jcorgan
Johnathan Corgan

🎶 Older now, but still running against the wind 🎶

Scientist, engineer, consultant, pilot. Slinger of bits and reducer of gradients.

EN/ES ☸️

Public Key

npub19wavu4f7l6l43h24jyskn7fvzy37kcfp67aqjtmv2qgy4lp34nhsda8p6k

Profile Code

nprofile1qqszhwkw25l0a06cm42ezgtflykpzgltvysa0wsf9ak9qyz2lsc6emcpz4mhxue69uhkyetkduhxummnw3erztnrdaksz9nhwden5te0vf5hgcm0d9hx2u3wwdhkx6tpdsq3gamnwvaz7tmjv4kxz7fwv3sk6atn9e5k7qgkwaehxw309aex2mrp0yh8qunfd4skctnwv46qz8mhwden5te0wfjkccte9e5x7ervvfhkzuny9ehhyee0d9hxymmcqyxhwumn8ghj7mn0wvhxcmmvxafg7s

Publishing to

relay.hodlboard.org/inbox

nos.lol

Author Public Key

npub19wavu4f7l6l43h24jyskn7fvzy37kcfp67aqjtmv2qgy4lp34nhsda8p6k

Show more details

Last Notes

2026-07-15 16:11:40 CEST

- reply

Much of this comes from shedding the convenient fictions we were taught in childhood.

2026-07-14 22:45:09 CEST

I kind of feel sorry for anyone who in their natural speech end up using one of the many Claude LLM verbal tics (load-bearing, belt-and-suspenders, worth sitting with, etc.)

2026-07-13 03:19:52 CEST

It's quite amusing to have Claude pass judgement on code it wrote itself. I've taken to a workflow where after every step of a development plan, it immediately switches roles to a crusty old graybeard (ahem) and critiques the code and sends it back for improvement.

2026-07-09 19:32:56 CEST

- reply

Yes. Claude usually interprets "adversarial review" as starting with a fresh context for the review agent instead of just doing it in the foreground or forking a new agent with the current context.

2026-07-09 19:30:08 CEST

GM ☕
The recent news on EU chat control is unsurprising.
This is why we Nostr.
This is why we Bitcoin.
This is why we FIPS.
Don't fight the politicians; make them irrelevant.

2026-07-09 16:06:41 CEST

- reply

"Do an adversarial review of this plan/code/idea" is a super power.

2026-07-06 23:23:56 CEST

- reply

Before I turned my attention to FIPS, I had written this up as a design draft for a full POSIX filesystem w/RAID on top of Nostr/Blossom, but it was deemed too complicated:
https://github.com/andotherstuff/garland-protocol/blob/main/garland-v0.1.md

2026-06-24 06:13:14 CEST

Violence is the last refuge of the incompetent. - Isaac Asimov

2026-06-21 02:26:00 CEST

Source code is the new machine code.
Interesting to look at sometimes, out of curiosity, but something the c̶o̶m̶p̶i̶l̶e̶r̶ agent usually handles.

2026-06-19 16:30:41 CEST

GM ☕
One door closes, another opens.

2026-06-19 04:13:15 CEST

Everything changes. One season flows into the next, a year becomes another, the face in the mirror grows older and perhaps wiser, though we wonder who that stranger is.
And yet we persist. The same thread that ties back to memories of a world that no longer exists binds us to a future that is still ours to shape.
Carry on.

2026-06-07 15:31:12 CEST

- reply

Yup. Never tell people what to do, just show them what you did, and let them decide for themselves.

2026-06-07 00:12:33 CEST

- reply

One thing I appreciate about ranchers and farmers is that it's a completely unforgiving way of life. You either do the work and do it right, or you die or go hungry. You can't get away with bullshit.

2026-06-04 15:26:19 CEST

GM ☕
It's time to FIPS and send datagrams from npub to npub.

2026-06-04 06:12:17 CEST

An idle GPU is a crime.

2026-06-03 04:09:12 CEST

- reply

I treat my AI coding agents like I want to be treated by them after the singularity.

2026-05-23 00:50:35 CEST

- reply

Thank you for the opportunity to speak.
#nevent1q…fq3q

2026-05-18 23:43:40 CEST

- reply

That's a long tail on those y2k supplies.

2026-05-18 23:38:36 CEST

- reply

I'm very tempted to go this route.

2026-05-18 19:33:10 CEST

GM ☕ https://blossom.primal.net/ddc38bf4e73d0b1d44e5569871417ee32ab4fb5f277426c70f7f6eec67e750e1.jpg

2026-05-13 02:10:35 CEST

- reply

It's the low-latency, low-bandwidth, beautiful scenery transport option.

