semisol

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

all opinions are my own.

Public Key

npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj

NIP-05 Address

Profile Code

nprofile1qqs99d9qw67th0wr5xh05de4s9k0wjvnkxudkgptq8yg83vtulad30gpp4mhxue69uhkummn9ekx7mqpz4mhxue69uhkzem8wghxummnw3ezumrpdejqzyrhwden5te0dehhxarj9ekxzmnyqy28wumn8ghj7un9d3shjtnyv9kh2uewd9hsj0qyls

Publishing to

Author Public Key

npub12262qa4uhw7u8gdwlgmntqtv7aye8vdcmvszkqwgs0zchel6mz7s6cgrkj

Show more details

Last Notes

2026-05-26 03:41:25 CEST

- reply

it is good yes
however I still find that the deep research module produces a different but good perspective
+ both of them suck on a lot of things but I push limite more than the average usecase of “I want to be a successful startup make me a startup make no mistakes don’t tell me it is hard”

2026-05-26 03:39:37 CEST

- reply

please do not post the ai slop without oversight 🥲

2026-05-26 00:36:38 CEST

- reply

Oops, the AI companies already bought out another 500 weeks and the automotive ones another 500
Would you still like to wait 1999 weeks?

2026-05-26 00:32:03 CEST

- reply

Sorry, functional features are out of stock, with a lead time of 999 weeks. Would you like to add yourself to the waiting list?

2026-05-26 00:02:14 CEST

- reply

In cryptography this horse has been beaten to death.
You don’t experiment. You make it secure, even at the cost of inefficiency, for the sole purpose of preventing the slightest chance of anything happening.
Please read these before you eventually shoot yourself in the foot:
https://soatok.blog/2026/02/25/cryptography-engineering-has-an-intrinsic-duty-of-care/
https://www.schneier.com/essays/archives/1997/01/why_cryptography_is.html

2026-05-25 23:58:20 CEST

- reply

Is a massive footgun worth it just to not publish 1 event or add it to an existing one

2026-05-25 23:52:20 CEST

- reply

https://i.nostr.build/tO9qCFUILMf2j7AI.jpg

2026-05-25 23:50:46 CEST

- reply

I have never liked the approach of “fixing” security issues by just marking them out of scope.
Cryptography needs to be resistant to footguns. You should NOT be able to misuse without going out of your way.
This trivially allows that with a simple oversight

2026-05-25 23:45:16 CEST

- reply

Any integer x over GF(p) have a modular inverse x^-1 so that x * x^-1 = 1.
So the view private key v = H(S || “magic number”) * s (where s is your nsec, and S = sG = npub) can be reversed, by testing each possible S’, and if it holds, then it is the view key for S’:
S’ = H(S’ || “magic number)^-1 * v * G = H^-1 * H * s * G = sG = S
Then you can trivially reverse the key
s = H(S || “magic number”)^-1 * v

2026-05-25 23:17:45 CEST

- reply

The issue is that there is no way to derive a distinct key from an npub.
All tweaks are reversible by the server and so they can derive your nsec easily, so the electrum server would know your nsec (and that is why I compared it to Anigma). A list of candidate npubs’ tweaks can be subtracted and checked.
Even asking an LLM would have found this issue. Cryptography is hard to get right and so unless you have a good reason, don’t roll your own schemes.
If you do, do some research.
Just make a new keypair.

2026-05-25 22:54:18 CEST

This is a good moment to sign up to nostr.land. We will not delete your notes.
And you can get back your notes with a click even if they were nuked, and download local backups. 😉
https://nostr.land/subscribe
#nevent1q…jtzl

2026-05-25 22:43:17 CEST

- reply

works

2026-05-25 22:01:14 CEST

- reply

Click change receiving address in profile

2026-05-25 21:47:03 CEST

@npub1lxz…svxa something about your rx is wedged please change it

2026-05-25 14:56:42 CEST

- reply

Websites actually work, at least

2026-05-25 14:52:32 CEST

- reply

I will be implementing this but I am also concerned that it will end up like those SMS UIs

2026-05-25 02:49:42 CEST

- reply

codex will consume all

2026-05-24 23:22:34 CEST

- reply

Silent payments via an external wallet work fine. Use them!

2026-05-24 23:21:35 CEST

- reply

It’s not. Unlike a separately managed wallet, this is the same nonsense as onchain.

