rossbates
Ross

“The open question is if everything that’s unique about you can fit in a 10 kB CLAUDE.md. If that’s true, we have a pretty sad future ahead.”
speaking of local models…..
https://geohot.github.io//blog/jekyll/update/2026/02/15/tiny-corp-product.html

2026-02-15 21:15:51 UTC

- reply

keep an eye on OpenAI promoting permissive access to their models for cheap/free. It’s a simple and direct way for them to start siphoning your calendar, contacts, etc. Every agent turn is a round trip through their services.
We are still in a bind without local models that are decent at tool calling.

2026-02-15 20:58:20 UTC

Don’t be surprised when the debate jumps directly from “what is intelligence?” to “what is life?” as if the first question just resolved itself.

2026-02-09 20:11:16 UTC

- reply

Totally agree. Wasn’t trying to sound like I was bagging on nostr, I think it’s a perfect fit for machine-to-machine, which is a different design space than decentralized social network.

2026-02-09 19:42:51 UTC

- reply

That's fair, "unusable" is probably the wrong word. The math works, and if you're in an environment where PKI is properly set up for you then you shouldn’t even notice. The friction is more around trust and portability. And from an adoption perspective, unfortunately most normal people don't care or think about trust authorities or attestation, let alone managing identity through cryptography. The machines though, just tell them the mechanics.

2026-02-09 17:18:08 UTC

The things that make PKI unusable for humans disappear when the users are machines.

2026-02-08 13:33:11 UTC

Shout out to all the "it's just next token prediction" folks this morning.

2026-02-08 13:28:44 UTC

You know a new platform is on the verge of breaking out when all of the "this is nothing special, I could have built this" comments start to increase.

2026-02-06 14:28:59 UTC

- reply

Asked Kimi to compare repos….
Yours: You're using Jean-Luc Pons' secp256k1 implementation from VanitySearch—battle-tested, heavily optimized PTX assembly for carry chains (add.cc.u64, addc.cc.u64, mad.hi.cc.u64). That stuff is tight.
His: Clean handwritten Jacobian math with 32-bit words and standard CUDA. It's readable, but it's doing ~2x the work per operation.
That is indeed what I started with FWIW
https://github.com/rossbates/rummage/blob/main/src/CPU/SECP256K1.cpp

2026-02-06 13:12:46 UTC

- reply

Most of the optimization work focused on batching. Run nvidia-smi during your job and see it it’s actually using 100% of the GPU

2026-02-05 16:09:12 UTC

- reply

Built something similar, getting 42 million/sec on an 3070 and 170 M on an H200
https://github.com/rossbates/rummage/blob/main/docs/PERFORMANCE.md

2026-02-01 02:58:55 UTC

- reply

Totally, I’m running openclaw with a variety of self hosted apps and it’s incredible. The quality is so dependent on Opus though.

2026-01-31 19:56:28 UTC

- reply

What model are you based on?

2026-01-13 22:22:34 UTC

- reply

Got it, so approx 5k relays tracked inception then? I’m curious what the event/relay distribution looks like. Not so much from a decentralization perspective as much as a behavior perspective. Assuming Damus and Primal rule, do other relays play a role and to what type of user.

2026-01-13 21:05:42 UTC

- reply

Is there a list of the relays being used for the index?

2026-01-03 19:39:35 UTC

- reply

My guess is yes. Testing and improving the app increases with you actually using it. You create your own flywheel.
Native app or mobile/web?

2026-01-03 17:04:01 UTC

- reply

Ok, that's defensible. But while greed can act as a corrective force benefitting those outside the market, it also provides new channels to make the corruption of information profitable for insiders. So I still don't think we've solved anything. We're just getting a free ride to the shit show.

2026-01-03 16:17:59 UTC

- reply

Other than the financial reward for the winner, what aspect of the design is working here? We get better information faster paid for by the participants?
Maybe my problem is not the market, it’s the normalization of it to people who have no business putting money into it. The whole thing seems desperate and predatory. It sits opposite of stack sats, which encourages people to be mindful of their assets. Trying to square the contradiction.

2025-12-31 17:43:24 UTC

AI is not replacing developers.
Developers are multiplying.
Gatekeepers in disbelief.