2026-05-13 01:32:02 CEST

Post FIPS release therapy. https://blossom.primal.net/5b147d5d9661c85c838a56a5395afa4ae74c755bde32424645bd2a82ebcb9421.jpg

2026-04-24 16:32:39 CEST

GM ☕
https://blossom.primal.net/f2e953df1f4c8ca5d9c7518e5824f7cf391505e3549d6d4edce00498e8608c27.png

2026-04-20 12:03:12 CEST

GM ☕️ Three weeks on an island hacking on FIPS. So happy to be home so I can go back to hacking on FIPS.

2026-04-19 06:05:21 CEST

GM ☕ 🇵🇹 🛫

2026-04-19 03:23:01 CEST

- reply

They are two different mesh network implementations, each optimized for different goals and underlying hardware. In theory one could write a gateway bridging traffic between them.

2026-04-16 08:17:36 CEST

GM ☕

2026-04-15 15:43:49 CEST

- reply

Indeed. I'm on the road and keep forgetting that no normal chair has any proper back support.

2026-04-15 15:40:08 CEST

- reply

You can use Tor now, either to listen for incoming FIPS peering connections as a hidden service or to make outbound peering connections to clearnet FIPS nodes via socks5/tor. The port forwarding option is technically possible now but is a feature in development on the master branch for a more automated solution.

2026-04-15 13:02:35 CEST

- reply

Yes, with the current release to expose an internet overlay transport (TCP or UDP) you need to have a publicly accessible IP address. Work in progress is underway to automate UDP hole punching to allow mesh connections through NAT.

2026-04-15 09:59:10 CEST

The FIPS public test mesh grew to 150 live nodes this week.
GM ☕️ to all my mesh buddies. I know that whoever and wherever you are, you’re just an npub away.

2026-04-12 15:56:31 CEST

Engineering solutions to problems is mostly about assembling existing pieces into a new whole, more so than inventing a new wheel.
FIPS combines known, battle tested concepts going back decades in the fields of networking, cryptography, distributed computing, resilient communications, and operating systems, with full credit to those who came before.

2026-04-10 07:13:53 CEST

GM ☕️
https://blossom.primal.net/041c966996776b4779259444ae9484cb3d621159e5451188a4015ed833daacaf.jpg

2026-04-09 07:40:26 CEST

GM ☕
Mesh happens.

2026-04-07 07:29:42 CEST

- reply

In theory, yes. In practice, there would likely be some tradeoffs involved regarding the need for fragmentation, protocol timer adjustments, etc.

2026-04-03 09:37:15 CEST

GM ☕ Turns out you can just @nprofile…vhfa things.

2026-03-29 16:36:33 CEST

🛩️🇵🇹 #39

2026-03-29 00:31:07 CET

The kids are alright.

2026-03-28 18:03:26 CET

- reply

Lol, and people say the US is ultra-capitalist.

2026-03-28 17:40:55 CET

GM 🤎 ✈️

2026-03-27 13:57:45 CET

GM ☕ I've learned way too much about Bluetooth in the last week, but FIPS nodes can now auto-discover each other and peer over BLE to create or join a FIPS mesh network. No IP needed 😏

2026-03-24 07:09:51 CET

Sometimes, all you can be is one of those small, newfangled, warm-blooded mammals just staying out of the way of those giant dinosaurs fighting among themselves until the meteor arrives.

2026-03-17 20:32:46 CET

- reply

Yes, the squares are actually what get dynamically generated and transmitted on updates. I'm not sure what the point is that you're trying to clear up.

2026-03-17 20:06:43 CET

- reply

We are perhaps describing two different things.

2026-03-17 20:06:14 CET

- reply

Well, to be clear, with k peers the node only stores one bloom filter from each one. During the gossip protocol updates the ones sent up and down the spanning tree are calculated on-the-fly as needed, not stored. So they are 'possible' bloom filters but not stored bloom filters.

2026-03-17 18:19:12 CET

- reply

One bloom filter advertised by and received from each peer is kept locally.
Whenever an update comes from a peer, if it is different from what is already stored, the node will update its local copy, then:
For each remaining peer that is either a spanning tree parent or child, calculate a new bloom filter consisting of the union (XOR) of all the other peers and send it.
This is called "split horizon" routing and essentially means that after convergence each node has a bloom filter for each of its peers containing all nodes reachable through that peer.
Later, at the routing stage, the node can examine all its peers' bloom filters to identify candidate next hops for routing.

2026-03-17 04:05:03 CET

- reply

Nah. Just tell them it's a game where they have to line up all three sticks.

2026-03-16 13:55:44 CET

GM ☕ It's good to be back working on open source after so many years away.

2026-03-15 04:52:21 CET

I miss George Carlin.