2026-05-24 23:15:39 CEST

- reply

This is a horrible idea. If it happens, it will cause a disaster worse than Anigma.
Use Payment Targets.

2026-05-24 22:55:04 CEST

- reply

Both of them? 😄

2026-05-24 16:33:01 CEST

Looking for some interesting accounts to follow, any suggestions?

2026-05-24 16:22:42 CEST

Pubky is owned by Tether, explains a lot

2026-05-24 16:16:21 CEST

Wtf?!
#nevent1q…ymxj

2026-05-24 14:45:00 CEST

gm nostriches
dogma kills innovation.

2026-05-24 14:32:14 CEST

- reply

For a thousand bitcoins they better ship a plug-and-play preconfigured rack of 5 servers with the latest hardware + 3 AI servers, that runs my own reliable Pubky server along with the source code for everything. *everything*.

2026-05-24 14:18:13 CEST

I cannot fathom the idea of speed-limiting paid users to 1MBps.
Nostr.land will never have such a thing.
#nevent1q…cs8x

2026-05-24 11:56:01 CEST

- reply

That is a slop LLM bot

2026-05-24 04:15:03 CEST

- reply

1. You ask
2. Download file
3. It opens in browser with infinite resolution

2026-05-24 04:13:35 CEST

- reply

Ask it to make it as an SVG

2026-05-24 03:33:36 CEST

- reply

✉️

2026-05-24 03:06:02 CEST

- reply

13768 65974 93907 99623
86615 57013 89975 13389
54160 85275 58478 48261
87561 56514 03777 26

2026-05-24 02:56:13 CEST

- reply

aagh ok
https://smp19.simplex.im/a#Ar3HnujM-Q3oh2Zmvip2CjzS4d24surQ7hgcTnzHT8Y

2026-05-24 02:54:39 CEST

- reply

signal?
also, I do not use giftwrap DMs

2026-05-23 22:27:54 CEST

Nice to see that it is in fact being used
#nevent1q…zcs5

2026-05-23 22:21:50 CEST

- reply

🫡

2026-05-23 20:33:42 CEST

- reply

Relays don’t need much bandwidth

2026-05-23 20:10:40 CEST

- reply

Doesn’t matter if they do. Outbox allows relay names to change, .onions exist, and we could use “pubkey URLs” where relays can publish a latest location document that is signed.

2026-05-23 20:06:44 CEST

- reply

Don’t bother they already have me muted for idk what.

2026-05-23 20:06:29 CEST

- reply

And yet you can still access them via domains outside US jurisdiction, proving what I said.

2026-05-23 19:50:53 CEST

- reply

These sit outside ICANN jurisdiction.
And the root zone file is distributed *everywhere* and the servers running the root nameservers are ran by many independent parties.
I think that the idea of a coordinated attack against Nostr via DNS is absurd. Instead of worrying about such nonsense (which can already be circumvented via Tor etc.) we should work on other solutions.

2026-05-23 19:48:27 CEST

- reply

I should remind that countries also control their own TLDs

2026-05-23 19:45:36 CEST

John Carvalho has always wanted to corporatize a Bitcoin-adjacent social/apps layer for their own benefit and for profit.
There was Synonym when LN got popular, and now there is Pubky now that Nostr got popular.

2026-05-23 17:54:58 CEST

- reply

a good choice as well, everything on top runs debian anyway
I like the additional VM isolation, though I use containers for the nodes to benefit from ZFS compression.

2026-05-23 17:47:09 CEST

Is there anyone using Payment Targets?

2026-05-23 17:43:56 CEST

- reply

In general, that sector is obsessed with their
✅✅⬜️✅⬜️⬜️✅✅✅✅⬜️✅⬜️⬜️⬜️

2026-05-23 17:42:01 CEST

gm nostriches
I migrated my node to Proxmox 2 days ago.
I recommend you do the same, and throw out Start9 or Umbrel or whatever you are using

2026-05-23 17:40:08 CEST

- reply

And even then, automotive security people are weird

2026-05-23 17:39:07 CEST

- reply

I am still trying to find one time where I preferred using anything else when Lightning was available.

2026-05-23 17:33:08 CEST

- reply

I don’t think there is a reason to be happy regardless of the temperature if you are forced to use Fahrenheit.

