Lots of bashing here so I feel compelled to say my piece that GoPros, despite their flaws, are still the best action camera in the market for many people, including me. They have a powerful set of customizations simple not possible on the their competitors via the Labs firmware. Not that DJI couldn't do that if they wanted to, but to this day, they don't.
This is going to happen, but it's too expensive for your LLM to do the scanning, and instead someone needs to build and maintain the index while allowing other people to subscribe to concepts. The problem is no one has sorted out the embedding space this all lives in.
It could be that soon we're gonna get a fully personalized briefing on the topics that we're interested in, or maybe a new kind of feed, replacing social media.
keywords are a start but not enough imo - consider a concept subscription such as "any of my political representatives making statements about firearm control"
That's why I reached for Apples own local LLM to fool with similar ideas like this: https://pageforth.com. Apple is better than I expected at this. Right now it filters through things like hacker news articles and whatever else you point it at to summarize and find things that match your interests. Apple's LLM reminds me of Claude like 3 years ago. It's weak for sure. But useful for small dose kind of problems.
I disagree, their findings should generalize to the frontier. Even if the latest can deal with the extra complexity, it stands to reason it will take more tokens to do less. This could be a useful insight into the next generation of evals.
It's so obvious to me states need to create a soul bound identity system, replace social security numbers with it, and then let everyone else use cryptography on top of that (which is now cheap when you don't care about sybil attacks) to do private stuff.
The places you actually need an ID are so rare, I don't think it's worth it to build such a system (and no, porn or social network definitely aren't valid use cases).
I think it would make the web MORE anonymous, not less!
The reason it's hard to boot up a secure social network (such as Signal) is the handshake for (re)identifying people. Signal makes a ton of conceits here (the UX essentially asks people to assume phone numbers are securely held) in the name of low friction and it's why they grew so fast. The "real" secure social networks are essentially too difficult to get real adoption because they don't make these conceits around phone numbers, and demand real key exchanges.
But if you had a L1 set of private and public keys the government works to maintain and defend, the L2 social networks like Signal (or banks, or markets, whatever) can do this cheap and easily.
First I'll say the government already has an ID system with a backdoor they mandate you use (your federal social security ID and state ID). The backdoor isn't very interesting because anyone with your ID in hand also has it.
So how about this:
1. State assigns citizens an ID at birth
2. State allows citizens to submit a public key along with their ID at any time
3. Citizens can go to their bank / private social network / whatever and say "this is my public key, you can use it to sign messages to me, and you can verify someone a) alive and b) a citizen of $state is reading it (from here you can bootstrap whatever protocol you want)
4. The state<>citizen network established in (2) is constantly under attack as stealing someones private key valuable so you also need a legal and technical framework to defend it
The protocol for submitting private keys and defending it from attack is a much longer post, I'm convinced there are ways to do it that drastically favor defense over offense, but that's not the point here.
Our question is can a government force it's way into the protocol you bootstrapped on top
How would they?
1. They could reset your public key to one they control the secret to, and then impersonate you digitally to break into your bank or social network. However I don't think they could do this secretly (the key update would necessarily be publically visible), so it's not really a back door. They can already do this with a search warrant. And if you're paranoid you can bootstrap your secondary cryptographic networks with multiple factors. So, this is on net more secure for you.
2. They could try to recover your secret key by force or warrant - but again not a back door.
I think the real concern isn't backdooring it's blacklisting, if this system becomes the L1 for every L2 crytographic interaction, they can practically remove your ability to freely transact. But that's a political problem you address with political means, I'm convinced from a technical perspective this is more secure and far cheaper for everyone.
Whatever clever crypto system you think of: if it needs to work for the general population, it needs to go hand-in-hand with UX.
Say your example: a user generates a pub/priv keypair locally and shares the public one with the government.
How does the government know you’re rightfully sending the ID?
How does the user know what they are sending? Can the app/website/tool/person at post office they are using to generate+store+send the public key be trusted by the user? How can the government give trust to the user that this tool/person can be trusted?
And there we have attestation again. Or walled app stores, or certification as we have for physical services.
My driver's license should have some anti-tamper identity proof that can do a challenge response. Or let me go pay a few bucks for an identity proof at the post office.
There must be a dozen other ways smarter people can think of but identity verification kills profits so the smart people don't work on them IMO. It's more profitable for social media to be an astroturfed shithole. It's more profitable to remove control of your PC.
End users should be authenticated so you can prove you're selling real eyeballs in the demographic mix you claimed to marketers and to provide lip service for the 'think of the children' regulators.
But anyone who's paying for ads should have as little friction as possible to dropping money and spewing garbage.
I'm surprised nobody is looking at some sort of "corporations are people" angle here-- we've attested the device ownership, but it's owned by the Lorem Ipsum Corporation, which is a legal/demographic dead end and spawned just long enough to buy the device.
