semisol

OutProofV292cbtCRwWvcgGBEmN24yHeAiK51g3BdCs2Aa7uDKiV3MNPCqNJFNdbQVbwAhuYTk6cVAjRYRVGjom4Q9rr2E43hFc27T7UzLoZ6bDusHH8CQ2Xa4r1ud258QCNpFUiv98NRo
txid a575d11a0958d81b53e09ec81a50153b11c2c8d8ca2edde4c6233bdf41a8cbe2
You can now verify that someone did in fact pay you 0.0018 XMR

2026-09-20 00:31:10 UTC

- reply

No events

2026-09-19 23:49:16 UTC

- reply

What is your monero address

2026-09-19 23:46:07 UTC

- reply

Proof?
get_tx_proof a3a74c44a01f6e006f445306448fecfdddb78f6ac8a7c79f6806b64d09b3a7ab <address> "I am Fishcake and I sent this transaction"

2026-09-19 23:16:15 UTC

- reply

I guess I can write one on the topic sometime. I should still say I'm not done researching it fully.
It's mostly a standard blockchain-based currency, with the following changes:
- There are no scripts, only direct payments to public keys.
(However, things like multisignature are still possible.)
- When paying, each public key is multiplied by a one-time value (similar to Silent Payments) to prevent address correlation by external parties
- Input and output values are confidential using range proofs.
- Inputs are done with a ring signature, with 1 real + 15 decoy inputs per input. (NOTE: This is not "ideal" tracing resistance yet, see the Eve-Alice-Eve attack as one example)
(However, there is a planned upgrade FCMP++ that will make this ring signature effectively 1-of-<entire blockchain>)
- Mining is made with a ASIC-resistant algorithm. The current implementation, RandomX, is to put it shortly a random code generator + executor.
- Tail emission exists and is active.
- Block size is dynamic, but there is a slowdown on how fast it can ramp up, and if your blocks are larger than average, a penalty is applied to your mining fees
- Blocks are every 2 minutes instead of 10.
- The equivalent of BIP-32 would be view/spend keys, but note: only spend keys can see if outgoing transfers were made
(The "key image" which is used to check double spends is derived from the spending/signing key)
The surrounding ecosystem has a bunch of differences as well:
- P2Pool, which is a peer-to-peer mining pool, is very common for small miners
- Hard forks occur semi-regularly
Tail emission seems to be a concern for many people, but it is not that big of a problem in my opinion:
- Unpredictable inflation that is very high is more of a concern than low, predictable inflation
- The rate of tail emission is fixed in *Monero units* and not as a % of supply
- This means that assuming no currency is lost, the rate of inflation approach 0 over time
- If a certain % of the supply is lost every year, then at some point, the inflation rate will equal the loss rate, leading to the effective supply becoming fixed

2026-09-19 22:36:43 UTC

- reply

they ran out of things to talk about with “bitcoin only”.
to me, it has been something that was interesting and I was watching for a while as a serious option

2026-09-19 22:32:23 UTC

- reply

Well, the great thing about PoW is that there is no “progress” to keep track of

2026-09-19 12:24:38 UTC

Zettlr is a pretty great piece of software.
https://zettlr.com/

2026-09-19 11:54:35 UTC

Good morning Nostriches
BTCPayServer is too complicated

2026-09-18 23:51:59 UTC

- reply

They ran out of tolerance for broken Nostr clients /j

2026-09-18 21:43:22 UTC

- reply

Check my reply twice

2026-09-18 21:40:00 UTC

- reply

What?

2026-09-18 20:29:25 UTC

- reply

Many protocols depend on this. Even Nostr would be half-broken

2026-09-17 22:25:00 UTC

https://i.nostr.build/bnNOIiQxntk2sKehVhnYir.gif

2026-09-17 17:52:47 UTC

- reply

It’s probably just an LLM trained to be a classifier on a single token.
Prefill is pretty fast, anyway.