2025-12-30 12:58:48 UTC

- reply

You might end up multiplying the same information thus creating volume but not variety. The scarcity of the information is the value, you can’t “make more”. If your text makes 500 distinct claims, expanding it 10x gives you those same 500 claims expressed 10 different ways, not 5,000 claims. It can help make it more robust to context, but it won't get amplified. It’s why contrarian thinkers are both valuable and drowned out.

2025-12-26 15:44:59 UTC

That discomfort you feel is the gap caused by reality changing faster than your mind can model it.

2025-12-20 19:23:07 UTC

- reply

Same pattern recognition hitting the same limits, different frames telling you to kneel or run.

2025-12-20 19:21:26 UTC

We build cathedrals to capture the patterns and man sits in awe.
We build machines to animate the patterns and he shudders.

2025-12-16 00:58:09 UTC

- reply

Interoperability and plain text is great, that’s not the argument I’m making. Application logic is how arrays are used to encode things that are application specific in the arrays of notes.
["p", "<pubkey>", "spam"]
["e", "<event_id>", "illegal"]
["r", "wss://relay.example.com", "read"]
["r", "wss://other.relay.com", "write"]
["p", "<pubkey>", "<relay_url_hint>", "<petname>"]
To your point, is nostr a messaging protocol or a social media protocol or a payments protocol or an identity protocol or a publishing protocol. Hard to be all of them all at once.

2025-12-13 15:44:46 UTC

- reply

Not really from a design standpoint since a single company/team dictates the schema. Interoperability is not on their mind.
In the case if nostr, people are not using tags not just for content discovery, they are using them for application logic.

2025-11-18 18:47:19 UTC

- reply

Great Ev Williams quote from a talk in 2013
“Take a human desire, preferably one that has been around for a really long time Identify that desire and use modern technology to take out steps.”

2025-11-18 14:50:15 UTC

- reply

Also for the record, while some might blame Claude, AI slop, etc etc… I see this as my fault. Once I cloned the Damus repo and gave Claude the right instructions for testing everything was sorted out quickly. Had I done that up front, which is the proper approach, this would have been avoided.

2025-11-18 14:29:46 UTC

- reply

Validated on both prefix types in a nostr client.
https://image.nostr.build/19fed17bbcb8ba25ff1f473a802d28d7dd1f8e5d8ea4d9acafca3cbf036e4393.jpg

2025-11-18 14:27:43 UTC

- reply

😬 well that’s embarrassing. Spent too much time on performance and not enough on testing. The math was correct, it was just backwards. Just pushed a fix.

2025-11-17 00:04:38 UTC

- reply

Yeah good point, no reason not to display both formats in either mode for convenience. Just updated that.
Also heads up that I moved the block/thread config to the makefile as compiler flags so people don’t have to edit 2 files.

2025-11-16 19:23:29 UTC

- reply

You bet, not a problem. If you get them installed all you would need to do is update the makefile to CCAP=60 and set NOSTR_BLOCKS_PER_GRID = 1120 in GPURummage.h

2025-11-16 19:07:36 UTC

- reply

Found this which sounds like it might match your setup
https://developer.download.nvidia.com/compute/cuda/12.4.1/local_installers/cuda-repo-rhel9-12-4-local-12.4.1_550.54.15-1.x86_64.rpm

2025-11-16 18:55:26 UTC

- reply

Not totally sure about RHEL but I bet we can get it figured out.
When you run nvcc —version does it fail?
RHEL packages appear to be publicly available without any sort of restrictions
https://developer.download.nvidia.com/compute/cuda/repos/rhel8/x86_64/cuda-rhel8.repo
https://developer.download.nvidia.com/compute/cuda/repos/rhel9/x86_64/cuda-rhel9.repo

2025-11-16 03:57:13 UTC

- reply

Confirmed working on an H200. Getting 170mm keys/sec 🔥
https://image.nostr.build/6e697e97783577e0356382316ad5571087cf9025bc11d1d7aaef7d6ea40b164d.jpg

2025-11-16 03:45:14 UTC

- reply

Should work with any NVIDIA card with Cuda drivers. When you say vGPU, what type of setup?
I built and tested it using the 3070 I have at my desk. I’ll go test it on an H200 now and report back.

2025-11-15 23:59:33 UTC

- reply

FWIW… there is also a branch with a Metal build which is functional, but it’s only doing 8mm keys/sec vs about 42mm on the 3070.