Yeah, agents are making self sovereign identity so much more relevant. We have all the technology. But identity is the main driver of the monopolies, they won't give it up unless forced to, maybe not even then.
We also need liability. Every time someone’s data is lost, the company losing it must be held accountable. They owe us huge amounts of money, and executives + board members should be jailed. No free pass.
Let’s see then if they really want to collect all our information all the time. Right now, they take it and handle it irresponsibly because they’re free from consequences.
The dependency tree for anything in the software world is so large, that liability like you describe is not feasible. Tomorrow Anthropic's latest model will find a RCE in SYNs being sent to a server? Who is "liable" when you lose your Google account, your bank account, access to your car and all ways to prove to the government you are who you are all at the same time?
You just need to deploy auditable (source-available, reproducible-build, firmware checksums LCD on-chip) biometrics booths that generate private keys from normalized biometric inputs, and then use those ephemeral private keys to generate and sign portable identity keys. Most people have fingerprints and retina patterns and that’s twelve signatures on an identity alone, allowing for continuity across severe biometrics events like regrown fingertips etc.
A nonprofit business could do this if backed by all existing dotcom and bitcoin billionaires. But they’d all want to profit from it, so either non-profit (NGO) or governmental it is.
Fun fact: this is already a core function of USPS. They serve as an identity verification hub for both US passports and their informed delivery and PO box services. They just have a human-dependent process rather than an identity-generator booth. So they’d be perfectly positioned to take your ID, hand you an attestation request QR code, and get your identity-signatures on it — without being able to reverse-engineer your biometrics from those signatures, but still being able to detect gross variances when someone else tries to lie about being you in a future verification.
Anyways, none of this will likely ever happen, but the rich tech folks could make it happen at any time if they cared to. Instead we get THE ORB which is doing retinas as a for-profit without auditable artifacts or hardware. Sigh.
I think you can do it without any biometrics at all, although using it as a second factor could make it smoother.
I'd propose the primary factor is social - when a child is born there is a recorded attestation from the family and care providers about the minting of a new soul. When keys are compromised you similarly seek attestations from your social network (or social worker) that you need to furnish a new key.
The network could be attacked by literal force, blackmail, or deception, but it's very expensive compared the defense (strong legal punishment for attempts to subvert the network)
That last part is why I think the state has to do it, not technologists. There has to be a strong legal and cultural immune system in place to defend the network.
That’s adjacent to birth certificates and passports already, with some variations on a theme per country, but certainly I don’t object to it. But I’m still infuriated at having to provide a birth certificate to LinkedIn to support a legal name change, so I encourage further design at the interface between “citizen identity” and “online identity(s)”. Your idea has merits and isn’t like others I’ve seen, so it’s worth considering in more detail!
>biometrics booths that generate private keys from normalized biometric inputs
Isn't this basically worldcoin? Aside from the fact that worldcoin is run by people I wouldn't trust to watch my cats for an afternoon, the core principle with well thought out ZK crypto could work well.
Everytime I've asked a model to write it's Agents/Claude file it's been pretty bad actually, are you sure writing these files is actually in distribution right now?
We really need proof of soul systems to exist, extended to also have a proof of citizenship. While the proof of soul systems can plausible be done in a decentralized manner, proof of citizenship is much harder, and in my opinion this is one of (the few) things the government should really do.
> What ZKPs don’t do is mitigate verifier abuse or limit their requests, such as over-asking for information they don’t need or limiting the number of times they request your age over time. They don’t prevent websites or applications from collecting other kinds of observable personally identifiable information like your IP address or other device information while interacting with them.
Interesting. While that is true I don't see how it's an argument against. Over-asking + ZKP certainly seems superior to over-asking + without ZKP. Without ZKP in a world where you constantly need to identify yourself you have absolutely no privacy.
And going forward I think that any communication without establishing some kind of trust boundary will just be noise.
Yeah one reason I think the government has to offer this is usability. While you can imagine a purely p2p protocol between cypherpunks, for everyone else there needs to be a way to social workers, DMV staff, etc can deal with edge cases (such as your id being stolen and needing a reset). Furthermore it helps if it's super illegal to tamper with this network (consider how rare check fraud is, despite being easy).
Sorry the term of art is really soulbound identity right now, I use POS but it's less common. Definitions vary but I say a useful system must allow people to endorse statements with evidence they are a) alive b) not able to be represented by more than one identity (id is linked to your entire soul, not a persona or facet of your being) c) a kind of socially recognized person (human in the expected case)
and then layer on citizenship on top if you want to use this for polling, voting, etc.
All you have to do is flip the tortoise back over.
> You’re in a desert walking along in the sand when all of the sudden you look down, and you see a tortoise, it’s crawling toward you. You reach down, you flip the tortoise over on its back. The tortoise lays on its back, its belly baking in the hot sun, beating its legs trying to turn itself over, but it can’t, not without your help. But you’re not helping. Why is that?
