hnaccount_rng

Do you happen to know how they compare to Trumpf the German laser manufacturer?

I would have formulated it less dismissively than my sibling but yes. There are ~no scenarios, where the Cloud act is not the more straightforward solution. Even factoring RSA-1024 will take non-zero compute time. And that can be spend more productively

You are right that there are better options and there is no excuse not to choose something better today. But that’s not what GP is arguing! It’s just that: For any given interest level there is no way that expending the resources to attack the RSA key is valuable! You are either willing to expend resources then attacking the owner is far cheaper or not then you are also not investing weeks-to-month of cluster time.

And from that perspective RSA-1024 is still perfectly adequate

Because there is nothing not already automated here. RSA will either be cracked because some assumption about prime numbers doesn’t hold or because we throw more and more brute force on it.

And if it were the former it wouldn’t be the “next” in line that would be cracked…

Sure, and in a zero marginal cost world that would be fine and dandy. In practice, it would mean significantly increased development and documentation costs for a low percentage increase in sold units.

Selling spare parts was worth when things broke often and labor was cheap. We don’t live in that world anymore. That means a broken device will no linger get fixed. But on the plus side there are far fewer broken devices. I’m pretty sure that has the same problem shape as “there is no glory in prevention”

You have trade offs though. Both with added effort during the development but also just physical limits with the amount of additional losses you have over clicked vs soldered connections

I think the term information is misleading here. In any encrypted message there is the original information, it must be because we are able to decrypt it. So from an information theory point of view that the server can "distinguished my ciphertext enough to produce a coherent reply" is always possible. The whole point of encryption is that we have Math that allows the identification of that information only under presence of the (or at least some kind of) secret.

The price that HE is paying is essentially "have to run every possible contributing computation", which means it has to scale extremely badly for anything real world complex

You are aware, that the normal Balkonkraftwerk is plugged in with a standard Schuko plug? This _only_ works because there is no licensed electrician involved. The personal cost would add at least 50% of the total expenditure and it is made safe by reducing the safety margin by the up to 800W that the Balkonkraftwerk supplies on that circuit

They need the grid frequency to generate AC. Those inverters are only allowed to be grid-following (at least for the plug-in part. I think there are systems that provide a -to-load socket that can be a PSU)

That’s largely correct but misses the point that the battery+pv system is legally only allowed to feed in 800W [1]. On our circuits you can draw ~3.5kW without tripping the breaker. This brings you to 4.3kW max load, which - as far as I understand - is still what all the circuitry must be rated for anyways.

The downside is that we have three phase power supply anywhere halfway recent. And that only gives you a single phase. In _most_ cases that’s fine, because you get a net-sum of all three phases at the meter, but technically you are only drawing/supplementing one out of three phases. In practice this likely doesn’t matter since different house put theirs on different phases and it all washes out.

[1] the DC side can have up to 2kW of panels. But that’s not visible to the grid

You are of course somewhat correct. But - I think - only technically. In practice what happens is that the “maintenance”[1] costs on solar production is extremely limited and should (that’s not yet possible to reliably say) be much lower than the rest of the grid.

If that’s the case, then we really don’t have to care about usage amounts _during the surplus time_. That in turn will enable completely different patterns. I think the correct analogy is the movement from candles/gas lights to electric lights. Before electricity was widely available nearly everything in life had to be done during daylight hours. Today we don’t even think twice about whether a meal can be cooked “after dark” or to move to a different location or ...

And yes that means I say: We will provide electric power for free [2], just like we do with street lighting

[1] let’s just call all replacements of cells and inverters maintenance for this discussion here [2] at the point of consumption. But that happens if the cost is smaller than the implementation of the sales system

But it’s also environmentally better for you to take the resistive heating thing. As long as you never need to heat it up outside of the noon-window that’s strictly positive. Because the “additional” solar panels will be necessary anyways to cover the night/late evening usage. The optimal buildout will always have superfluous energy at noon. That’s fine. We just need to get over the whole “energy costs anything” thing. That’s only true if you need to spend fuel to generate it

But transit only solves your problem in cities like London. Some people - for some reason I’m still not entirely clear on - seem to like this. But other people - so far the majority - don’t. And for those, self-driving cars solve the transit problem. That’s valuable. And you only need to beat unit economics of taxis. So there is a significant margin to capture

Isn't the US example: Medicare/Medicaid? It has far less costs (i.e. more of the money spend ends up with patient care) and has a reasonably good reputation?

But that is leas fundamental then you make it sound. LLMs work well with human language because that’s all they are trained on. So what else _could_ an ideal language possible look like?

On the other hand: the usefulness of LLMs will always be gated by their interface to the human world. So even if their internal communication might be superseded at some point. Their contact surface can only evolve if their partners/subjects/masters can interface

Depends on the future expected value that they assign to you. But.. yes

I love that example. It’s a basic exercise in “for how little money can you break any amount of trust”. Not sure how they could avoid that (besides being competent in the first place..)

