How could I have missed this beauty?

https://github.com/aws/aws-toolkit-vscode/commit/1294b38b7fade342cfcbaf7cf80e2e5096ea1f9c

A great post about Rust bootstrapping from a LinkedIn comment: https://graydon2.dreamwidth.org/317484.html

I vividly remember my time in Switzerland, when if you see a bright flash while driving — a speeding camera — you're waking up terrified for the next few weeks, since the letter with a fine and possibly more might arrive in the mail.

Easily one of the most negative emotions. And it's lasting. You really feel stressed while walking to the mailbox. I got cured of this in my first half a year in the US or so.

Apparently it's a European thing. I am flying to Europe tomorrow; to the UK through Europe, to be precise. The UK has the ETA process in place, and I have the ETA filed and approved from the previous trip. Continental Europe ... not clear.

I repeat: NOT CLEAR.

Those bright lawmakers and whoever else can't come up with a single web page that shows up for the search query of "Is ETIAS for Europe in place", and this web page should show a huge green NO, FREE TO TRAVEL or a huge red YES, NEED TO FILE ETIAS.

As of today, different websites provide different results. What is the official website is not clear. And I found some seemingly official page. It does not say yes or no. But you can "subscribe to alerts" about ETIAS. Which I did. And here's the email I just received from them.

Seriously, click on this link yourself: https://etias.com/articles/

And tell me: if it OK to travel, or does one need to complete any paperwork?

All I'm asking is the world where these answers are received unambiguously.

Ideally, as some on-chain statements, that are programmatically verifiable. So that if the service tells one "yes" and then the border control guard tells them "no", they are entitled to compensation, both for changed travel plans, and, a far larger sum, for emotional distress.

Why do you hate your users? Why do you think your users are scum? (c)

I guess the world just wants people who are not as naturally curious and demanding as I am. Which is sad to say the least. Since all the stories that "when it comes to personal liberties it was better in some 1960s and 1970s" are indeed true.

Uncharacteristically good for an almost-randomly dowloaded episode for a transatlantic flight:

https://www.youtube.com/watch?v=4-hj5uwclmk

I particularly like how Richard outlines different cohorts of values in America, with their histories, breakdowns of core values, and desires — both populist and long-term.

Would not be an overstatement to say that — on my journey to understand the world better and build mental models with better-than-instinctual predictive power — this episode will easily rank in the tops for a ~year.

(The previous episode of this magnitude for me was Marc Andreessen opening up on the previous administration wanting to make all AI and all Web3 a government-controlled thing.)

This also resonates with me taking a drive through Montana and Idaho a few weeks ago, where many Don't Tread on Me flags can be seen right next to the Stars and Stripes ones. Didn't think I'd see so many of them IRL, and oh boy I was wrong.

Here’s my take on the whole AI “doom” narrative.

I’ve been known for overly optimistic predictions. Like saying back in 2015 that SQL would be dead by 2025. Or calling Docker containers just a phase, with unikernels being the real future.

Here’s an even better one. In 2016, I spent a month in Thailand. Loved it — so I stayed another. And while I wasn’t bullish on self-driving cars, I was pretty sure Southeast Asia would be “basically the same, but with all-electric scooters” within a decade. They’re quieter, cheaper, GPS-enabled — what’s not to love?

Yeah, no. Didn’t happen.

Good things, oddly enough, tend not to happen.

People with agendas, though? They show up fast. Especially when there’s money to be made selling their side of the story to the public.

Take Big Climate. Greta sails across the Atlantic — heroic, symbolic — and then flies back home, undoing any measurable “gains.”

Or take Big Tobacco. They bought another decade of “smoking doesn’t cause cancer” by simply ensuring both sides were “represented” in the media — including the “we’re not sure yet” camp.

So yes, there’s a lot of money to be made in AI fear-mongering.

That doesn’t mean everyone involved is a paid actor. Some genuinely believe in the Cause — be it veganism, banning plastic straws, tracking carbon footprints, or pushing to pause or ban AI. I’ve met some of these people, and I deeply respect their sincerity.