It's funny to think of how the US government is effectively a decentralized web of trust system. Building one that works, that has sufficient network effects, auditability, accountability, enforcability, so that when things are maliciously exploited, or people make mistakes, your system is robust and resilient - these are profound technically difficult challenges.
The US government effectively has to operate IDs under a web of trust, with 50 units sitting at the top, and a around 3,000 county sub-units, each of which are handling anywhere from 0 to 88 sub-units of towns, cities, other community structures.
Each community then deals with one or more hospitals, one or more doctors in each hospital, and every time a baby is born, they get some paperwork filled out, filed upward through the hierarchy of institutions, shared at the top level between the massive distributed database of social security numbers, and there are laws and regulations and officials in charge of making sure each link in the chain is where it needs to be and operates according to a standard protocol.
At any rate - ID is hard. You've gotta have rules and enforcement, accountability and due process, transparency and auditing, and you end up with something that looks a bit like a ledger or a blockchain. Getting a working blockchain running is almost trivial at this point, or building on any of the myriad existing blockchains. The hard part is the network incentives. It can't be centralized - no signing up for an account on some website. Federated or domain based ID can be good, but they're too technical and dependent on other nations and states. The incentives have to line up, too; if it's too low friction and easy, it'll constantly get exploited and scammed at a low level. If it's too high friction and difficult, nobody will want to bother with it.
Absent a compelling reason to participate, people need to be compelled into these ID schemes, and if they're used for important things, they need a corresponding level of enforcement, and force, backing them up, with due process. You can't run it like a gmail account, because then it's not reliable as a source of truth, and so on.
I don't know if there's a singular, technological fix, short of incorruptible AGI that we can trust to run things for us following an explicit set of rules, with protocols that allow any arbitrary independent number of networks and nodes and individuals to participate.
Some friends just made this: https://www.congressionalrag.com/ - they need help from anyone interested, especially around pulling in more data sources.
This is not obvious to me! For example, if you locked me in a room with no information inputs, over time I may still become more intelligent by your measures. Through play and reflection I can prune, reconcile and generate. I need compute to do this, but not necessarily more knowledge.
Again, this isn't how distillation work. Your task as the distillation model is to copy mistakes, and you will be penalized by pruning reconciling and generating.
"Play and reflection" is something else, which isn't distillation.
The initial claim was that distillation can never be used to create a model B that's smarter than model A, because B only has access to A's knowledge. The argument you're responding to was that play and reflection can result in improvements without any additional knowledge, so it is possible for distillation to work as a starting point to create a model B that is smarter than model A, with no new data except model A's outputs and then model B's outputs. This refutes the initial claim. It is not important for distillation alone to be enough, if it can be made to be enough with a few extra steps afterward.
You’ve subtly confused “less accurate” and “smarter” in your argument. In other words you’ve replaced the benchmark of representing the base data with the benchmark of reasoning score.
Then, you’ve asserted that was the original claim.
Sneaky! But that’s how “arguments” on HN are “won”.
No, I didn't confuse the two. There is not a formal definition of "smart", but if you're claiming that factual accuracy is unrelated to it, I can't imagine that that's in good faith.
Exactly. I think people want to be treated like owners more than they want an obtuse legal status of being an owner. In other words, a seat at the table when strategic objectives are being defined is what will make them act like owners, not a RSPA document.
It all depends on the people. I know lots of people who prefer working for startups not because they're looking to get rich, but because the startup is doing something new that particularly excites them.
And in my statups, I want that group of people, and actively don't want the ones who are just looking for a big payday.
I assume you are either already rich or want to become rich because of your startups. Why wouldn't you expect people to want to also be rich?
I will say every time I hear someone claim that their employees should want to join them for non-monetary reasons, I assume its to try to keep more of the rewards for themselves.
> I assume you are either already rich or want to become rich because of your startups
You assume incorrectly. I am not rich by the standards of rich people (although I am rich by the standards of many others, but I think it's safe to say that most of the people reading my comment right now are wealthier than me), and I don't form startups because of a desire to gain wealth. I have no actual interest in amassing wealth as such.
I form startups because they give me the opportunity and freedom to engage in projects that deeply interest me.
"Not rich by the standards of rich people" is a moving target as your social class improves and you rub elbows with the more wealth as you make more money. There are people making $500k a year in the Bay Area or NYC who don't think they are rich. But the simplest dividing line I can think of is "has fuck you money". That is, could you (financially, not socially or being bored) retire tomorrow if you felt like it.
I never said anything was wrong with it. I only claimed that's not the motivation for everyone, and in my own startups, I prefer people who are motivated by passion for the work over those who are motivated just by income potential.
That's not a value judgement, that's just being interested in finding people who are more on the same page as me.
reply