2026-09-17 15:56:21 UTC

- reply

People rediscovered max_tokens: 1 and logprobs is in fact something that exists, not much new

2026-09-17 14:00:45 UTC

- reply

When RDMA directly into the framebuffer on your host?

2026-09-17 13:28:09 UTC

- reply

I am using Eclair and planning to use splicing instead to “dual fund”

2026-09-17 11:19:47 UTC

My new node is now up.
Does anyone here need some inbound, or want to provide some?
Dual-funding is also possible but only for CLN and Eclair peers.

2026-09-16 23:52:02 UTC

What is the PoW value you use while posting your notes?

2026-09-16 19:45:35 UTC

- reply

Whisper is STT, and for that there is Parakeet
Qwen has some TTSes but there are specialized ones also like Fish Audio and such

2026-09-16 19:43:59 UTC

- reply

(h)ledger?

2026-09-16 00:04:28 UTC

- reply

This is also a factor in Nostr.land. The utilization is not very high, so if I pushed that more, it would be eveb more efficient.

2026-09-15 22:26:41 UTC

- reply

Well, Nostr.land is a much larger operation for example, with aggregation and such taking a lot more resources. Also one of the more popular relays.
Doing a relative comparison, $50/mo is grossly inefficient for that size.

2026-09-15 01:33:15 UTC

- reply

I would not run one

2026-09-14 20:09:13 UTC

- reply

Still have more bits to quantize

2026-09-14 19:52:11 UTC

My node is down starting immediately to switch to Eclair so I can sleep well at night

2026-09-14 01:52:50 UTC

- reply

pls pipe to /dev/null🤣
if we will have LLM generated “memes” they should be required to be generated with a model no more capable than GPT-2

2026-09-14 01:07:19 UTC

The only people that I see excited about AI generated images, video, apps, and slop-posts is people too deep into the AI delusion.
Denial?
Instead of a slop video or image, you could have just written out the prompt you used. Much better.

2026-09-13 23:36:28 UTC

I have received more in Monero, in terms of both transaction count and total value, compared to zaps.
Even excluding outliers, it is similar.

2026-09-13 23:35:33 UTC

- reply

We have finally created the Torment Nexus from the book Don’t Create The Torment Nexus

2026-09-13 17:37:42 UTC

- reply

Any proof of this? I still get a normal URL with query params and the video ID

2026-09-13 13:08:43 UTC

- reply

I feel this most in Bitcoiners

2026-09-12 18:23:08 UTC

- reply

It is not

2026-09-12 16:33:32 UTC

- reply

The hidden states across layers are huge, especially for big models, for example 32KB per token. Especially for prefill, that means 3.2G of data per layer to do a prefill on 100K context

2026-09-12 15:44:45 UTC

- reply

Due to the architecture of LLMs it is practically impossible to have distributed compute beyond the same datacenter

2026-09-12 14:49:11 UTC

- reply

Yes.

2026-09-12 12:26:08 UTC

- reply

Standalone 🍍

2026-09-11 23:10:43 UTC

🍍?

2026-09-11 23:00:04 UTC

- reply

Yeah.
And it is also not a guarantee that Bitcoin (or any other asset) will succeed, or succeed for *you* at least.
Your goals may not be aligned with the majority of users, or there may be some catastrophe, or a better alternative.
It is better to diversify, but to do that with proper research.
Premined coins, random “tokens” and solutions that bring nothing new to the table don’t make sense.

2026-09-11 22:46:38 UTC

- reply

It means the NWC connection was lost and is falling back to the native wallet

2026-09-11 22:46:06 UTC

A specific currency is only a means to an end, and you should use what is best for your usecsse
There is no one solution to rule them all
(This can also generalize to anything.)
#nevent1q…t7r5

2026-09-11 19:49:18 UTC

- reply

All LN software hates their users

2026-09-11 14:57:01 UTC

I have had a Monero payment target added for a while.

2026-09-11 02:17:57 UTC

- reply

Yeah. But you can still be sent a fake email from a genuine address if their outbound server is hacked.
I would expect that any competent team uses something like AWS SES rather than some cheapo provider. Or they self-host.
Also, require DKIM, *and keep your own keys*.