Still — there’s a difference between real concern and a manufactured agenda. And in the case of AI, it looks a lot like the latter. Far more than in other cases that I’d rather not state out loud.

Now, before I forget, a semi-joke: I’m actually a fan of Greta. The more EU taxpayers fund research into renewables, the better it is for us, Americans. If the narrative says Net Zero requires maxing out solar and wind, and Europe foots the bill for that R&D — well, I call that a solid ROI on my American tax dollars.

On a serious note, dear Europeans — I pity how self-incriminating your politicians have trained you to think of your own lives. But that's a subject for a totally different conversation. All I'm saying here is that [macro-]economically the German taxpayers funding better solar panels production so that Texans get more power for [electric!] pickup trucks — thank you, we are grateful, please do keep sponsoring American well-being as opposed to caring for yourselves.

But with this “let’s pause AI” thing — I’m not so sure. To me, the worst possible move for the West is to start regulating itself on AI too soon.

Honestly, it’s weird. The public can cheer on preemptive airstrikes to stop a regime from getting a nuclear bomb — yet turn around and want their own country to slow down on AI. Even though a) it’s clearly going to matter geopolitically for decades, and b) rivals are full steam ahead.

PS: Dear Europeans, please do use air conditioning! I get that your politicians can convince you to accept all sorts of nonsense — but suffering in 30+°C heat? That’s just bad for your health. There have to be red lines. This one is way past mine.

One of the most eye-opening podcast episodes for me last year was Marc Andreessen talking about how the Administration tried to pressure a16z into avoiding investments in AI startups.

The top-down "big idea" was to let only a few giants survive and regulate the rest so tightly that smaller companies couldn’t make it.

On one hand, thankfully, that plan never fully played out in terms of market, products, and regulations.

On the other hand, when it comes to investments, cash flow, and talent — we’re pretty much living through the worst-case scenario.

Venture capital for software startups, as we knew it, is essentially dead. Reasonable seed and Series A rounds are almost nonexistent. What we see now are inflated valuations if you’re in AI, and rock-bottom ones if you’re not — both equally unreasonable.

In AI, the gigantism is hard to ignore. Small and mid-check investors are openly squeezed out. In big shops, employees are closed to not even receiving stock options — just the occasional generous bonus, more of an appeasement than real ownership.

It’s the worst possible flavor of late-stage capitalism. The number of investors hasn’t changed much, but, weighted by volume, money mostly flows through “strategic deals.”

Don’t get me wrong — I’d still take this over a committee-based, USSR-style allocation system. But the true spirit of capitalism that once drove innovation? Long gone.

If you’re an AI founder, you don’t think about pitching a hundred small investors anymore. You think about how to please the big one. And even if you do take smaller checks along the way, you’re ready to say “sorry” and part ways once the giant steps in, overfunds you, and forces them out.

For tech founders outside AI, it feels like venture-backed products are dead. Better to bootstrap, go solo with AI, consult for cash flow — or do all three.

The market works best when the Invisible Hand does its job — when many small players with modest amounts of money engage in countless voluntary, positive-sum transactions. That’s how real innovation happens.

To me, committee-based allocation of centralized funds is stagnation. It’s "great" for extracting value from ecosystems already dominated by behemoths. For creating new value in a flourishing, highly competitive with tons of innovation to unlock — not so much.

I can't stress this enough: when the "rules of the game" are well understood, the top down approach does work. But when much research is needed — be it in the world of atoms, in the world of bits, or in the world of understanding human desires better — we just have to keep enabling individual entrepreneurs in quantities to get to the bottom of this. So that we all benefit from the journey and from the end result.

The fact that the world seems comfortable with the committee-based top-down status quo is … depressing.

Good thing I’m in distributed systems. I’ve learned it the hard way that solid deep tech doesn’t raise big money. So I focus on what works: strong engineering teams, clear links between tech-enabled milestones and business value, and mountains of realtime data.

