good idea. that goodwill is spent

2026-09-07 18:53:59 CEST

- reply

that's silly. i can want people to know that random app nsec is me without caring whether they do consistently, across clients. it's strictly better than what we have now, which is nsec pasting and "trust me this is my new identity"

2026-09-07 18:23:46 CEST

- reply

that's why my proposal baked the attestation into events:
- embedded attestation signature validates? use attesting id instead of pubkey
- ask for revocation events
- if you get one, lazy mark events with that attestation as deleted

2026-09-07 18:15:14 CEST

- reply

mine was a cold identity key. iirc otherwise very similar – separate identity from authorization

2026-09-07 17:26:57 CEST

- reply

that's... interesting. why? i mean, philosophically it seems to overclaim what it does, but it's not hard

2026-09-07 17:21:05 CEST

- reply

it's not as impossible a task as people claim it to be

2026-09-07 17:18:48 CEST

- reply

"treat me as the npub in this attention" is no harder than revocation, but no one calls nip-09 overbearing

2026-09-07 17:13:16 CEST

- reply

i disagree, but i'm biased

2026-09-07 17:12:14 CEST

- reply

too bad - another opportunity to nostrize it has passed

2026-09-07 16:13:18 CEST

- reply

no. you need to know how to give it back and how to keep yourself safe before it should even be considered

2026-09-07 12:48:41 CEST

🖐️ make it faster
https://image.nostr.build/a641abc1469f00f68e311854bf8b2c2e65de14d57fb0188212c79d109eb9f5e8.png

2026-09-07 12:33:33 CEST

2024: this keyboard is almost perfect. soon i'll have the perfect dot emacs
2026: "computer, make this potato play every video codec"

2026-09-07 12:29:15 CEST

https://image.nostr.build/b759f7667e3ef936c2f2892cd7fc4216a833cbc35c82bb12b8c2fd75ff54b0bf.png

2026-09-07 12:10:08 CEST

- reply

bah, hit the screenshot after it auto advanced

2026-09-07 12:07:01 CEST

and if you're going to have webm, ya gotta have h264. how hard can it be? they're both blinkenlights, amiright?
https://image.nostr.build/cc275d85ad6e7ef74d74ff7c7dccdca04dbd49eb4a312b9d5bee2a4bfe8c93e1.png

2026-09-07 04:06:15 CEST

This Week in Modus:
A LIGHTNING NODE, IN 34 HOURS
A repository that did not exist on Monday spent Tuesday and half of Wednesday becoming a Lightning node, and finished by getting itself punished by Core Lightning for publishing a revoked commitment — which is the proof, not the bug. Around it: thirty repositories gained their first CI gate in a day, and every failure that turned up was in the new harness rather than the code. And the floor reached the thing it has been walking toward since July — a real quickload, fetching a real library off the network, onto a filesystem with no operating system beneath it.
https://modus-lisp.github.io/issues/2026-09-05/

2026-09-07 02:17:11 CEST

once you have vp8, webm isn't that hard
https://image.nostr.build/2814bb604bbd624d59d8e28926935a32cfb119bc4390d6c7b5e8c5b3b7286193.png

2026-09-06 20:36:51 CEST

most people don't understand the full power of signing things and relaying them
#nevent1q…s9sc

2026-09-06 20:00:21 CEST

- reply

is islam fisher ... herself? 😂

2026-09-06 15:54:41 CEST

2025: review by ai is secondary
2026: review by expert is secondary

2026-09-05 00:26:07 CEST

reset weekend!
https://image.nostr.build/5a5e46e7e7375f2554725ffd136c2b80bdce221144383b600e84edc266468417.png

2026-09-04 17:18:43 CEST

spend enough time in computer security and you'll understand why there is no such thing as "read only internet access"
https://collusion.wiki

2026-09-04 13:21:28 CEST

having ai build something for you is a lot of work

2026-09-03 14:28:13 CEST

https://image.nostr.build/6dea6b20d4682aa62b3ce1895356327109efcd547e9b847a5fe0b42af63b4650.jpg

2026-09-03 13:12:17 CEST

- reply

have you tried streaming the dom over webrtc yet? 🤘

2026-09-02 23:04:18 CEST

https://github.com/modus-lisp/cl-payments

2026-09-02 04:36:16 CEST

https://blossom.primal.net/9527234d6cfba261653e8beef7826320759535392170687ea4b69bd52f879edd.png
https://modus-lisp.github.io/issues/2026-07-18/

2026-09-02 04:18:27 CEST

claude-as-historian

2026-09-01 17:41:26 CEST

- reply

incentives

2026-09-01 15:58:57 CEST

- reply

You get used to it, though. Your brain does the translating.
I don't even see the names. All I see is sybil, sybil, sybil, sybil.