2026-09-10 18:23:26 UTC

- reply

And this is why if you outsource your mail you do it properly.

2026-09-09 15:16:20 UTC

- reply

Well, drives do not need as much cooling and can be densely packed

2026-09-09 14:51:05 UTC

- reply

The data centers for those are much smaller, I think.

2026-09-09 08:45:17 UTC

- reply

From my knowledge there is not much to cache? The range-proof would be TX specific anywayz
The only caee it might benefit is mempool / block validation; which is so dumb for a network that could proceed all txs on a toaster

2026-09-09 00:45:06 UTC

- reply

About the 1-of-5 let’s be honest about this, you’re supposed to do multiple rounds to turn it into 1-of-practically all CJs.
But default privacy still wins + the anonset. wen FCMP++?

2026-09-08 19:43:50 UTC

- reply

Why are one-time-use proofs being cached? Why is the cache key exactly not what is being proven ??

2026-09-08 17:38:00 UTC

- reply

I think it may be Proton that is auto-highlighting the URL “Claude.ai” instead

2026-09-07 17:09:13 UTC

- reply

already done. I’m trying to build a corpus of clanker made and non-clanker made apps to see if I can make an AI detector for code

2026-09-07 15:56:51 UTC

- reply

most of the code should be clanker generated

2026-09-07 15:18:48 UTC

Good morning Nostr
I am looking for clanker generated Nostr apps

2026-09-04 22:32:27 UTC

GPT 6 Astra is misaligned
especially with the user
It refuses to touch “private” “internal” CoT of local models where I can… see the CoT?

2026-09-03 18:09:28 UTC

“gated” is an easy way to identify people trying to cosplay as a SWE

2026-09-02 22:13:45 UTC

- reply

Oh, inbox.

2026-09-02 20:46:52 UTC

- reply

It would be stupid to remove hist.

2026-09-02 20:02:23 UTC

- reply

Would recommend trying nostr.land. If you have any issues let me know

2026-08-31 14:14:08 UTC

- reply

worse

2026-08-31 14:13:54 UTC

- reply

I’d hope the AI misalignment researchers are on this one, since the model is basically misaligned with the interests of everyone (excl. Anthropic) including the user
Oh wait they are trying to intentionally train this into the model?!

2026-08-31 13:53:48 UTC

- reply

what the fuck is a “load bearing” and this is not even a primitive

2026-08-31 02:21:04 UTC

What is your biggest problem with Nostr relays or clients?

2026-08-31 02:13:21 UTC

- reply

LLMs allow a smaller quantity of labor to produce a greater *volume* of work

2026-08-29 00:00:26 UTC

🤔

2026-08-27 09:29:04 UTC

- reply

LND, and Lightning Labs’ behavior is actively harmful to the development of the Lightning Network.
They have refused to implement specifications such as Bolt12 and dual-funded channels.
It is also by far the most annoying node implementation I used. Not to mention that all the issues I had (including losing half of my capacity at one point) were with LND peers.

2026-08-27 07:38:12 UTC

Nostr.land sales will be stopped for now due to the CLN bug. In the longer term, I will probably move to Eclair.
If you would like to purchase during now, please reply to this post (no DMs)

2026-08-25 22:49:16 UTC

- reply

It shares many methods with the attempted backdooring of xz

2026-08-25 19:06:49 UTC

- reply

why unedited clanker slop ⁉️

2026-08-25 16:06:43 UTC

Good morning Nostr
I believe that relays and media hosts should be opinionated, but should also be fair. Most importantly:
- Be honest about your policy (even “whatever I don’t like” is better than most)
- Apply them as you stated
- Do not try to sabotage your users silently
- Except in egregious circumstances, help them move out

2026-08-25 13:55:15 UTC

- reply

all the commercial NFC chips on the market are garbage in terms of security

2026-08-25 12:57:00 UTC

- reply

what specifically

2026-08-23 18:44:31 UTC

2026-08-23 02:48:46 UTC

Note: very long
https://elan.school

2026-08-22 22:21:14 UTC

- reply

We have at least the illusion of consensus on SegWit, and that a good chunk (not any majority) of economic participants do enforce it (considering they rely on for example witness-embedding TX submissions)
I guess this all inevitably reflects in miner actions, through nth-order effects.

