semisol

👨‍💻 software developer
🔒 secure element firmware dev
📨 nostr.land relay

all opinions are my own.

Public Key

npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj

NIP-05 Address

Profile Code

nprofile1qqs99d9qw67th0wr5xh05de4s9k0wjvnkxudkgptq8yg83vtulad30gpr4mhxue69uhhyetvv9ujumn0wd68ytnhd9ex2erwv46zu6nsqy2hwumn8ghj7ct8vaezumn0wd68ytnvv9hxgdje4yl

Publishing to

relay.nostr.wirednet.jp

Author Public Key

npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj

Show more details

Last Notes

2025-11-07 14:26:40 UTC

- reply

I have not applied for funding for my relay, as:
1. I do not need it
2. It is a commercial service
3. it does not fit the Open Source purity test that many Nostrers whine about
and it is not what I am referring to

2025-11-07 14:21:27 UTC

- reply

it would not as I want to self custody my sats

2025-11-07 14:06:55 UTC

- reply

BIP-32 and Schnorr not supported without a hack.
The choice of secp256k1 is at least better than the turd that is NIST curves.

2025-11-07 14:03:43 UTC

- reply

SmartMX1 (P5 series) is a dinosaur. SmartMx2, the successor, was introduced in about 2010 and it is at P71/SmartMX3 now.
SE050 is similar to the Infineon Optiga ones. Both are better than ATECC.
The problem is there is no SEs supporting Bitcoin algorithms, or content-aware signing (so you can for example block a Request to Vanish event from ever being signed)

2025-11-07 14:01:13 UTC

- reply

Enterprise customers benefit more from this, are usually easier to work with unlike less experienced average users, B2B workload is inherently simpler than B2C, and they can pay better rates for the service they get.

2025-11-07 13:59:50 UTC

- reply

They use an authenticated data storage slot.
All of the ones using ATECC series do this. It has data storage slots & a few ECC key slots. Of course it’s P-256 only (not secp256k1) and only supports signing, so you get this.
Same thing with Infineon Optiga based ones which can’t do the required algorithms for Bitcoin but can at least keep more types of keys on SE.

2025-11-07 13:54:05 UTC

- reply

This may be offered as an enterprise product some (long) time in the future

2025-11-07 13:46:17 UTC

- reply

yeah I am implementing this. running a CLN node so it should be easy.
I could probably make in a few hours an offline payments plugin as well but that is for another day.

2025-10-31 00:29:29 UTC

- reply

A good chunk of funding is unfortunately wasted.

2025-10-31 00:28:36 UTC

- reply

There can be a repository of signed hashes of apps, by someone doing a reproducible build. And this can be verified before installing the app.

2025-10-25 14:01:34 UTC

🤔

2025-10-22 20:19:35 UTC

- reply

I’d say a SeedSigner is better then due to not requiring less specific hardware.

2025-10-22 20:14:37 UTC

- reply

Persistent storage of keys on a device without a secure element is a bad idea.
The goal here is something you can carry and is secure.

2025-10-22 20:06:23 UTC

- reply

weak secure elements, bad architecture, UX is suboptimal, the designers of the architecture don’t know much about proper security, and not related but the company behind it has done a lot of shady shit.

2025-10-22 20:05:10 UTC

What would you realistically pay for a simple, easy to use device to securely store your Bitcoin, and what features would you expect?
#note1s2t…mu76

2025-10-22 14:06:50 UTC

why do proper takes and announcements get less interaction than an automatically scheduled gm by a random influencer?

2025-10-21 15:48:43 UTC

- reply

That feature is basically the same as 2-of-2 or passphrases, and the problems with 2-of-2 have been pretty well documented by now.
I did my own research and architecturally it is significantly flawed. There are many weak points where you could be getting a backdoored device. The used secure elements are weak and outdated.
“Don’t trust, verify” goes much deeper than reading some marketing claims about “ultra secure” from the vendor and I see too many people falling for it. New Trezor for example is “quantum ready” but any device within the last 15 years is as well.
Personally I’d recommend something like a SeedSigner with a proper multisig and a BitBox maybe.