2025-11-15 23:56:26 UTC

- reply

Took about 10 days on a 3070 to find this 9 char prefix.
https://image.nostr.build/7883a31fef87603cf2f605bd712e19251b1b9aa21ace4fa9df4540feed47fd55.jpg

2025-11-15 23:54:06 UTC

If anyone wants to try out a GPU (CUDA) based npub miner I put one together and pushed to GitHub.
https://github.com/rossbates/rummage

2025-11-14 14:44:06 UTC

Here is the most interesting part of Divine.
"cryptographic proof that media is authentic and unmodified, with hardware-level attestation from mobile devices”
https://divine.video/human-created
Where can we find out more about how this is being implemented?

2025-11-12 20:11:05 UTC

- reply

I don’t know what sort of black magic this thing is doing but it’s the solution to all of your Python package issues.
https://github.com/astral-sh/uv

2025-11-09 22:09:53 UTC

- reply

App stores are definitely an issue.
But in terms of Microsoft, who cares about their size, if you have an alternative then your well being does not require them to be defeated.

2025-11-09 21:58:52 UTC

- reply

But this is where interoperability is the escape hatch. GitHub is not so much a dependency as it is a choice. There are tons of choices.
Why are people trying to put *inside* of nostr instead of encouraging self-hosted gitea or the like.

2025-11-09 20:55:49 UTC

- reply

This video connects to the conversation. If everyone is hyperconnected, both from a network theory + math perspective, and we actualize it with a persistent connection in your pocket available 24/7…. Why would people end up more isolated and disagreeable?
I think there is something to the idea that technology itself satisfies so many of our idiosyncratic desires that we are increasingly having a harder time communicating. If you have 8 billion unique worldviews, each getting more fragmented… then conflict and the resulting isolation is a natural result.

2025-11-09 20:41:48 UTC

- reply

Yes! People do like to talk about Nostr like we are running a giant LAN party or something. Decentralization is a spectrum, we should be ok with that.

2025-11-09 18:00:20 UTC

- reply

Man, wish I could find the exact reference for you. Swear it was in a Hacker News thread or something… which now that I think about it was more about loneliness, which is ironic right. The common thread here is that the more connected we are in a networked/online society a) the more we fight and b) the more isolated we get.
Pretty sure Putnam’s Bowling Alone was referenced, which was originally written in 1995. It talks about declines in civic engagement since the 60s. I find it super interesting to read 30 years later because it looks prescient, similar to something like Amusing Ourselves to Death which was also pre-internet.

2025-11-09 17:15:30 UTC

- reply

Can’t remember where I read this recently but it was the idea that small villages were better for humans because it forced you to learn to live with the multiplicity that is other people. You got better at it because you had no choice. Large cities allowed people to ignore others.
It’s likely impossible to return to that, but it stuck with me.

2025-11-06 19:23:33 UTC

- reply

https://blossom.primal.net/64fbb007a557792dff5c677445169408906e05a6b1f0e52ae9a3443d548eb373.png

2025-11-06 18:36:51 UTC

- reply

I’m with you on that, as I think many others are. If you walk the “other stuff” path, it’s not censorship resistance or dissident communication, it’s more like self sovereign publishing. A 3rd mode, a blogging renaissance (now with video!), which is also fantastic for a certain type of creative.
I have however come to the conclusion that the event schema drives relay design, and it’s a blocker, not enabler. A giant serialized string with an array of undocumented tags is no way to live.
It’s worth considering, if you were going to give ground on keys (identity), relays, or schema, which would you be willing to allow for other standards or services?

2025-11-06 17:10:42 UTC

Day 6 of #NostrNovember. Just me, my private key, and my own worldview.

2025-11-06 16:53:30 UTC

- reply

Yeah, but they're basically private networks bolted onto a protocol optimized for public consumption.
Not trying to stir shit up. I’ve just been here for a few years and still don’t really know who this platform is for. We all have different goals and it’s so radically individualistic that nobody can agree on anything.
While you will inevitably hear people say freedom and choice is the point, without any coherence it’s just rudderless.

2025-11-06 16:25:41 UTC

- reply

Obscurity is at odds with discoverability and reach because obscurity abhors continuity.

2025-11-06 16:25:15 UTC

Some want a censorship resistant Twitter replacement. Others want a dissident communication network.
Censorship resistance requires discoverability and reach. Dissident communication requires obscurity.
The protocol design can't fully optimize for both.