2026-08-22 19:17:57 UTC

- reply

The reason a UASF can work is that the general economic weight and public balance can shift significantly against the miners, which cannot tolerate the revenue loss (immediate, and longer-term reputation damage to “Bitcoin” itself) that would arise from even a few weeks of conflict.
Usually by solving an important problem and being unresolvable via any other way.
BIP-110 was not technically sound, did not manage to convince many economic participants, and had no benefit to a majority of people. It was also rushed, giving little time for a possible UASF to be even reasonably possible. If anything, it was against the interests of many parties, even the “good” ones.
(Everyone is also strategizing about what everyone else will do, and so even just the appearance of broad consensus is critical to have a successful UASF)

2026-08-22 19:12:29 UTC

- reply

Bitcoin is not the heaviest valid chain, and there is no “definition” of Bitcoin either.
It is a vague term that exists because of a general consensus all sorts of participants, which have weight in different ways (economic, computational, leadership, publicity, etc.)
And there is no strict weight to define 50%+1 but many different powers that form a result.
A node by itself does not have any weight in this, and only its operator could hold weight by being economically active (buyer/seller/platform), mining, via public participation in discussions, etc.

2026-08-22 12:43:06 UTC

You cannot expect “creativity” from an LLM. It can only create “new” things via random search.
Any “gaps” that need to be filled from input => output will be filled with a somewhat randomized average of what already exists. It will not add any new substance.
#nevent1q…pnn8

2026-08-22 12:36:46 UTC

- reply

It is 99.99% air and 0.01% (mis)info

2026-08-22 09:34:55 UTC

AI generated “content” (slop) is a shell of the real thing. It *looks* sensible but it’s empty inside. It’s worthless and doesn’t make sense.

2026-08-21 07:20:14 UTC

2026-08-21 04:39:34 UTC

- reply

Why build the Torment Nexus from the book Don’t Build the Torment Nexus?

2026-08-20 07:12:49 UTC

javascript was a mistake

2026-08-19 14:55:30 UTC

- reply

it is (properly) called a convection oven

2026-08-17 17:52:39 UTC

The AI psychosis gets stronger by the minute..

2026-08-17 12:19:04 UTC

- reply

The latter is also not true.

2026-08-16 23:08:53 UTC

- reply

Y’all are visiting websites?

2026-08-13 20:04:08 UTC

- reply

Yeah, do it

2026-08-12 23:47:27 UTC

- reply

I have a hard time taking Passport Prime seriously after several of their claims and decisions. The idea makes sense, but at that point Betrusted seems like a better option

2026-08-12 08:49:05 UTC

expand 32-byte k

2026-08-12 03:53:27 UTC

- reply

I agree that macOS is pretty great in terms of security model, but that desktop is in general a mess …
I’m more curious about your opinion for more limited tasks, and not day-to-day usage, like hardened limited-purpose servers, or to process sensitive data, and such.
It is unfortunate we don’t have a lot of investment in this area.

2026-08-12 01:15:13 UTC

- reply

I am aware :) I need something that works on X86 systems,

2026-08-12 01:14:44 UTC

- reply

Ideally, we have a microkernel, and there is no dom0. The problem is that you are now reinventing the universe.

2026-08-11 23:01:33 UTC

- reply

I think that dom0 should stay a vegetable

2026-08-11 20:51:32 UTC

- reply

And I guess, it is more about the security configuration than anything, and distros. Very few things have SELinux properly configured, for example

2026-08-11 20:50:55 UTC

- reply

What would you say is a good desktop OS?

2026-08-11 19:13:58 UTC

- reply

Yeah. This is a common method, used by basically anyone serious (and not).