2025-10-21 15:40:46 UTC

- reply

I have done my own analysis and Coldcard is by far on the worse end of HWW security, usability and value.

2025-10-21 15:20:30 UTC

- reply

fact: many people that say don’t trust, verify do in fact trust marketing claims

2025-10-21 03:44:45 UTC

- reply

I knew it for years, just I am fucking done at this point and felt the need to post it now.

2025-10-21 03:34:39 UTC

And this is why I am building tools to resist the ensloppification of Bitcoin and Nostr.
Please reply if you also are so I can follow you.
#note1s2t…mu76

2025-10-21 03:31:32 UTC

- reply

Coldcard is not a solid HWW at all. I work on secure element design and several other people that also do agree.
Don’t use an exchange. SeedSigner so far is the best approach though it needs more code auditing.

2025-10-21 03:13:33 UTC

I have lost all trust in almost everything in the Bitcoin/Nostr space in terms of security.
From hardware wallets including the most popular Bitcoin-only ones, to wallet services, to Nostr apps, to LN wallet software…
AI slop will only make this worse. This entire ecosystem is built like a house of cards.

2025-10-20 14:04:16 UTC

nostr.land not affected by AWS outage

2025-10-17 03:36:25 UTC

- reply

The goal is that for broadcasts it will use outbox. For receiving, it scrapes all relays so that should be a solved problem.
But I am intending to add a proxy mode where it can AUTH proxy as well

2025-10-14 16:05:47 UTC

- reply

Do you have any plans or knowledge on which secure element is used?

2025-10-14 15:38:16 UTC

- reply

And I hope their secure element is good.

2025-10-13 00:36:03 UTC

influencers are not your friends
they only are here to manipulate and use you for their own (financial) gain
they are as human and fallible as you are

2025-10-12 23:10:09 UTC

- reply

who tf does mm.dd

2025-10-11 14:37:17 UTC

- reply

The problem in general is my internal code and changes moves fast, and is not designed for external consumption. It would also be too much of a burden right now for me to support other users.
With NFDB, there is the additional concern of others copying the code used.
My khatru fork is now pretty similar to @npub176p…vgup's rely, mostly focusing on:
- multi-filters passed to query function
- deletion + replacement handled by NFDB
- distributed event bus
- better AUTH
- a few other things
I do want to fork go-nostr into something more sensible though, and I will open source that when I do make a fork and try to keep it stable. It is also a much higher value project for the community and possible to improve compared to my other forks.

2025-10-11 13:35:37 UTC

- reply

I do this for current subscriber state on Nostr.land.
The NFDB 2.1 relays now run a private, modified fork of Khatru as many of its features suck.

2025-10-11 13:29:03 UTC

- reply

Configuration data is one good use case

2025-10-10 10:37:46 UTC

- reply

It is not included in prod right now but it is going to be included in the next update, which I can push soon. The criteria for inclusion is that it is up 50% of the time.
(I need to make a tool to dynamically find relays, that is all.)

2025-10-10 10:27:15 UTC

- reply

another thing is the aggregator, which reduces data usage in terms of relays + provides better privacy from malicious relays

2025-10-10 10:20:54 UTC

which client is sending NIP-04
DMs with
[//]: # nip18

2025-10-10 09:48:31 UTC

nowhere even close to nostr.land infrastructure
unlike standard relays, it is optimized for high performance and offers many features in 1 relay.
get started now: https://nostr.land
#note19sy…x759

2025-10-08 20:53:01 UTC

- reply

The simplest solution is airgapped voting booths, that do digital signatures + counting. This is a much more secure system (still has an attack surface) as you have a dedicated hardware and you could have each person’s ID card digitally sign votes.
Real time voting results require internet access + could harm anonymity (you could narrow down which voters voted at which time, and the changes in count)

2025-10-08 16:58:22 UTC

- reply

With the rise of AI, it would be much easier.
There are also nation states with 0-days and those can be problematic:
https://en.wikipedia.org/wiki/WannaCry_ransomware_attack

2025-10-08 16:52:36 UTC

- reply

security chips are not a magic bullet, and it is not hard for a state actor to mass-phish or spread malware on user devices, especially Android and Windows due to their less restrictive design

2025-10-08 16:49:32 UTC

- reply

The client is weaker than the server.