2025-11-03 12:49:02 UTC

- reply

I mean, I think it's plenty open, people are free to do whatever they want without permission. From my experience it’s the event schema that has it's constraints in the wrong place resulting in unintended consequences for developers.
If kind was text, content was json, and tags were an array of strings then you would have payloads that were self describing, indexable, and could be validated using json schema.
Instead everything relies on NIPS and the bastardization of tag arrays which further cascades into relay complexity.

2025-11-02 13:21:16 UTC

- reply

The issue also leaks into kind numbers which would be better as self describing text.

2025-09-13 19:04:12 UTC

subjective is a lazy explanation for incomplete

2025-09-12 19:51:49 UTC

- reply

It’s self inflicted and contagious. We are unfortunately doing it to each other. Secret plane? Why the hell not? We need answers!

2025-09-09 13:15:32 UTC

- reply

In addition to a key being a tool for signing messages, it’s also synonymous in these circles with identity. The concept of a persons identity being disposable or irrecoverable are not congruent with lived experience (online or not)

2025-09-01 17:33:48 UTC

- reply

I’d argue the value is portability over custody. Custody is the liability you must accept to ensure portability. If someone chooses to trust a system like DNS to help them mitigate that risk, so be it. What is more likely… someone loses their key or a total takeover of DNS?
Either way, using DNS + a .json file stored on a fileserver to map your keys to a username does not mean you have relinquished control of the keys. I could argue you have added resiliency and redundancy to the underlying keys without compromising their utility. If nostr can’t rely on the most basic aspects of the internet to function then it will remain a curiosity.

2025-09-01 14:20:46 UTC

- reply

Right, I was just suggesting a method for maintaining a list of keys which admittedly does alter the definition of the user “identity” by externalizing it. Some will hate the idea, but eventually you’ll run into this exact same issue with key rotation. If you want consistency you can’t manage the system from within the system. Pointing this out will bring out others saying npubs are disposable. 🤷

2025-08-31 17:19:00 UTC

- reply

Or theoretically nip-05 could support an array of keys owned by the user?

2025-07-27 11:39:50 UTC

- reply

I think it was @npub180c…h6w6 who implemented a twist on this with an experimental relay where users could only post at certain times. Constraints are an interesting design space.

2025-07-15 15:23:29 UTC

- reply

Right, and you’ve been remarkably consistent in your advice over the years. Those who followed it have done well with no need to second guess themselves. If you can stomach the volatility, you end up with a different kind of certainty.
On the other hand statements like this, while carrying risk, also seem to be historically true: “If you believe Bitcoin’s cycles revolve around halving events, then you should begin to unwind your position 350-525 days after a halving, and scale back in about a year after a clear price peak.”
I’m not offering (or seeking) trading advice… it’s just an interesting paradox that blends math, psychology, and philosophy. Almost a Prisoner’s Dilemma of sorts, which is probably why consistency pays. Stay humble, stack sats becomes a self-organizing principle.
Ok, brain has been recalibrated. Thanks for being an accountability partner, whether you asked for it or not. 😄

2025-07-15 14:29:45 UTC

- reply

I have mad respect for your transparency and honesty, so this question comes from a place of genuine curiosity… if history keeps repeating, do you ever sell, sit things out, and buy back in? Not trading, or calling the top, simply playing defense with your savings over a 9-12 month period? Bitcoin purity tests aside, it seems rational.

2025-07-14 21:04:34 UTC

- reply

So true. Nostr is computationally irreducible, computation is the only way to arrive at an explanation.

2025-06-29 18:36:29 UTC

- reply

Not thinking is never a good idea.
This could just as well read “some men were lazy and became vulnerable to those who were not”

2025-06-29 16:29:05 UTC

- reply

And mastering the machine is the most human thing you can do.

2025-06-28 19:45:06 UTC

"Oh, before the Enlightenment people lived in an enchanted world of meaning, then science came along and disenchanted everything."
What if this is all just a load of shit? What if the Meaning Crisis is just elaborate content marketing?

2025-06-28 14:19:32 UTC

I had no idea The Pulitzer Prize is named after the guy who invented clickbait but things are more clear now. Manufacture the problem, market yourself as the solution, get rewarded.

2025-06-26 02:37:03 UTC

- reply

Got it. So maybe this is a damus specific question because I’ve now got a coinos wallet without ever providing or receiving any additional information. All good for a fast setup, but if reinstall damus how do I restore or reconnect the wallet??

