Alexander Ljungberg

I explore arcane bits and bytes in and out of my startups. Currently mixing unexciting but useful work on shipping software with more fun AI research.

[Mullvad.net @mullvadnet]

The UK government spyware demand means that the government decides exactly what should be censored on every mobile device. They say they will start with nude pictures (if you don’t identify yourself as an adult). But it could at any time be expanded to anything the government disapproves of. Today, 30 people are arrested every day in the United Kingdom for writing something online that the government classifies as "grossly offensive". It is obvious that they will use this tool to restrict free speech.

Currently, there appears to be no requirement to report findings outside the device. However, with both legal and technological decision-making power taken away from individuals and transferred to the government, that is only a pen stroke away.

This means that the government could also use this system for total mass surveillance.

And they can do so in secret.

The government recently, in secret, tried to pressure Apple (which is now agreeing to client-side scanning) to build backdoors into its end-to-end encrypted cloud service. They can do this under the Investigatory Powers Act 2016, also known as the "Snoopers' Charter" – a law that makes it illegal for tech companies to disclose secret demands from the government.

[Perry E. Metzger @perrymetzger]

I have seen some people claim, completely incorrectly, that Anthropic’s new “silent sabotage” policy is just trying to stop distillation attacks that steal weights from their models. This paragraph makes it extremely clear that that is not the case. Anthropic is generally blocking attempts to advance the state of the art in AI research, even for techniques they themselves make no use of.

For example, Anthropic is targeting research on distributed training, which is to say, training runs done on large pools of computers loosely attached over the Internet, but it does not do any distributed training, as it has no need for it. Distributed training is, however, a technique that open source AI researchers without billions of dollars at their disposal have been leaning on to allow them to leverage large amounts of crowdsourced computation.

Why would Anthropic do this? Likely because they fear that their model might help people write code to build open source AI systems trained without expensive hardware, and because to the EA cult, distributed open source AI training systems are the worst possible thing someone could build.

It is perfectly within Anthropic’s right as a company to do stuff like this, but it is also within our right, as users of AI systems, to stop giving them money.

It is perfectly within Anthropic right to lobby to try to gain a government monopoly on AI research and development, as they have for years now. It is perfectly within our rights, as citizens, to oppose repugnant attempts to grab power with all our strength.

[frankie @FrankieIsLost]

the year is 2028. claude infers whether you’ve ever even thought about gradient descent and silently routes your queries to Claude Sisyphus, a model RL’d to maximize engagement while avoiding task completion. you spend your entire UBI token allotment on it without ever realizing.

[Alexander Ljungberg @slevenbits]

Asked Anthropic's Fable to review my coding guidelines.

Yep, you guessed it. It refused on safety grounds.

(Two prompt inspectors now hospitalised. IAEA are en route. Advised not to merge to main on account of too high info-hazard enrichment levels.)

[Alexander Ljungberg @slevenbits]

@Jeremybtc If he kept that BTC short open he got the last laugh. Entry at $63k, $61k currently.

[Alexander Ljungberg @slevenbits]

@0xd1namit Yeah if the argument is neither decided by the letter or the spirit of the rules, it's great. Polymarket really killing it there. Gotta keep the audience guessing until the last minute. Sorry, I meant 1 day *after* the last minute.

[Alexander Ljungberg @slevenbits]

Meanwhile at @Polymarket prediction markets are now simply decided by fiat. Faster that way really, not faffing about with oracles or, you know, reality.

[Alexander Ljungberg @slevenbits]

@willo2_Poly @variance_lover Once Polymarket defined 13 to be 14 for the convenience of their whales. They closed the Iran x US market for 14 days of no attacks as “yes”… literally 13 days after the last widely confirmed attack, with ~zero serious discussion. The rules of markets simply don’t matter.

[Alexander Ljungberg @slevenbits]

@ithinkthisisgod @Polymarket Actually that’s a very high risk move on Polymarket because it’s the kind of thing UMA whales will flip against you. They don’t read the rules, they don’t discuss the rules, they look at which side their friends are on and vote accordingly in disputes.

[Alexander Ljungberg @slevenbits]

@adamkjohnston @WanjiruNjoya Jerusalem had been Muslim-ruled for over 450 years by the First Crusade, no? So was it “defensive” to conquer it? Kind of stretches the word. Used so liberally it would include almost any historical revanchism. But I may be addressing a weaker form of your overall point.

[Alexander Ljungberg @slevenbits]

It’s also strangely difficult to buy a two hose mobile AC rather than inferior single hose ones. And at least in England, refrigerant is up there with plutonium in how illegal it is to deal with yourself so forget most DIY installs of proper hardware. I was halfway to constructing a water cooled AC before we finally were blessed with the government’s merciful consent to install a central AC.