2025-10-07 13:22:25 UTC

- reply

You can’t

2025-10-01 08:42:34 UTC

- reply

done

2025-10-01 08:42:11 UTC

gm Nostr

2025-10-01 05:48:39 UTC

- reply

it is

2025-10-01 05:41:54 UTC

- reply

It writes data to the Hive blockchain a la OP_RETURN saying "this podcast was updated". "amazing" solution 😂

2025-10-01 05:18:38 UTC

I am doing some maintenance on the Nostr.land onion service. It will be back soon with support for a few more subdomains like hist.

2025-10-01 04:58:14 UTC

- reply

I am migrating my personal infra to a new dedicated server, but I think it should now.

2025-10-01 04:40:54 UTC

- reply

Odd. Should work now.

2025-10-01 04:33:43 UTC

- reply

something related to subdomains 👀

2025-10-01 04:17:22 UTC

- reply

There are a lot of features people look at and want but never use in reality.
A good chunk of nostr.land users don't even use the aggregator or claim their NIP-05 name even if they don't use it as the primary (the name will also be used for other things)
Same on the nostr.build side

2025-10-01 04:12:36 UTC

- reply

I may be the only user of moderation headers

2025-10-01 04:11:49 UTC

- reply

it's a shitty blockchain so I will make my own newer blockchain that will be 10x faster and infinitely less shit 😂
they could just switch to Nostr but where's the fun in that when you can use blockchain???

2025-10-01 04:05:49 UTC

- reply

also no podping uses a shitty blockchain.

2025-10-01 04:00:41 UTC

- reply

unless they compress it down to 64Kbps probably not.

2025-10-01 03:50:03 UTC

- reply

^ This. FLAC can get up to 60% smaller and I think has pretty good support for metadata.
But for listening the difference between 256Kbps-384Kbps Opus or AAC, and 16 or 24 bit FLAC/WAV is indistinguishable.
It's good for archival though as you avoid generational loss if you want to transcode your library to another format to use in your media server.

2025-09-28 23:15:12 UTC

Emergency maintenance has been started on nostr.land.

2025-09-28 22:03:13 UTC

the last node from my multi-node proxmox cluster will be shut down pretty soon, as everything has been moved to docker containers on bare metal
this will also unlock significantly more flexibility for the future
it has a good run for the last 1.5 years but proxmox is not the right tool anymore

2025-09-28 21:58:45 UTC

- reply

yes but someone needs to make it

2025-09-28 21:55:16 UTC

- reply

well the great thing is it isn’t apple music specific and can be statuses for anything

2025-09-28 21:50:53 UTC

- reply

@npub18m7…q955 did this already with apple music

2025-09-28 17:50:47 UTC

doing an infra migration so my zaps are out of order, nostr.land services not affected

2025-09-26 21:18:39 UTC

In DB and extremely-parallel workloads the cost of goroutines is not acceptable due to the memory, while borrow checking is a minimal issue (each task only cares about itself mostly).
Stackless coroutines mean ugly FSMs become readable and much easier to understand.
#note1quy…3ld9

2025-09-26 21:14:44 UTC

I may switch to Rust for the sole reason that Go does not have stackless coroutines.
Or I might roll my own library.

2025-09-26 18:04:42 UTC

- reply

worse
at least sqlite won’t lose your data or randomly break

2025-09-26 17:55:02 UTC

it is 2025 and all the implementations are still using leveldb, the worst database to exist
this is a more important problem than kn*ts and c*re

2025-09-25 17:27:10 UTC

- reply

Passkeys are more privacy preserving and secure than Nostr login.