Still, part of me is hoping for a Renaissance — the spirit of the late ’90s and early 2000s Internet. Not the “social” part, but the era when global tech produced things people truly wanted to pay for. Not things users have to keep paying for because a monopoly has left them with no choice.

Humanity spends $1T annually on everything software.

That’s 2x tobacco, or same as alcohol.

Or one third of farming, or one fifth of mining.

Also, half of pharma, or half of medical devices. But barely one tenth of the entire healthcare industry.

In other words, if this AI thing can make healthcare 10% more effective, the entire IT sector would double (!) in size.

Keep this in mind when making back-of-the-envelope estimates on how effective our planet is, on where it is going in single-digit years, and on whether the AGI is around the corner.

What’s the state of the art for Trusted Execution Environments (TEEs) in Web3?

From my research so far, TEEs seem to be emerging as a major R&D direction — and they pair surprisingly well with the growing use of stablecoins.

Done right, trusted execution of blockchain and near-blockchain code blends the best of both worlds. From the centralized side: high throughput, low latency, tight SLAs. From the decentralized side: the emergent trust property, where customers don’t need to trust the service provider or their preferred cloud.
I'd even view them as the next step after ZK L2 rollups — highly cost-efficient, low-latency, and with a remarkable degree of trust. You could see 100x+ throughput improvements without the need for mutual policing and staking&slashing.

That said, I have two concerns:

⒈Regulatory exposure: TEEs sit close to areas that could be heavily regulated. I’d use them as augmentation engines, not the sole execution layer, so products can survive any disruption by falling back (at higher cost) until the community re-decentralizes.

⒉Hardware trust: if large chipmakers ever get caught with backdoors in TEEs, how quickly can we react? This could push us toward ASIC-based “secret sauce” for Web3 sooner than expected — a lot of fun as well btw!

I’m curious:

∙What does success here look like in one, three, and ten years?
∙Who are the major players to watch?
∙What other risks should we keep in mind?

Tell me more pls!

Thanks Jordan for yesterday, and our SysDesign Meetup episode from yesterday is now live on YouTube!

Yesterday I learned Oregon — the state between California and Washington — banned grades in schools.

Where exactly are we going as the civilization?

Also, more Europeans die of heat than Americans of heat and guns combined.

Not taking sides. Just think it’s valuable to be aware of useful facts.

#Russian

Попробую успеть сделать несколько слайдов — через два часа поговорим с @volyx про Web3 и Secure Enclaves / Trusted Execution Environments.

Привет! Сегодня стрим в 21:00 мск

Поговорим про Веб 3.0
- blockchain / decentralization / web3 products
- задачи web3 для инженера
- систем дизайн веб3 продукта
- эффективность
- расширяемость, надёжность

В гостях Dima Korolev

Кликайте Notify чтобы не пропустить

https://youtube.com/live/D1GeYGCcrSo?feature=share
https://youtube.com/live/D1GeYGCcrSo?feature=share
https://youtube.com/live/D1GeYGCcrSo?feature=shar

An adult revelation of the decade.

By my mid-30s, I converged on a simple belief: to become very wealthy, one must be extremely money- and status-hungry. Pathologically so. Because, honestly, what rational motivation exists to work this hard just to earn your second billion?

For most people, the threshold is far below the "three commas club." Beyond a certain level of comfort, most would rather focus on living well, spending time with family, and enjoying life.

There's even a term for this: F-U money — the point where you can walk away from games, politics, and the grind, and finally focus on what matters.

For years, I believed — and still mostly believe — that the relentless drive to keep chasing billions is inherently unhealthy. Yes, there are rare exceptions: people with a genuine mission requiring vast capital. But they're outliers. Statistically, I thought the answer was simple: billionaires and high-profile CEOs must be psychopaths.

The stereotypical ultra-wealthy operator thrives on power games. They can fake empathy convincingly, but when a bigger win appears, they'll backstab and betray without hesitation. That level of detachment felt almost genetic — as if their identity is inseparable from multiplying their wealth.

But now, in my 40s, I'm rethinking this model.

