Bali

I'm considering this AI-coding bootcamp program: https://www.gauntletai.com/

Anyone have any thoughts or run into someone who has done it?

The following example encodes the ascii char "h" (or the bits "01101000") with a chunk_size=2.

Innocuous is lilting around in a "calm before the storm". Basically need some test infrastructure code before one last big refactor which will add headers to encoded/decoded data which include a length, nonce, and version along with the actual data.

Been talking to a lot of people IRL about the project but not many seem interested or can think of use-cases. I did talk to one lady this week at in at ETH conference that had written papers on this underlying subject called "steganography". On a walk I realized one of the primary use cases is procedurally generated NFTs.

I've also been able to develop an easy way ti explain the concept in markdown which I'll paste below.

Thanks for doing these @deSign_r!

This is fascinating, thanks for sharing.

Reading through, I'm reminded of the infamous DropBox Show HN initial comments where the programmers were like "this is just rsync" but the product ended up being successful because the UX/usability was superior for non-programmers.

Isn't version < 1.0.0 still supported via: https://github.com/stackernews/stacker.news/pull/2162 ?

Yes models must match, and it's trickier because there are often dozens of "levels of quantization" for each every model. But the particular model / quant-level expected can come baked into a version code-number or its own meta-paramater and be checked rather rigorously by asserting on the model-weight's hash at load time.

The opportunities are pretty wide in my opinion, but will take years and many researchers to ideate on. For example, take the Spaces Protocol below, the what, how, why of creating identities or address space on a blockchain took a decade from when the earliest of adopters started thinking about blockchain technology.

Basically if more of communications continues to go into/out-of/in-between LLM's there's this secondary "data layer" that can be hidden or exposed with ideas like this. There are freedrom and opensource concerns too since the major similar art is Google and the like developing internal system to watermark their model output: https://deepmind.google/science/synthid/

Thanks for taking a look! The idea that you're getting at is quite important. I've imagined there's a "standard" where the first two bytes will represent a "version number" which will set a value for the free floating paramteres, and have lots of different initial_prompts to produce a variety of texts.

version=96version=96

prompt = "Once upon a time, in a kingdom far away, there lived a"
chunk_size = 3
model = Mistral7Bv0.2Q4

version=154version=154

inital_prompt="The algorithm processes data by first analyzing the input and then"
chunk_size=2
model = Llama4.1-8B-Q6

You will only need to remeber encoded_prompt which will have the data + version number encoded in it. So it should eventually be able to work like opening up .docx with MSWord. If it's a valid text create by the encode, it will open up the message, if it's invalid it will fail or show random characters.

I get that people's instincts are let me hide my seedphrase in there because it's the most obvious thing to hide but it's not really the correct fit (IMO). There's other opportunities that I see opening up after people think about this concept for a week...

My takeaway is that stablecoins won't be allowed to defeat G7 capital controls but can offer an "American bank account" to the developing world. Sounds promising...

But will this be allowed in China? Or India, or Russia, or Brazil? All have moved pretty heavily to block disruptive technology and financial exports in the past decade.

And if not any BRICS, then maybe you get an Argentina or Nigeria or Vietnam? But at this point, you are developing an enitre industry where permission/adoption is an outlier instead of where the original vision of a universal rails where rejection is the outlier.

General Wellington said The Battle of Waterloo was won on the playing fields of Eton.

And Director Aaron Sorkin implied The facebook social network was coded during fraternity rush week.

What I recall most positively about college was the gentle and most open-ended call to develop agency. It was a world where you could, indeed, just do things. Spend $190 on throwing a party and try to feed yourself for the rest of the week on $10. Sleep fourteen hours. Sleep zero hours. Burn out on cheap weed, or wage-max with three part time jobs.

The credits, the grades, your major and minor gave the adults a rationalization for what you were doing. But really you were like kids playing house or dress up. There was a vague recognition that your were in a temporary developmental stage, but what you could become was an open path that wasn't just decided by "the system" anymore, that you could have a say in.

I think these increasingly common cases of 18 year olds foregoing college to do a "tech startup" are going into the one of the harshest agency training programs - a triple black diamon if you will - and are likely to emerge more jaded, more cynical, and less filled with hope than they could have if they tried a few years on the bunny hill first.