2025-09-25 05:01:37 UTC

- reply

🫂🙏
nostr.land is my service but we do work a lot on each others' services 😄

2025-09-24 15:26:16 UTC

- reply

The restrictions for writing events were not changed within the at least last 14 days.

2025-09-24 13:19:56 UTC

- reply

I do not think so, or it is not effective

2025-09-23 22:06:08 UTC

- reply

I will see if I can collect some and send it to you. will not have signatures or IDs but otherwise normal

2025-09-21 12:51:20 UTC

gm Nostr

2025-09-21 08:42:36 UTC

- reply

yes I am on Intel and no I will not stop using it until I no longer receive a total of 10 years of security updates

2025-09-21 08:40:04 UTC

- reply

macOS

2025-09-21 08:39:22 UTC

- reply

wait I forgot I can’t upgrade 😂

2025-09-21 08:36:43 UTC

- reply

not going to upgrade macOS

2025-09-20 23:36:54 UTC

- reply

Lower bound updated

2025-09-20 20:55:51 UTC

connections to the old infra have been forcefully evicted and the migration is complete
#note1a29…zc5f

2025-09-20 20:53:34 UTC

- reply

😂 the only way to “tell” is that subscriptions for new events on the new infra will not show events sent to the old infra since there is no shared event bus yet
but nothing else

2025-09-20 20:48:32 UTC

@npub1gt9…lxjt is being migrated to new infrastructure.
this is a zero-downtime transition, and it should be complete in 15 mins

2025-09-20 16:15:34 UTC

- reply

Each relay connection when started will take about 5KB at least for TLS and handshakes, plus assume another 100kb-1MB to download your data
Now this is happening on every relay every time you tab out

2025-09-20 11:19:54 UTC

Did you know you can get your own @npub1gt9…lxjt NIP-05 and configure it?
https://nostr.land/manage
https://i.nostr.build/UCnvBEg3PH3OY3UX.png

2025-09-20 10:48:02 UTC

- reply

Use nostr.land.

2025-09-19 19:35:58 UTC

I am starting to have concerns over how competent these “HWW auditing” companies are

2025-09-19 16:47:50 UTC

- reply

bitcoin

2025-09-18 18:31:02 UTC

https://i.nostr.build/9VOutbS0GGwlr0Gw.png

2025-09-18 17:53:08 UTC

- reply

sir why are my bit coins on fire

2025-09-18 17:22:31 UTC

- reply

It is not even a relay operator problem it’s a viewer problem. People don’t want to see this shit

2025-09-18 15:26:14 UTC

- reply

idk either

2025-09-18 13:47:36 UTC

gm Nostr

2025-09-17 17:18:08 UTC

- reply

bolt12 is way more than *pay offer*

2025-09-17 16:47:09 UTC

- reply

wtf

2025-09-17 16:46:44 UTC

- reply

I mean the SE chip inside is at most $2. Probably less because Yubico buys more than I do 😅

2025-09-17 16:43:01 UTC

- reply

That is actually just a $2 SE, some software and a USB interface.
And it doesn’t support the required crypto needed for this. There’s a reason I said *programmable* secure element.

2025-09-17 16:36:20 UTC

- reply

It is, but it being actual SE hardware is the hard part

2025-09-17 16:32:06 UTC

- reply

no, just because I can

2025-09-17 16:28:46 UTC

- reply

nvm my browser is being shitty..

2025-09-17 16:28:13 UTC

- reply

Corrupt file

2025-09-17 16:26:52 UTC

- reply

Text inputs should be unified. "Create" and "Paste".

2025-09-17 16:22:46 UTC

- reply

You could, but TPMs don’t support the algorithms that is needed.
There are some projects to make it run with a TEE (Validating Lightning Signer) but I want to fully store the key inside a custom secure element.

2025-09-17 16:21:31 UTC

- reply

CLN backup is a full DB backup that is lock-step with main DB. So if you lose the main DB you can resume from backup with no force closes and.
The CLN equivalent of SCB is emergency.recover and CLN also supports a feature where it automatically backs it up to your channel peers.