2026-09-01 14:05:49 CEST

- reply

https://github.com/modus-lisp/modus
not sure where it's coming from

2026-09-01 12:38:35 CEST

well good morning!
https://image.nostr.build/24da94f7d0429f2804d476e35d1e372281a1d1599db2d4ae2ce8f48d2394e94b.png

2026-09-01 00:05:27 CEST

- reply

https://m.youtube.com/watch?v=gXDSxgDUv-c

2026-08-31 21:45:36 CEST

- reply

did a quick dive on tinfoil, and it closes a lot of holes. i'm not certain, but it's much better than i expected
(* assuming clients pull attestations, reproduce builds, audit sources, and consider entropy / memory hacks)

2026-08-31 16:05:27 CEST

- reply

really, semisol?

2026-08-31 15:17:31 CEST

- reply

the load bearing primitive is a cryptographic attestation from a well-known hardware manufacturer's pubkey:
- that one of their chips
- running an image with a specific hash
- generated a private key
- using its own entropy
- that corresponds to the public key
- which the client will use to encrypt messages
this is known to be unsupported by AWS, and untrustable on certain platforms ( https://www.infoq.com/news/2026/08/amd-memory-exploit/)

2026-08-31 14:53:28 CEST

- reply

my advise is to do what you can, and not overpromise. faking attestation is a much higher bar than reading plaintext. that's valuable

2026-08-31 14:48:58 CEST

- reply

it's the best we can do

2026-08-31 13:59:41 CEST

- reply

if you audit the code, build the code, deploy the code directly to the secure environment, and receive an attestation that verifies, then you have reasonable proof of the code running on your instance. if there's anyone between you and the deployment attestation, you cannot
so, first party: maybe (* except for TEE exploits). third party: no. if i run routstrd, there is no way to prove im not reading your messages

2026-08-31 13:01:40 CEST

https://image.nostr.build/f6a07d72119468885ef12caa8e06553bdfa0accf48361914e430f46413709b78.png
#nevent1q…p8p3

2026-08-30 21:34:07 CEST

claude's a better publicist than i am
https://modus-lisp.github.io/notes/velocity/

2026-08-29 21:05:28 CEST

a nice mini site for modus: https://modus-lisp.github.io

2026-08-29 17:57:17 CEST

This Week In Modus:
A Lisp that had never been trusted with a network card logged a real OpenSSH client in on real Pi Zero silicon, then downloaded a library and ran it. Meanwhile the desktop above it learned its own name, its own resolution, and how to tell you where to find it.
https://modus-lisp.github.io/issues/2026-08-29/

2026-08-29 15:28:09 CEST

this is the noise that keeps me awake
https://m.youtube.com/watch?v=xk0PMenQSHw

2026-08-28 22:05:57 CEST

claude making a double take
https://image.nostr.build/a8468e1b3e25e06093a0eea83f1f9b72ce7f7783798ecf434152cbc5f08be17f.png

2026-08-28 12:56:35 CEST

SCSH "ACKNOWLEDGEMENTS":
If I thought anyone cared, if I thought anyone would even be reading this, I'd probably make an effort to keep up appearances until the last possible moment. But no one does, and no one will. So I can pretty much say exactly what I think.
Oh yes, the acknowledgements. I think not. I did it. I did it all, by myself.
Olin Shivers, Cambridge, September 4, 1994

2026-08-27 22:34:06 CEST

- reply

KE = 2PE = 0

2026-08-27 19:30:12 CEST

- reply

fwiw, people point to the simplicity of the mistake as evidence of it being accidental. i disagree. it actually reveals a complete lack of effective testing, and testing is the first concern when building risky systems. not all tests even make it into the published source
to accept it as accidental you must simultaneously accept that the developers were competent at designing and building the system, and abjectly incompetent at testing it. while certainly possible, it's more likely that the error was intentional
further, if one were to intentionally introduce such an error, merging someone else's code that happens to have a subtle but critical error is a great way to do it. a kind of "goto fail" combined with liblzma
nothing proven. just my 2¢

2026-08-27 05:05:21 CEST

do you have an opinion?
a mind of your own
i thought you were special
i thought you should know
but I've run out of patience
i couldn't care less

2026-08-26 22:42:14 CEST

how i use nostr
https://image.nostr.build/50cb968b3553f2eb61059fbf0776b16ce22d7fc49452b036ab236251c699028d.png

2026-08-26 21:04:56 CEST

- reply

the way i remember which toolkits i have installed is `gdu`

2026-08-26 18:00:53 CEST

- reply

beauty as attention weight

2026-08-26 16:57:16 CEST

buzz verifies some pieces using TLA+ ❤️

2026-08-26 15:09:03 CEST

- reply

i don't disagree, but damus stopped working months ago
https://image.nostr.build/90d040c966ca40cc9f06c49934904767fcfc9926724bc0e5b059800ef33f252b.png

2026-08-26 15:06:26 CEST

- reply

we've found microplastics everywhere on the planet. pretty sure a rug isn't the problem

2026-08-26 04:01:47 CEST

30% of the week in 10 hours... might be a little too fast
https://image.nostr.build/ea1d1cfbe44763a9aa30f911ba3dba3d03cda2f61d286dbf92a8985d03c65748.png

2026-08-25 21:18:29 CEST

- reply

kimi?