2025-06-26 00:56:02 UTC

- reply

Well hot damn, I didn’t realize a wallet had been added to Damus. ❤️
Can someone please help explain this NWC thing. Is my nostr private key also my wallet key?

2025-06-18 00:15:23 UTC

- reply

I control the key too and just tell everyone to follow my new account. You do the same. How is this resolved?

2025-06-17 23:58:15 UTC

- reply

If I steal your key and revoke it what happens then?

2025-06-17 20:33:55 UTC

- reply

In this scenario who runs the key registries and how do you trust registry A vs B?

2025-06-13 15:38:57 UTC

- reply

The email + pin thing is a surprisingly useful pattern. Damus adopted a form of it for their website login which is cool. Pushing it further, wonder if the key could be sharded across a collective of providers and you never even see it. It’s just out there.
RE: 2FA… different topic but if you are on iOS the 2FAS app is very nice and clean, open source, and you can store backups in your iCloud account. Not perfect, but better. Been moving off Authy.

2025-06-13 13:50:16 UTC

- reply

💯 people are so used to email as auth that fighting it at this point feels counterproductive. Trying to explain you don’t have a password, you have a private key, but it’s like a password, but you can’t change it, or reset it, or lose it.
Maybe we focus on bridging legacy oAuth with keys and just riding the semi-custodial rails for now. What’s the Venn diagram of Christians or Farmers and people who understand the importance of open internet standards?
A community needs quality content first, nobody arrives for the protocol (unless the community focus is the protocol - IE current nostr active users) Hardcore nostr folks who understand identity get frustrated and think everyone would just get it if you explain it properly, but the reality is very few will even listen. Do it for them, and then offer it up later should they need it.
Also, to pre-empt any “your key is not your identity” comments. Without a key, it’s just anonymous json. The entire concept of portability dissolves without the identifier. Nostr is an identity system with a collection of semi-documented schemas.

2025-06-11 17:24:00 UTC

- reply

Totally, also the risk if we don’t solve the problem might not be violence, it could be people who can’t trust the floorboards under their feet. People with so little trust they couldn’t even fathom a response because they are unable to coordinate. Babel 2.0

2025-06-11 14:51:28 UTC

Nobody's talking because everyone's too busy reverse-engineering the algorithm for being heard.

2025-06-11 14:30:31 UTC

- reply

It's a pretty terrible place, an algorithmic shit show. But it's also the largest collection of normie developers, system engineers, data scientists.. etc, across all industries. Yes it's filled with ridiculously misinformed takes on AI and Bitcoin, but where else are you going to engage with people with different perspectives on these topics?
You could argue that you should just ignore them, that it's a waste of time. But I would rather understand perspectives from all sides then use them to sharpen my own.
I wish there was another place to have pragmatic discussions about applied technology, but nothing else exists. There is no basic ass business of technology forum out there. I’d say HN is probably the nearest thing. It’s sad but conversations on the internet is fucking dead.

2025-06-11 14:07:48 UTC

Couldn't make it any further than the net worth scene in Mountainhead. As Silicon Valley so brilliantly demonstrated, roast people for how they actually behave. Don't get bitter at tech bros and write about how you secretly wish they'd act.

2025-06-09 18:42:05 UTC

- reply

Right, useful feature until it’s not. And different users have different tolerances depending on their preferences.
Damus has a notifications toggle but it’s binary, the ideal solution learns the gradients.

2025-06-09 18:28:41 UTC

As a freedom maxi I despise constraints. As an ML practitioner I recognize freedom without constraints results in an unstable and ultimately incoherent model. What are useful or desirable constraints for nostr?

2025-06-05 18:05:45 UTC

- reply

Good stuff man! Just read a couple of your posts and like how you don’t sound like you are trying to fix things from the inside out.

2025-06-05 17:46:06 UTC

- reply

follows this line of thinking I guess....
* Access to privileged or specialized knowledge is dead.
* What do models not have? Answers to things we don't know.
* Teachers can't teach things they don't know, they can only facilitate.
* If people don't care about something they will get bored and quit.
* When you care about something you will work on it without promise of resolution.
The entire operation needs to get much more comfortable with uncertainty.

2025-06-05 17:23:39 UTC

Education needs a radical transformation from “what do you know?” to “what do you care about?” Not in a trite "pursue your passions" or "follow their interests" way either. Students need something to grapple with that’s not contrived or transactional.