2026-05-22 14:22:03 CEST

- reply

as of this moment I retroactively invalidate any and all recommendations I might have made for this thing

2026-05-21 22:57:57 CEST

Onchain zaps are not new, and the idea of using npubs for bitcoin addresses has existed for a while.
There is a reason no one did it.
https://github.com/nostr-protocol/nips/pull/143
https://github.com/nostr-protocol/nips/issues/1851
https://github.com/nostr-protocol/nips/pull/258
and probably many more that discussed on Nostr and never made it to a NIP

2026-05-21 22:10:17 CEST

https://i.nostr.build/kRGp4uC4v5uzR3Lr.gif

2026-05-21 21:56:51 CEST

- reply

I think our points were the same, just bad communication :)
It would be best if the LN address service is integrated into the mint.
This is simple and can extend the existing LN receive path, plus maybe zap broadcasting.
The npub passed as the localpart is decoded and used as a P2PK.
From a liability perspective, it also does not increase the risks compared to what mints already tolerate.
Then there would be an “inbox” endpoint to check for any unspent tokens, which the app can then sweep up.

2026-05-21 20:49:31 CEST

- reply

So, it’s fully automated, but it retains all of the work already done on LN and avoids incompatibility and mint management issues.

2026-05-21 20:48:34 CEST

- reply

They need to set up npub.cash for receive, which *can be fully automated by the client including claiming new tokens that are received*. That’s the thing.
This would be a server that anyone can host just like mints work right now (and can be integrated into them)
For send, you need no setup either way, as you can just use a pile of Cashu tokens out of the box with some good way of picking mints.
So, to the user, it is still just “get started” and done, but it still retains compatibility

2026-05-21 20:45:40 CEST

- reply

ga 🍵

2026-05-21 20:11:23 CEST

- reply

there exists many services, like npub.cash and similar, that would bridge different mints and ecash. mints could even provide this as an out-of-the-box service.

2026-05-21 20:06:09 CEST

- reply

Yes! It is basically an xpub that is safe to share and send-only

2026-05-21 19:46:56 CEST

- reply

nut*zaps* and wallet is not the same thing

2026-05-21 19:46:16 CEST

ga nostriches
hmm

2026-05-21 19:00:41 CEST

- reply

you could encrypt it and store it, it’s just a pile of tokens really.
and it does not require nutzaps

2026-05-21 18:38:11 CEST

- reply

Yes

2026-05-21 18:15:26 CEST

- reply

It’s basically a way to make an address for a recipient without interaction

2026-05-21 17:46:00 CEST

- reply

The sender knows the address to derive it…
The point of silent payments is that it’s like an xpub that you can actually publish yourself.
Someone with your xpub can see all your txs, but someone with your SP address can only make new addresses

2026-05-21 17:44:53 CEST

- reply

Cashu can be your wallet, that’s fine.
We don’t need Cashu itself leaking onto zaps though with nutzaps, as it provides 0 benefits, and locks us into a low-adoption protocol.

2026-05-21 15:22:37 CEST

- reply

Zaps don’t care what your wallet is

2026-05-21 07:29:20 CEST

- reply

Please try again

2026-05-20 23:35:11 CEST

- reply

Try again

2026-05-20 23:35:08 CEST

- reply

Try again

2026-05-20 23:10:02 CEST

- reply

Anon zap?

2026-05-20 22:46:46 CEST

- reply

back then nwc also was still a newish thing

2026-05-20 22:43:28 CEST

- reply

what lol
this can be done in ln as well

2026-05-20 22:33:30 CEST

meanwhile I have a pi extension that blocks rm
GPT 5.5 refuses to even consider bypassing the rm block once the runtime says no
#nevent1q…p0w3

2026-05-20 22:30:35 CEST

- reply

anything but nodejs 😂

2026-05-20 22:27:18 CEST

- reply

Huh. There was a 30m maintenance period yes due to a change in the code.
I thought I added some things to handle errors?

2026-05-20 22:02:49 CEST

Testing moved LN node, please send a test zap

2026-05-20 16:39:27 CEST

- reply

gm
idk

2026-05-20 16:39:19 CEST

- reply

gm!