Just as General Wellington credited his officers discipline and teamwork inclulcated at college as the deciding factor in his military success, I think a generally whimsical and agentic nature is the critical ingredient which powers much of today's success in innovation. I think this will be difficult to create outside the dorms for now.

Still working on Innocuous, a way to encode/decode messages in LLM generated output: https://github.com/sutt/innocuous

I like my new example. This text:

Amidst the ancient forest, dwelt a wondrous Wizard renowned for his arcane might. One day, he unearthed true power not in enchantments, but in wellaimed words. Fearing misuse, he penned this power three ways in his magnum opus obscuring it with tedious trifles.

Time passed, the wizard departed, while countless seekers puzzled over his laborious tome. Until one sage, whose curiosity was matched only by fortitude, finally discovered the cryptic keys tucked deep within these words un

Decodes to: "pip install innocuous"

This is the perfect resource. Thank you.

I think SN has a standout system for FOSS PRs that most other projects lack:

• Disciplined maintenance of issue tracker: this takes a lot of time to do right, and OSS devs want to write code, not "be a PM", so the priorities of the project maintainer become illegible to outsiders looking to get involved.
• Comprehensive dev environment: "works on my machine"-friction adds a lot of costs to collaboration. NB: i think this is what's holding back a lot of "drive by contributions", I dont think there's any AI coding system that can handle the docker environment without a lot of expert customization.
• Consistent response time and courteousness: On other projects you'll often see a hostility to even earnest contributions, or it's a side project that got big while the maintainer has a full time job so PRs go un-reviewed for weeks or rejected for petty reasons.
• Not too large, not too small rewards and tasks: ONE BIG REWARD type bounties (or hackathons) are good for marketing and attract lots of contributors but here, the median quality will be quite low.
• The contributors are almost always users of the product: whereas for many other projects it's specialized software that each person uses differently in private on their own computer, e.g. MoviePy pakcage.

Overall this seems to hit the sweet spot of being welcoming and soliciting useful things. But the cost to duplicate these features for other projects also seems high.

Some other incentivized contributions I've come across that I think could be relatively AI-resistant are perf-based challenges like tinygrad bounties and kaggle competitions where a test-harness can be applied in a semi-automated to see if the contribution boosts accuracy or perf on some metric, and if so, then look into the code quality.

That's a good point. But even so...

If I recall correctly, other than now defunct Mutiny wallet, there's no lightning wallets PWA's?

And I think their devs stated it was a really difficult problem to build a mobile lightning wallet without the native app features.

This is an automatic kill switch to ever hit mass adoption for MoE.

Change my Mind.

Who's gonna L402-pill hacker news?

Update, I've written the first draft of my idea (cross-posted here: #1085919)

This address: 12Wfw4L3oPJFk2q6osDoZLYAwdFkhvgt4E (which can be seen on mempool here) can be represented as:

The king, whose power and pride, were known through the realms of earth to shine, In harness proud as eagle with the morning wind entwin. Now with what grace he did command his people in this wise. Their daily toiling lives for them to thine

or

The king, a man of noble grace, did dwell in state within a towr that loomed above all other loft, where eagles, soaring free in azured heights would cast, in envied sight, a winking glace at him who ruled, and in his hand a golden rod of rule he held alight, with sceptered power to rule and right, and with

It doesn't have to be "The king..." that's just the first prompt I was trying.

Source: https://github.com/sutt/innocuous

Thank you!. And yup any arbitrary data can be embedded. Seeds, addresses, love letters, the secret ingredient to coca cola.

The tricky part though is both parties (sender / reciever) will need to run the same quite complex code suite, same LLM weights, same LLM inference process, and same version of my "innocuous" algortihm on top. Otherwise they might misunderstand each other.

The benefit is it's decentralized, unlike something like cashApp where you can send me coins to my $username, but they could always de-bank me or alter the address unbeknownst to the sender and steal the coins.

I think my target for now is addresses and nostr pubKeys. But I want to get those tool into other peoples' hands to see what they can think up with this.