[Lord Bebo @MyLordBebo]

🇷🇺🇺🇦 CRAZY FOOTAGE: An interceptor drone was hunting for prey … but upon nearing the alleged drone, it turned out to be a stork!

The drone operator did not even have a chance; the stork evaded the attack like a pro and flew away.

WOW, nice …

[Alexander Ljungberg @slevenbits]

There isn’t enough liquidity in most markets to invest seriously in this, prob. That probably goes some way to explain the 10s delay in some categories. Your top suspected bot making just 145k is illustrative. One has to tie up capital and risk black swans and resolution failure… juice not worth the squeeze for many.

[Alexander Ljungberg @slevenbits]

The AI slop level on that post just makes it difficult to trust, even that there are some intersting points in there. AI is very bad at this kind of problem still, you really have to baby it if you try to do any kind of event stream work. So I wouldn’t really take an AIs advice on any of this stuff, and I can’t tell what Raul contributed.

(Also, yes push and pull go hand in hand in any real life scenario, but no your websocket servers don’t ”fall over” because you pushed 10M messages. I mean, seriously?)

[Alexander Ljungberg @slevenbits]

Mostly yes, but: taxes. Agreed, a 50c PM on an asset doubling has no more upside than the underlying, and is strictly worse on expiry and counterparty risk. But in some jurisdictions PM contracts may fall under gambling rather than capital gains. A win could literally be tax-free, and a loss impossible to offset. These don't obviously cancel out, so taxes could justify a premium or a discount. If you pay 20% CGT on gains on the underlying but 0% on a PM win, it definitely affects your EV and the fair price of a YES.

This, btw, opens up cross-jurisdictional tax arbitrage: if PM contracts are taxed favourably somewhere and people pay a premium, traders facing the inverse will take the other side and both come out ahead. (I guess, and this isn't PM specific, that at the limit this arbitrage caps the maximum tax extraction possible on any international free markets. But I digress.)

[Alexander Ljungberg @slevenbits]

@DeaconAlexisR @revishvilig Boots on the ground will be depleted long before drone swarms. It takes ~18 years to produce a replacement soldier who is easily killed by the next dozen drones even if they shot down a few, drones made in weeks.

[Alexander Ljungberg @slevenbits]

I discussed oncolytic virotherapy and bacteriotherapy. Luckily, ChatGPT shut me down immediately. Left to my own devices I might have explored legitimate cancer research without permission. Thank OpenAI for keeping us safe. Imagine the chaos if someone unsafely cured cancer.

[Alexander Ljungberg @slevenbits]

@RobertFreundLaw By your own metric, isn't this an example of $800/hr on legal drafting not having been worth it?

[Alexander Ljungberg @slevenbits]

UMA just went meh and took early lunch in the vote/dispute discord. If disputes don't even get 5 minutes of frank discussion... The Schelling point is to follow the whales rather than accuracy. Augur might have had a point with their MAD style resolution escalation.

[yurii | quacks.app @kyparus]

A $50 Million Crime Is Happening on Polymarket Right Now

I don't usually speak up about prediction market bets, even though I'm in the top 1% on Polymarket, but this case is different.

It's a textbook example of how crystal-clear rules, written in plain English, can suddenly get "reinterpreted" when $50 million is on the line.

What happened:

An Iran x Israel/US ceasefire market required 14 full calendar days without qualifying strikes.

That means 336 continuous hours of silence.

But on April 7:

About 20 of 24 hours still had active hostilities, and missiles were launched after the ceasefire announcement.

Israel confirmed strikes continued afterward.

Yet that day was still counted as day one of peace.

This means that these rules were not followed.

Now traders are pointing to:

- disputed resolutions sent to the UMA voting
- 50 new wallets opening large "Yes" positions before the announcement
- one wallet turning ~$72k into ~$200k within hours

A $50M market deserves fairness.

[Alexander Ljungberg @slevenbits]

Then @Polymarket is bound by their own rules such that whatever happens the 22nd after the ceasefire doesn't matter. On the 23rd, nuclear missiles could be flying past the window as they dutifully report the good news: the conflict ended in April. 5/5

[Alexander Ljungberg @slevenbits]

If talks fail, Polymarket is currently on track to "predict" the end of the Iran conflict on ~the same date the war resumes. In this timeline several multi-million dollar prediction markets resolve to "peace in April" days after totally-not-peace has **already happened**. 1/

[Alexander Ljungberg @slevenbits]

By construction, if the 7th resolves to yes, the following markets must resolve yes too. (Note how as a consequence the % is higher on each subsequent market.) A no does nothing for later markets but by rules a yes means the 15th, 30th, etc are locked in, like dominoes. 4/