These people are often cold and calculating when it comes to money, and generosity is rare. If they truly believed in sharing success, they'd give away meaningful chunks of wealth to those who helped them — but that almost never happens. Most grand gestures of giving are strategic: PR, taxes, optics. Rarely because it "feels right."

And yet, I've realized you don't have to be a full-blown psychopath to succeed at that level. There's another trait that works just as well: the ability to rewrite your emotional past.

Imagine this: you once called someone a "friend for life," praised them endlessly, celebrated them — and then one day, they become a "liability" instead of an "asset." Most people would try to salvage the relationship, reconcile, accept imperfections. But highly ambitious people with just enough empathy won't bother. They'll erase the past — consciously, deliberately. "I never said that." "We were never that close." "That chapter doesn't exist." Problem solved. The game continues.

A sure way to alienate someone like this is to challenge their rewritten narratives. Try to make sense of a past conflict, ask why things happened, or remind them of the promises they made — and you'll hit a wall, or worse, become persona non grata because you are "not of their caliber."

You're not arguing with their rational self anymore. You're arguing with their reconstructed ego — built to protect the story they now must believe and the role they now must act out.

That's why, for many leaders, truth dies quietly — not on battlefields, but in boardrooms. Every step up the ladder, the percentage of people sincerely committed to truth and empathy drops sharply. At adulthood, maybe one in ten are their authentic selves. One level up, it's one in twenty. VP or C-suite? One in a hundred. Generational wealth? One in a thousand, if you're lucky.

The irony is, most of these people aren't actually psychopaths. They can care. They do feel. But their sincerity is fleeting. They can care deeply today and erase you tomorrow if it benefits them. That's the adjustment we, the "ordinary" people, have to make: accepting that high-status sincerity isn't the same as ours. Expecting it to persist only leads to rude awakenings — often at painful times.

And yet, I find this worldview oddly comforting now.

I no longer believe everyone "up there" is a soulless sociopath. Many are capable of genuine warmth and generosity — in the moment! We just need to recalibrate what those moments mean and avoid attaching permanence to them. Their narratives shift. Their reality is fluid. And if we don't understand that, we're the ones living in a false story.

Although, on re-reading this, I may have just redefined what "psychopath" means. Sigh. Well, forewarned, forearmed regardless.

A story related to the previous post is something I heard many years ago. My memory is blurry — it could have been something I saw personally, or something someone told me about themselves, or something they overheard. These details are not important though; it’s the gist that matters.

The setting is a VC pitching session.

There is a protagonist — a trustworthy person in the room, an investor with dignity and integrity. And there is a young founder, the antagonist — arrogant and, frankly, incompetent.

The protagonist has already passed on investing in the antagonist, for a combination of factors: many objective, plus quite a few subjective.

And now the antagonist has taken the stage and is pitching. Their pitch involves something demonstrably false — something our protagonist can easily disprove. For example:

We are in talks with ${protagonist}, and he is considering investing.

The person telling the story happens to sit next to the protagonist and knows them. Naturally, the question arises:

Why didn’t you speak up and say this isn’t true?

It would cost nothing. You’d earn even more respect from the room for keeping the event honest. You’d send the busted crook home empty-handed. And other investors — who might otherwise have invested in this person’s company — would quietly thank you later.

Here’s the protagonist’s answer:

Look, you’re right. This person is not telling the truth. But they’re arrogant, cunning, and they hustle well. There’s a nonzero chance they’ll become wealthy and influential over time. I do not want to be remembered as someone who stood up to them.

In other words, there are pros and cons to speaking up versus staying quiet.

In a reasonable community — which I would imagine such a noble gathering to be — the penalty for speaking the truth would be negligible compared to the value it brings, while the value of staying quiet would be negligible compared to the penalty of doing so. So speaking up should be the automatic, genuine, sincere, authentic thing to do.

Alas, this is not the world we live in.

The world we live in almost exclusively promotes the behavior of being a selfish, egocentric, arrogant bastard — the kind who will “fake it till they make it,” lying all the way to the very top; at which point, it no longer matters to them whether their past sins are exposed.