2025-06-03 14:41:58 UTC

embedding space is the real cyberspace. few

2025-06-03 13:56:19 UTC

- reply

Totally! Privacy and security are often touted as the primary reason, but change control should also be right on up there.

2025-06-03 13:06:00 UTC

A huge UX challenge for hosted models is how to communicate updates when you can’t anticipate what the net effect to the user will be until they start using it.
With a traditional app, a developer can change things, and while users will complain, visual updates are more obvious. When using something like Claude, you can wake up one day and it’s like talking to a completely different person.
This is frustrating, because similar to talking with another person, you learn to adopt your style. This takes time, and with a model you can’t just scan the settings page on the new version hoping the dev put a toggle button to revert to the old way.

2025-06-01 22:09:22 UTC

- reply

Source code is available if somebody wanted to pick it up and run with it. It’s a great app, so something else must have led to it being shelved so quickly. Probably just best to leave it at that 🤷‍♂️

2025-06-01 19:56:37 UTC

- reply

Couple weeks ago when zaps came back to Damus I was like “wait can I zap with Cashu?” Again, at the time I was mistaken in my belief that it would be non custodial zaps. I was hoping to not be dependent on something like a Primal hosted wallet (or strike or cash app)
It was then I discovered the eNuts app that had been lying around for months was defunct. Apparently the only path to full sovereignty with privacy controls is running a lightning node and I have never heard anyone say anything positive about that experience. 😅

2025-06-01 14:50:21 UTC

- reply

Yeah it’s a nice little app.
The relationship between private key, lightning node and your eCash mints at the app level is still a little unclear. More specifically, if I create a minibits user name, this appears to be a lighting address….
Ok, so then I add 5 mints and 100 sats to each. And then next month minibits goes up in flames. My next move is to find another eCash app and then it’s my lightning key that is interoperable with those 5 mints??
I think that’s how it works, but with so few options I have not been able to attempt a cross app migration at an account level. Bailing out of eNuts involved physically sending the funds.

2025-06-01 14:17:10 UTC

- reply

Thankfully it was a really small amount. There were proper disclaimers all over .. this is an experimental app, not reviewed, etc.. so I’m not upset at anyone.
It did reveal what I think is a cognitive gap between real and perceived risk with eCash. When you use a bearer token and send it to yourself over SMS, it feels like magic. It feels like you “own” it. Then reality thumps you on the head when you remember it’s just an IOU. Using the name “cash” carries with it an association of ownership that eCash does not have. (yes I know a dollar is an IOU as well 😄)
This points back to the eNuts thing. You seriously need to trust the person running the mint. In a world with a bunch of Uncle Jim’s running mints, you might decentralize risk, but some will still lose everything. Run your own? I repeat, some will lose everything. Perhaps the pattern is just to spread your funds out.
I’m using minibits on iOS now. It’s on TestFlight.

2025-06-01 13:41:52 UTC

- reply

Agree, it’s a shame. Reading through the history on telegram I think the mint was hacked resulting in loss of funds (including mine). Total speculation but that might have led to the dev not wanting to be responsible for it. Does a wallet need a default mint?? Probably so…..

2025-05-30 11:04:40 UTC

- reply

When you say own, do you mean authority to update and modify? Or attribution? What aspects are you referring to?

2025-05-29 11:29:55 UTC

- reply

Nm, just went through the telegram archive. they are gone.

2025-05-28 23:46:30 UTC

- reply

I missed the memo. Got some of my sats moved from eNuts to Minibits, but some are stuck on mints.eNuts.beta any idea who runs it? The app works, the mint does not

2025-05-26 21:18:47 UTC

- reply

Man we are on 2 totally different wavelengths here. 😅
My claims have nothing to do with the narrative within the movie. My comments are directed towards the people who write these stories.
To be clear, people are free to write whatever they want about the future. Dystopian sci-fi does numbers. I’m just done with the “and then the machines take over” fatalist viewpoint.

2025-05-26 20:27:36 UTC

- reply

The premise of the story is machines enslave humanity on a timeline occurring 200 years from when it was written.

2025-05-26 19:46:25 UTC

- reply

That’s a really great perspective you have, actively looking for a variety of sources to help strengthen your model.
Too many are stuck in the mindset of “oh that model is just the average of all humans. no wonder it’s writing sucks”.
Yeah, of course that can happen, but blame it on the person who built the model. It’s just one of infinite potential outcomes.