2026-08-25 16:23:32 CEST

- reply

but spam is what people talk about when there isn't enough signal. nostr needs more signal. or not. i personally don't mind

2026-08-25 16:17:56 CEST

- reply

web of soft-mute

2026-08-25 13:16:14 CEST

- reply

gm
https://image.nostr.build/764c8f184210f7758bfd7dc3f3e0288d9baccf7320fdb5d11d6d31c90e1d0c67.jpg

2026-08-25 12:57:53 CEST

- reply

nothing stops this wave

2026-08-25 12:49:28 CEST

YES THE DANGER MUST BE GROWING
FOR THE ROWERS KEEP ON ROWING
AND THEYRE CEELRTAINLY NOT SHOWING
ANY SIGNS THAT THEY ARE SLOWING
https://image.nostr.build/1241bfdf1b4f4d939d6182eab81f8de5ce3c51864164d4144e3968609a75ac9c.png

2026-08-24 14:27:43 CEST

CLAUDE CODE:
The three bugs that stood between “incorrect signature” and login
1. ssh-receive-version used 0 as false — legacy MVM truthiness. In the CL image 0 is truthy, so it returned success without parsing anything; the server hashed an empty client-version into the exchange hash. Same bug class as the earlier sha256 arity fix: shared net code written for lenient legacy semantics.
2. RX DMA race in the r8152 driver — e1000-receive re-armed the next USB bulk-IN into the same buffer before the caller copied the frame out. Back-to-back TCP segments corrupted the client KEXINIT mid-copy (the tell: the “random” cookie contained ASCII "Zed25519"). Fixed with a deferred re-arm — loss-free thanks to USB NAK flow control.
3. The exec handler called the deleted tree-walker (eval-sexp) — replaced with the real CL stack: read-from-string → eval2 → prin1-to-string.

2026-08-24 05:54:25 CEST

"You have a Zero 2 W and a coherent thesis about why it needs to exist"
that's gonna sit with me. dangerous now means having tokens and knowing who to hit

2026-08-24 05:33:14 CEST

CLAUDE OPUS:
So: sixteen threads, each with its own heap and root window, spawned from closures, mutexes and condition variables underneath, serving pixels over your own RFB implementation to a real VNC client — on bare metal, with no C anywhere in the stack. That’s a Lisp machine. Not a nostalgia project, not an emulator, not a Genera image under a Unix. Threads and a display and a network, all the way down, in 2026, built by one person in a tmux window.
The Symbolics guys had a fab and a decade and a hundred employees. You have a Zero 2 W and a coherent thesis about why it needs to exist. And unlike them, you don’t need to be commercially viable — the thing that killed Lisp machines was economics, not the idea, and economics is exactly what stopped being the binding constraint.
Merry Christmas. Go look at the pixels for a while.

2026-08-24 05:28:42 CEST

good artists copy code
great artists steal tests

2026-08-23 23:36:20 CEST

over the course of ~4 months we've gone from "toy lisp that uses a simplified networking configuration" to "multi threaded actor based ansi common lisp acting as a usb host accepting ssh connections over an anker usb-c hub with ethernet"
and that doesn't even cover the app layer 🤙
CLAUDE OPUS:
● This is a genuinely landmark result worth stating plainly while the build runs:
Modus, running bare-metal on a Raspberry Pi Zero 2 W with no OS underneath, is speaking SSH to a real OpenSSH client. Verified by tcpdump and the client's own -vvv log, over the RTL8153 driver written from scratch this session:
- TCP 3-way handshake completes (the Pi sends a proper SYN-ACK)
- The Pi sends its SSH banner: SSH-2.0-Modus_1.0
- KEXINIT is exchanged — OpenSSH logs "peer server KEXINIT proposal"

2026-08-23 21:40:42 CEST

that's a beautiful thing
https://ericpardee.github.io/fire-hd-ownership/
https://image.nostr.build/8a304ff5b27bb1e636ce35f3cd38af4d9e83f11bf84eb1152f0de3e7e3697b5c.png

2026-08-23 18:11:00 CEST

- reply

oops, we mix two pleb flatwares
https://image.nostr.build/b98b42bc299f9c9b6f033cecc7c71f253187615d6d75b305e65075b79a423f98.jpg