I think we are pretty much past the "LLMs are useless" phase, right? But I think "super-charged search engine" is a reasonably well fitting description. Like a search engine, it provides its user with information. Yes, it is (in a crude simplified description) better at that. Both in terms of completeness (you get a more "thoughtful" follow up) as well as in finding what you are looking for when you are not yet speaking the language.

But that's not what OP was contesting. The statement "$LLM is _doing_ $STUFF in the real world" is far less correct than the characterisation as "super-charged search engine". Because - at least as far as I'm aware - every real-world interaction had required consent from humans. This story including

If you are willing to ignore the mechanics, then the result rounds to: The operator will receive the auctioned value of x £/MWh for each MWh delivered to the grid

If you are interested in the mechanics, then _I think_ (i.e. not first hand knowledge) the operator will join each auction (once per day for each 15 min interval of the next day) and offer his energy amount there. He will then receive the integral of (auction price) times (volume) from “the grid”. If that is too little money he then goes to the government and asks for a top up to (contract price) times (volume). If that was too much he has to pay the government the difference.

But again. That is purely mechanical. The end effect of the contracts is you will always receive say 80£ per MWh delivered. Independent of the MWh was worth 500£ in that 15 min interval or -50£. It’s “just” a risk transfer

Ah I’m seeing a possible confusion. There are two different auctions in the description. The first one is for the CfD price and happens (for each project) once. The second one is the daily-price-discovery one and happens daily

That’s not necessarily true. In general a single windmill is more efficient at pulling energy out of the wind than two are. And the marginal costs of windmills are not zero. I.e. their maintenance cadence (also) scales with active hours. So it might be a “at this price-wind point it’s not profitable for us to run a second mill”

Think of it this way: as a windfarm operator you know your costs and you know your expected amount of energy produced. But you don’t know the precise timing and therefore the market value at generation time.

From the first two you can calculate what you need in terms of £/MWh (include whatever profit you want in there). Now you can go to the government and bid that price in the auction. If you win, you have a safe profit and all risk (and upside potential) now lies with the government. As GP said, in the case of 2022 you would have lost out on revenue. But that’s the price foe guaranteed margins

The CfD part is a technical detail. It ~ doesn’t matter whether you first sell the energy and then go to the government for reimbursement. Or whether you sell the energy to the government which then handles the follow up sale.

What I’m not sufficiently familiar with is whether you _have_ to go to such an auction (i.e. whether the auction also is the mechanism of capacity planning) or whether you are free to bypass this system and just hook up your wind park and carry the risk yourself. But functionally this is an insurance scheme for profits, with a market based pricing system

Why wouldn't a few-occupancy sleeper bus work? I'd agree that this is a bit beyond "normal cars are autonomous". But I don't think the driving part will be particularly special to this application. It's just the form factor that changes.

Whether it will be possible to be cheap enough operationally is an interesting question though. The price of an (autonomous) taxi is (probably?) largely lower bounded by cost to build the system divided by the number of times it's used. And that means the denominator largely scales inversely with trip length. So it might still be too expensive to offer hotel-price level fares for night-long drives

I mean in the case in Lebanon they knew. They sold those pagers to Hezbollah.

While you are correct, for any user this is completely irrelevant, right? I have the choice between picking up an MBA or a different laptop. One comes with a reasonable expectation of battery life. The other is a gamble

My initial reaction was: Oh no, who would have thought? But then... 21% is almost shockingly low. Especially given that there are almost certainly some false positive, given that this number originates with a company selling "detecting AI generated text"

Russia basically cancelled their support for the S-300 Turkey bought from them. A lot of the tourists in Turkey were from Russia, that basically stopped. A lot of collaboration projects (IIRC mostly energy) got cancelled/suffered unexplained delays.

Erdogan send a letter of apology to Putin and did a bit of grovelling to alleviate some of that

Most importantly of all: Turkey was left alone by NATO. Yes, there was no great confrontation. But that doesn't mean there is no price to pay. Real life isn't a video game. These kinds of low-level confrontations do not result in large scale war, unless at least one side wants that war. Another example of those is the Kashmir region. There it happens regularly that Indian and Chinese troops fight. That has lead to both sides agreeing not to have those troops have rifles only clubs, but not to a large scale war

But neither was it only one incursion into Turkish airspace, not did Turkey not pay a price for it

But you don't need 12kts, right? The ocean logistics is ~only about costs, as evidenced by the reduction in travel speed. That being said, cost is currently not dominated by fuel cost either. That means unless you reduce crewing requirements or build cost, there is probably not much savings that will pencil out. Plus you'd need more ships and the physical capacity to build more is limited

Yes, and the Cloud Act pretty much forces upper management to ensure that there is always a US IT guy that can be compelled to implement the wishes of The US Federal Government, as the penalties apply to executives of US companies, too.

We can quibble about whether the term "threaten", which implies some moral wrong doing, is correct though. It's a law with defined criminal penalties. That's how criminal law works

In addition to my sibling comment: The cost of the panels is a rather small fraction of the total cost of a typical installation. Most of that cost ist labor, some regulatory requirements and the inverter. Whether you pay a factor of 2 for the panels or not typically doesn't matter. In other words: Reusing used panels will only ever be able to safe you a minuscule amount.