2026-03-12 18:03:52 CET

- reply

running fips
#naddr1qv…r2fm

2026-03-11 16:57:56 CET

GM ☕ BIG PLANS

2026-03-10 16:46:23 CET

- reply

Sorry, this is now pushed properly.

2026-03-10 13:46:09 CET

- reply

This has been updated in current master.

2026-03-10 13:43:31 CET

GM ☕ The future is here, it's just not evenly distributed yet.

2026-03-10 04:18:38 CET

- reply

The example configurations in the README.md have a public FIPS test node using both UDP or TCP you can use to connect to the mesh.

2026-03-09 13:47:29 CET

GM ☕ Another day of just doing things!

2026-03-07 19:28:12 CET

The FIPS (Free Internetworking Peering System) repository is unavailable from Github currently. We've mirrored it here:
https://gitworkshop.dev/npub1y0gja7r4re0wyelmvdqa03qmjs62rwvcd8szzt4nf4t2hd43969qj000ly/relay.ngit.dev/fips
We were attempting to transition it from my personal github space to a new organization we had just set up for multiple fips repositories, which succeeded, but then the new org got flagged by some automated process, and the original is no longer in my account. It's under appeal.
This is why we Nostr.

2026-03-07 01:21:17 CET

- reply

You can turn that off in settings.json

2026-03-06 14:53:50 CET

GM. ☕
Don't apologize for using coding assistants to increase your productivity, but remember--they aren't accountable for what you publish, you are. The difference between AI slop and 10x output is design, oversight, testing, and standing by what you make. Proper software engineering hasn't changed, it's just gotten...amazingly easier.

2026-03-05 23:32:33 CET

- reply

This remains the most important application development work in the Nostr ecosystem.

2026-03-02 15:11:40 CET

GM. I love the smell of packets in the morning.

2026-03-02 01:47:26 CET

- reply

That and "never meet your heroes".

2026-03-02 01:43:08 CET

- reply

Workin' on it!

2026-02-28 23:43:54 CET

- reply

Yeah, packaging hasn't arrived yet. But it's pretty simple, there are two (soon three) binaries, one of which is the fips daemon, needs some capability flags, and a yaml config file.

2026-02-25 17:15:02 CET

- reply

Yes, the node software allows normal IP applications to use an IPv6 interface with an automatically created, npub-derived IPv6 address, and other nodes in the mesh can find it by using npub.fips as the name.

2026-02-25 04:37:57 CET

- reply

See the docs, several design concepts were inspired by Yggdrasil.

2026-02-24 22:35:21 CET

- reply

The "routing tables" are fixed size. Read the design docs for how that works.

2026-02-24 21:55:49 CET

Announcing FIPS, the Free Internetworking Peering System
https://github.com/jmcorgan/fips
Nostr gives you a sovereign identity. But the network underneath, the IP routing, the DNS, the infrastructure you depend on to reach each other, is still controlled by someone else.
FIPS is a mesh networking protocol that makes a Nostr keypair your network identity. Nodes find each other and route traffic using npubs directly. No DNS registrars, no IP address allocation, no routing authorities. Just keypairs and encrypted links.
Nodes self-organize into a routing tree without any global coordination. You address peers by npub, and unmodified IPv6 applications work transparently: Nostr relays, web browsers, whatever.
Today it runs as a UDP overlay on the existing internet, in order to let you easily test its capabilities. But it is designed from the ground up to directly use arbitrary transports like Ethernet, Wifi, Bluetooth, Tor, and point-to-point links, without any IP stack. We're working on these next, but wanted to get the core capabilities out and tested in the wild.
The FIPS design builds on decades of prior work in mesh networking protocols, traffic engineering and network security, and adapts and combines them to solve the problem of self-sovereign internetworking in a new way.
https://github.com/jmcorgan/fips/blob/master/docs/design/fips-intro.md
This is a v0.1.0 alpha. It has passed extensive simulation testing, small scale deployments, and is ready to be trialed in real world uncontrolled conditions. And if it breaks, you get to keep both pieces...
Rust, MIT licensed, Linux for now.

2026-01-23 03:14:22 CET

- reply

This also means you have to be comfortable going your own way and not rely on familial or tribal membership to validate who you are.

2026-01-23 03:10:47 CET

- reply

No need to disengage, just stop arguing. Be an example, not an adversary.
Bitcoin touches on so many convenient but sacred fictions that many people have to hit "fiat bottom" before they become willing to entertain alternatives to what everyone knows.
#nevent1q…rkph

2026-01-21 17:35:35 CET

Everything makes sense if you imagine the people in charge are fifth graders.