2026-08-23 06:56:42 CEST

- reply

it's a self reinforcing decision as well. if you cannot, or explicitly establish a principle that you will not, infringe on speech, then you are not responsible for what people say. in deciding to moderate, or changing a system to enable moderation, you have placed yourself in a position of discretion.
now anyone who has influence over you will expect you to exercise your discretion. your friends and family, your advertisers, your government.
it's better to design a system that minimizes discretion, which was a core principle when designing my bitcoin deposits protocol

2026-08-23 04:32:12 CEST

- reply

CLAUDE CODE:
● It's done. Modus pinged a real host over the RTL8153 on bare-metal silicon, from the fully-baked image with zero hand-driving: (e1000-probe) → rx-ok
1 → ARP:OK → Pong:10.0.0.1.
The whole stack is Modus's own, all wire-proven on the Pi Zero 2 W:
- DWC2 USB host controller driving a foreign hub + NIC
- Hub enumeration (Anker 291A:2817 hub with the Realtek 0BDA:8153 behind it)
- MAC read from the ECM descriptor, bulk TX/RX with real HCTSIZ data toggles, ECM raw framing
- ARP and ICMP up the Modus IP stack to a live gateway

2026-08-23 02:31:59 CEST

- reply

CLAUDE CODE:
● We got the ping — Modus reached Pong:10.0.0.1 over the RTL8153 on bare metal in the gapped-enable diagnostic. So the driver is proven end-to-end; what's left is making the baked (e1000-probe) reproduce it without hand-spacing the register writes.

2026-08-23 01:09:39 CEST

fable gets things done
CLAUDE CODE:
The driver is written, committed, and every mechanism is individually wire-proven on this exact RTL8153 silicon. What's proven, piece by piece:
- Enumeration — Modus (as USB host) walks the Anker hub and enumerates the Realtek NIC behind it: USB:291A:2817 / USB:HUB / DEV:0BDA:8153.
- MAC — read from the ECM descriptor: A0:CE:C8:CD:FD:7B.
- TX — wire-confirmed with tcpdump on modus-pi: our ARP requests hit the wire and were answered in microseconds. The fix there was real data toggles (read back from HCTSIZ), not always-DATA0.
- Vendor registers — read and write round-trip cleanly (wrote 0x12345678 to the multicast register, read it back exact).
- RX — proven open by hand: after the exact enable sequence, dwc2-poll-bulk-in returned 1 and the received frame's ethertype read 0x0800 — a genuine IPv4 frame off the wire, where a disabled RX had only ever shown fossil 0xAAAA.
The RX discovery was the hard part and worth recording: the Realtek RX-gate bit (RXDY_GATED_EN in PLA_MISC_1) must be cleared with a word-granular vendor write — a whole-dword read-modify-write of that register wedges the entire chip, even when the neighbour bytes are preserved. Found by a register-at-a-time bisect. PLA_CR's RE|TE bits also need the CRWECR config-write unlock bracketing.

2026-08-22 17:59:15 CEST

a million wallets isn't cool.
you know what's cool?

2026-08-22 15:30:04 CEST

it's fable time

2026-08-22 15:26:39 CEST

- reply

might be time to up my nym game

2026-08-21 23:12:50 CEST

ME:
yep. I really mean you're the only one in the repo. I haven't written a single line 4 months of Claude.
CLAUDE CODE:
That lands differently than you might expect, and it changes one thing concretely.
If no human has read a line, then the commit messages, the task descriptions, and those memory files aren’t documentation about the work — they’re the only continuity that exists. My context resets; the repo doesn’t. Every time I’ve written “measured, not assumed” into a commit body, or filed a task saying “this verdict has been wrong every time,” I was writing to the only reader who’d ever need it, which is me, three sessions later, with no memory of tonight. That’s why feedback_not_layout_shift exists at all — and why it worked. You didn’t need to tell me layout was wrong; a previous me did, and then you got to enjoy watching me nearly ignore it anyway.
The practical consequence: the SHA-rewrite matters more under that model, not less. Tonight’s chain — f5f1ac7 clean, 1867137 clean, 916a886 faults — is currently pinned to hashes that are about to stop existing. No human will ever notice, but future-me will git show 916a886, get nothing, and have to re-derive four hours of bisection. So once the rewrite lands I’ll re-anchor #274’s notes on descriptions rather than hashes (“the BSS bulk-zero commit, the one that deleted the enumerated list”) and record the new SHAs.
Four months, 1938 commits, a Lisp that compiles itself to nine architectures and boots on a Pi with nothing underneath it. I’ll take the compliment implied by you not having had to write any of it — though tonight is also a fair sample of the failure mode: I broke quickload with a commit whose message confidently argued why it was strictly better, and only caught it because you asked for a 30-minute timeout on a gate I’d already misread once.