And this, to me, is the major problem with human nature.

No wonder we have psychopaths rising to the top!

It’s not because psychopaths are uniquely “well-equipped” to make large-scale, cold-hearted decisions and extract profit from them. It’s because we, as a society, enable this behavior — through our shared, collective silence.

When I was younger, I hoped the world would change for the better during my lifetime. These days, I’m no longer so confident.

Sure, we have modern technologies that promise to help — Web3, after all, is meant to level the playing field for all parties involved. But human nature has been evolving for millions of years, and we’re still mostly monkeys — monkeys who form hierarchies, who need pecking orders, and who almost always show their dark side when lured by substantial profit.

Baboons, interestingly, have evolved a parallel hierarchy of “teachers”: individuals who choose not to play status games and instead become egalitarian evangelists. These teachers are respected enough that even the alphas of the tribe trust their kids with them.

When I read about this a while back, I had an a-ha moment — that not all is lost for people with my preferences and life choices.

But even this remote possibility seems to be fading away for more and more people of my “tribe,” as the world becomes increasingly winner-takes-all. And the largest winners — mostly governments these days rather than John Galts — appear all too eager to support anything but egalitarianism.

I still respect ThinkPad-s a lot, although touchpads are superior to ... these things for well over a decade, to my taste.

Physical trackpoint buttons right on top of ThinkPad touchpads were a very, very big reason I was reluctant to use anything but ThinkPads for 20+ years though — shame they are being phased out as we speak.

Blast from the past!

Evidently the Internet did change quite a bit, but it's due to the AI boom, not social media.

Which is cool and unique and shiny in its own way. But also ugly, but hey, no one promised this to be a smooth ride.

The largest lesson to draw is that both Google and Facebook/Meta are alive and well. Microsoft secured a win with Bing, by finding it strategically for years and years until an opportune moment showed up. It looks like Google is playing the same game with Gemini now.

Over time, big players who are willing to provide sustained investments and endure years-long losses do tend to prevail. Which ultimately is a good thing, if you ask me — since this observation alone ensures smaller players will also keep getting resources to keep trying.

On the growth mindset and its inseparable counterpart.

This is a long train of thought. I find it interesting not just for the destination but for the journey.

Many cultures, including the one I was raised in, believe money is earned through work — presumably hard work. Seen that way, the “career” of a car salesperson or, God forbid, a futures or derivatives trader is treated as selling one’s soul.

My counterpoint: tech began as automation, and automation matters. If someone who can crunch numbers lifts national crop yields by 1%, that’s an enormous success — easily worth many full-time farm workers’ salaries.

Another belief is that saving a dollar is as valuable as earning it. That makes sense through the lens of unit economics.

But for work to matter long term, it needs a scaling strategy. Saving 10% this year and another 10% next year has a hard upper bound. Capturing 1% of an emerging market, then 10% the next year while the market grows 1000%, is a different game.

Therefore, investments — especially in high tech — chase growth. Hence AI and robotics.

Here’s the thought: growth happens, but VCs often overfund the most promising players beyond reasonable forecasts. Call it hype or winner-takes-all logic, but valuations often run too high while revenues lag.

Not everywhere though. Conservative industries behave differently. Pitch a major energy company a credible plan to save a billion next year with new tech, and they will listen. And worldwide, many economies don’t fund moonshots without clear business plans. While those economies underperform the U.S. overall, people there often see their basic needs improve steadily.

We tend to be too optimistic. Hence debt cycles and crises — one reason Ray Dalio’s circa-2020 analyses feel increasingly persuasive.

Now the controversial part: what if actual growth sits far below projections? What if a large economy needs a strong signal to slow dangerous bets before the house of cards falls?

The practical lever is simple: raise the central bank rate!

Many believe expensive money suffocates the economy. The cure, they argue, is lower rates to create jobs and empower entrepreneurs.

But there’s a case for keeping rates high. In a high-and-stable-rate world, I can credibly argue that cost-effective infrastructure is a worthy investment, and improving developer velocity pays off.