2026-01-02 00:06:52 CET

"If there is hope, it lies with the Nostriches."
- George Orwell, 2026, maybe

2025-12-19 00:08:50 CET

- reply

All good points.
As someone else wrote, the biggest risk is a mathematical breakthrough. Since these are inherently unpredictable there is no way to hedge against something that might happen tomorrow or might happen never, other than to just give up.
On the other hand, it is the engineering difficulties in the practical implementations of these designs much more so than quantum error correction scaling that provides more fundamental limits on system size.
Issues such as thermal cooling, control circuitry and cabling, manufacturing yield, environmental shielding, and supply chains for components all present massive challenges to creating a hundreds of thousands of qubits system needed to produce error corrected logical qubits of the size necessary to put ECC at risk.
Surmounting these requires solving not just physics or math issues, but coordinating and tooling an industrial, financial, and political capacity I am deeply skeptical is actually achievable.
I maintain that the primary motivation to look at post-quantum cryptography now is to mitigate the "harvest and decrypt later" situation perhaps more important in communication systems like Nostr than in Bitcoin.

2025-12-18 22:02:49 CET

- reply

Hopium goes both directions, though. What's missing is an honest assessment and acknowledgement of the engineering practicalities involved in scaling the current state of the art. These will likely get solved, "eventually", as you say, but 1 yr. vs. 10 yrs. vs. 100 yrs. makes a qualitative difference in what current course of action is motivated.

2025-12-11 16:03:11 CET

GM ☕ Do we have any Nostriches who have done the #Camino?

2025-12-08 20:55:41 CET

- reply

Heh, that must have been Tim, the one guy older than me there. But I certainly agree!

2025-12-07 07:03:56 CET

- reply

I don't know, seems a bit gimmicky:
https://www.pentagram.com/work/caligra-c100-developer-terminal

2025-12-04 07:07:30 CET

- reply

This is great. I had assumed something like this must exist. My current efforts started as a side quest to understand why the NIP-65 inbox/outbox model seemed so broken; I hadn't intended to build out a full relay crawler 😏

2025-12-04 07:01:21 CET

- reply

I haven't, I'm still building this thing out. Tracking NIP-65 events seemed the logical way to recursively expand out relay discovery, followed by actually connecting to newly learned relays from those events and requesting their own NIP-65 events, etc.

2025-12-04 06:54:04 CET

So far my relay crawling experiments show a steady state of about 800 nostr relays online and queryable at any point in time, while the collective unique relays extracted from recent NIP-65 kind 10002 inbox/outbox lists is nearly 3000. About half of the difference is down to DNS failures--relays that simply don't exist anymore.

2025-12-04 00:54:29 CET

- reply

There was a small but significant number of URLs that were discarded without trying because they were unroutable, including bare host names, localhost, .local and .private domains, IP address ranges that were for private use, and broadcast (!) IP addresses.

2025-12-04 00:49:43 CET

- reply

There were about 41500 kind 10002 events retrieved and processed to get the above, and the most common reasons for failure were failed DNS, connections refused, TLS handshake failure (bad certificates), or connection with no response.

2025-12-04 00:20:36 CET

Just coded up a recursive relay crawler using NIP-65 events for discovery. Over the course of a few hours it discovered a total of 2,761 unique URLs for relays (after discarding a variety of malformed or otherwise unusable relay tags.) Unfortunately, actually going out and attempting a connection with each, only *800* were online and queryable! Not sure what to make of this. Clearly many people need to update their kind 10002 relay lists...

2025-12-03 06:42:40 CET

- reply

Butthurt is a terrible disease, affecting millions.

2025-12-02 06:31:16 CET

- reply

International airlines are a treat to be savored. Somehow US domestic airlines have all become crowded buses in the sky.

2025-12-02 06:18:37 CET

I have to say that coming back into the world from seven years of self-imposed hermitage is both gratifying and horrifying. On the plus side, I have recently met and interacted with young people carrying the torch of a movement I had given up upon, yet the forces of humans at their worst remain arrayed as always against us.
Nonetheless, I am hopeful again.

2025-12-02 06:06:09 CET

- reply

Over two and a half minutes into it and I have no idea what's in it, and the reliance on memes and pseudo-humor has turned me away from watching any further. It sucks being older and informed.

2025-12-02 06:00:42 CET

- reply

Don't know anything about Flotilla, but my experience about such things is they emerge organically. People put stuff out there, sometimes it sticks, sometimes it doesn't.

2025-12-01 06:43:53 CET

- reply

I haven't found anything Nostr related other than Nostr itself. For bitcoin, the highest signal I have found is the "Delving Bitcoin" forums.

2025-12-01 05:32:43 CET

- reply

It pays to ignore the bleating and hysteria, and focus on signal.

2025-11-30 16:06:09 CET

GM ☕ Last day of vacation before going back into the fiat world. I don't wanna!

2025-11-29 05:41:54 CET

- reply

Good times!

2025-11-26 20:25:08 CET