2026-05-20 15:45:54 CEST

gm nostriches 👀

2026-05-20 15:17:48 CEST

- reply

The point of this is not private entropy but public entropy, for example if you are doing a giveaway and want to make it verifiable.
This uses threshold BLS signatures, where the signed data is the timestamp (can be chained as well), and the signature is hashed to get entropy. The initial keys were also generated with multiple parties.
You can't really bias the output except at setup, if you knew *the future usecase you want to bias*, and bruteforced a key that would return a preferred value at that exact timestamp.

2026-05-20 07:59:07 CEST

- reply

What a waste of 7K

2026-05-20 00:27:59 CEST

- reply

Clients are basically lying to users, saying:
“You can use our proprietary method for sending “zaps”, to send money to this person!”

2026-05-19 23:49:58 CEST

Block withholding attacks could be used by a malicious “hashrate renting” provider to double-dip on profits (both on the hashrate, and by decreasing difficulty) while sabotaging activation

2026-05-19 23:45:31 CEST

This has already been invented (Drand). Unlike this project, it also does not say “even we can’t decrypt timelocked files”
(they can)
It’s also a multiparty protocol so it is available even if one party fails, or is malicious.
There is a production instance backed by many different companies and organizations too
https://drand.love
#nevent1q…mxx9

2026-05-19 21:49:37 CEST

- reply

Have you tried financing it instead? 12 mo

2026-05-19 19:50:22 CEST

Many so called “bitcoiners” here have a pavlovian response to being told that $NOT_BITCOIN does $THING better than Bitcoin.
Influencer slop, as usual

2026-05-19 16:00:15 CEST

We don’t need nutzaps either.
Cashu can already use regular LN zaps, and unlike nutzaps, does not have the interoperability or mint mismatch problem.

2026-05-19 13:41:26 CEST

gm nostriches
Translations for Damus Purple subscribers have been updated and should be better in a bunch of contexts

2026-05-19 03:37:11 CEST

L1TF CPU bug present and SMT on, data leak possible. See CVE-2018-3646 and https://www.kernel.org/doc/html/latest/admin-guide/hw-vuln/l1tf.html for details.

2026-05-19 00:17:48 CEST

Occasional reminder to add wss://hist.nostr.land to your relay list, to back up your follows list for free.
Unlike app support, it works across all apps, and is way less likely to fail.

2026-05-18 21:39:21 CEST

- reply

Well you should not be emulating a CPU arch in general

2026-05-18 21:37:56 CEST

- reply

Lockdown mode only moves WA to interpreted instead of JIT which is still orders of magnitude faster than simulating RISC-V, unless you have a trash implementation of “lockdown”

2026-05-18 21:37:12 CEST

- reply

“clear advantages when it comes to licensing and being a significantly more open platform”
What? Both of them are open specifications, that are openly licensed, and I’d say WA is even lower risk than RISC-V in any case

2026-05-18 21:11:57 CEST

- reply

What is the point of RISC-V? You are taking a massive performance hit as it is not a bytecode format designed for being portable but a HW instruction architecture. There are intentional tradeoffs such as the mapping of bits as a direct result of this, which is extremely expensive in SW, but is just wire swapping in HW.
I would recommend you use WebAssembly instead.

2026-05-18 19:14:45 CEST

- reply

I have lost confidence in all but 1-2 SE vendors, after seeing a lot of stuff.
I may be biased, but even recent “open source” secure elements are bad (and not really open source)

2026-05-18 19:10:32 CEST

- reply

I guess. The good thing though is that after so long, most of the crypto for privacy is stable, so it could be implemented directly.
also, I would be upset if someone made a client that allowed anyone to send any fund via any methods:
The recipient thinks that they are sending it to me, because the client is lying to them that I support this method, while I might not even know it exists.
reminds me of those annoying twitter tipbots

2026-05-18 19:04:30 CEST

- reply

yes. many signers have the problem that the SE is utter trash though including trezor

2026-05-18 19:03:47 CEST

- reply

I’d say it lies in both. If people are so concerned, take the split Zcash approach.
I have other concerns with “onchain zaps”, which is that it’s a stupid spec on so many levels:
- can’t specify a multisig / non-npub based address
- implies people that don’t have onchain zaps in their client / don’t prefer them can “receive” them

2026-05-18 18:54:53 CEST

“We made something awful, and we don’t want to stand behind it. We knew it was a bad idea. Can you please stop criticizing us for the mess we have created. Don’t you have some respect? Also no we’ll continue doing it”
— onchain zappers