Hype-driven companies rarely cut costs unless there’s obvious low-hanging fruit. I know folks who saved big companies millions and got no recognition. Most companies assume engineers should ship new features; in overheated, cheap-money markets, that’s a defensible position.

That's why I increasingly believe the best long-term cash-cow businesses aren’t in tech. They’re car washes and laundromats. Hype comes and goes, and markets can stay irrational longer than we can stay solvent. But basic human needs still have to be met.

I also recognize my privilege: I can work in one place and live in another — happy in fiscally conservative, high-rate Thailand while being paid by a hype-chasing startup in the money-cheap Bay Area.

Two takeaways.

First, there are real merits to high interest rates. In the short term, stocks may fall and unemployment may rise, but the economy needs periodic cooldowns. Preventing oscillations from wrecking the system is where “the invisible hand” fails.

Second, more people are learning to live well during cooldowns by staying fiscally conservative and steering clear of hype magnets. Those conservative products improve the baseline of life.

Perversely, I’d love to keep doing infrastructure architecture in an expensive-money world — perhaps for less money, but with better-aligned incentives to make tech more efficient.

For example, imagine a “boring company for food delivery.” Instead of gig drivers, build a metro for parcels that moves groceries automatically — cheaper than drones. A project like this is unlikely in a booming high-tech market, but very plausible in an over-cooled one, where cutting costs — not chasing hype — attracts capital.

An addendum to my previous post.

In the previous part: today’s [high tech] economy is over-leveraged and tilts toward ultra-high-risk moonshots instead of balancing star-shots with steady infrastructure gains.

The model I outlined — investors selling promises and inflating hot-air balloons instead of making tech better — describes the past ~20 years almost perfectly.

Why Docker then Kubernetes? Why are PHP and Node.js still common on the backend? Why is Postgres resurging despite being a [perfectly good, yet] single-node RDBMS?

Why are cloud providers raking in record profits?

In a sane world, within a few years, businesses like Amazon Web Services should see profits fall. The premium is too high, and with solid tech, running one’s own datacenter — and/or buying or renting hardware and going bare metal — would be cheaper.

The state of the art only makes sense in a world that aggressively over-promises returns.

And while I’m making this claim seriously, I’m also eyeing teams whose biz model is exactly this: better cloud solutions.

Imagine a Sr./Principal engineer saying: Let my team focus on infrastructure effectiveness, and in two quarters our hardware bill drops XX% — saving X million annually.

The response from most Sr. Directors and VPs will rightly be: Stop wasting time on spreadsheets and get back to moonshots. Even if you save seven- or eight-figures every year, the “real” value is shipping features — even if the net outcome is negative today and stays negative!

Right now, the path to “shareholder value” — for public firms and especially for venture-backed gigs — is to keep burning infra money as long as a plausible story says it will help the product line dominate. Then, someday, dictate “proper” prices. And finally turn a profit.

Hype used to be about network effects — see WhatsApp and Discord / Notion. Now, with AI, it’s "simpler": gigantic hardware spend, neatly captured by a particular cloud, is "the driver".

Conclusion: while the market is like this, working on improving infra primarily to cut costs — even dramatically — is absurd. No investor will fund it.

Two cases still make sense for me. One is Web3/DePin, where “average Joe/Jane” providers must be paid enough to keep supplying their computes as the nodes of the "decentralized cloud". The other is partnering with a hardware-first cloud provider with real skin in cost-effectiveness.

Both are about building for a post-crash, post-bubble tech world. Working on this today leaves money on the table — hype pays better. So I stay at the outskirts, backed by larger players hedging against this crash.

Shorting what’s “too big to fail” is the worst small-player gamble. Even if macro and unit economics say the current state shouldn’t last a few quarters, it might grind on for a decade — too many big players are invested into this hype.

Who knows — after the AI wave, maybe robotics. Then VR, or healthcare/genomics — again. Then quantum. Unironically, much of that could already be “solved” if even a sliver of AI money had been routed there.

My takeaway: for an engineer like me, step back from “market” forces and focus on hardcore tech, without expecting immediate returns.

Find a small, securely funded team, pick a problem that will still need experts in 5+ years, and dive deep — as I did with strongly consistent infra a decade ago and decentralized consensus last year.

The real motivation behind these posts: I think there are more people like me. Let’s find each other — conferences, online/offline, more meetups, more networks of folks happy to do this on weekends or as consultants.

Not to chase money or fame, but to keep our brains energized by worthy problems — hard intellectually and needed pragmatically.

Yesterday we had an offline conversation about the state of the art in moderation. It inevitably touched on censorship — but bear with me for a moment.

Three “disclosures” up front.

1) I used to moderate a fairly large forum. It taught me it’s hard but possible to uphold the principle: “I disapprove of what you say, but I’ll defend your right to say it.” I didn’t know that phrase back then — I was in my late teens and early twenties — but I learned the idea the hard way.

2) I’ve had run-ins with moderation elsewhere. The most notable was with Quora, where I tried to play by the rules, yet their “You have not done enough to have your content un-collapsed” remains, to me, a prime example of failed moderation and power abuse.

3) I’m not a free-speech absolutist, but I believe a system where platforms and users share responsibility for fairness is far healthier than… well, whatever we have today.

A decade ago, I used to argue it was a shame we couldn’t encode, in strict transparent rules, what qualifies as valid content. If only we had a “black box” — open-source, or at least with published binaries and signatures — hosted by the platform but fully auditable.

Here’s how it could work: before clicking “Publish,” your post gets flagged green, yellow, or red.

∙ Green: Safe. Your content complies with the rules and stays up — unless policies change.
∙ Yellow: Borderline. Goes into a transparent manual review process, ideally open, precedent-based, maybe even jury-driven.
∙ Red: Post at your own risk. The platform reserves the right to remove or hide it.

That’s the dream. And here’s the twist: we already have the tech. It’s called LLMs.

A platform built on this wouldn’t be hard to design:

∙ The moderation models are published regularly.
∙ They’re hosted by the platform, and responses are cryptographically signed.
∙ Anyone can request a signed statement: “This content passed under policy version N.”
∙ The platform must serve those requests — no tricks.
∙ If they delete content flagged green, they owe the author $1,000 per hour it’s down. Do your job if you host content.
∙ Full API access can be <$10/month, meaning feed ranking happens locally in offline clients.
∙ This kills shadowbanning: if User A follows User B, User A always sees 100% of User B’s posts — as long as they pass the model.
∙ PGP signatures + Merkle proofs ensure the platform can’t alter or reorder posts.
∙ Legal takedowns link directly to government request or court case IDs, with appeals supported.
∙ An “internal court” handles edge cases like doxxing.
∙ Moderators are accountable — community-elected mods can have their “stakes” slashed if they abuse power. They can also grow their reputation over time, among different cohorts of users — what's not to like in this, when filtering and ranking is performed on the client side?

Some would call this censorship. I don’t. It’s the opposite: leveling the playing field between platforms and users.

Users can still delete their own content, but its hash stays on a public ledger. That way, speculation like “they removed X” vs. “they removed Y” is instantly provable. And if content violates the law — doxxing, threats, etc. — the platform cooperates with law enforcement transparently.

I honestly can’t see why anyone would want less transparency than this.

Ultimately, we all want a shot at finding The Truth. Facts can be false or fabricated, but meta-facts such as:

∙ “User U published X.”
∙ “Platform removed X.”
∙ “X was restored, and U was compensated.”

… should always be preserved.

And one more thing: if User U2 follows U, there should be no scenario where the platform hides U’s content from U2. None. Run feed ranking locally; keep the servers blind to what you see.

I’ve never heard anyone sincerely say, “Yes, I want the platform to decide I’ll never see what someone I follow posted yesterday.” Not once.

Tell me where I’m wrong. From what I see, everyone wants at least this level of transparency and content protection. Some want more — but I’ve yet to meet a single reasonable adult who wants less.