sharpneli

Even though I couldn't make it myself by the usual orbit tools I also wouldn't be too surprised if it was possible to gather bunch of stuff with like < 100m/s delta v or whatnot.

KSP taught me that flying a rocket is "easy". But calculating the stuff is hard (well, computationally intensive. So it can be easy, but hard for humans in their head).

The rules needs to be written in a way that allow things, like integrated circuits, to exist but at the same time prevent intentional preventing of repairability.

Things like "Add lines 25a through 25c". Don't they have computers? Are the forms for Citizens like this too?

That's the reason why they introduced https://en.wikipedia.org/wiki/Bfloat16_floating-point_format

It's similar to normal half precision float but the exponent is the same as with 32bit float and mantissa has been sacrificed to get room for it.

Or is it so that if I criticize something that affects some group not specifically me it's actually punching down? It's not possible to criticize something that's being done to a group without it being a criticism of that group?

This is absolutely brilliant. PR departments must love this.

Not much. This comparison includes apparently also plugin hybrids (So BEV and PHEV combined) https://en.wikipedia.org/wiki/Electric_car_use_by_country

Out of all new cars 1.9% were plugins in US in 2020 18.1% were plugins in Finland. 74.7% in Norway.

And they were not full EV's over here. Only less than 2% is full EV here (which is still more than what's sold in US). The majority is basic family movers like Volkswagen Passats and whatnot that allow daily driving to work with electricity and then for summer cabin one can go with gas.

I do understand why consumers are not enthusiastic about them in US. As in some cases it can be cheaper to just go with gas than use electricity.

Over here its overwhelmingly cheaper to drive with electricity. As an example the RAV4 PHEV costs around 30-50% to drive using the EV mode instead of fueling using the average price of household electricity here.

> Sorry I mixed up PHEV and FCEV. Our dicussions is about FCEV but I wrote PHEV, that is just a hybrid with a plug.

Yeah FCEV's have a hard time going forwards. Unlike plug cars that can be charged at home with FCEV the full chicken and egg problem in refueling is there. Perhaps someday, but in general overcoming those problems is really really hard.

> Also, for PHEV the emissions are often understated. Yeah. Especially when there is no significant savings with charging.

As gas costs closer to $8 per gallon over here actually charging is far more common. It's so much cheaper to drive with the battery than by using gas.

It's basically all about economics. Make it cheaper and people will buy. Expensive status symbols only go so far.

EV’s have zero car tax and zero VAT on Norway. That makes them cheaper than equivalent gas cars

> Yes, and consumers so far have universally rejected PHEV unless massively insentivised to do so.

Same with EV’s as shown above. They’re only massively popular if the price is right. It all comes down to it.

In Finland PHEV sales are now higher than diesel sales iirc, which used to be super popular here. Due to how car tax works on models like Volvo the PHEV variant is cheaper than the gas variant, hence people select them. We don’t have no VAT like Norway has so BEV’s are still very expensive here.

Personally I do like emissions based taxation a lot. No need to complain about people not buying them as they will buy if the price is correct vs alternatives.

As a sidenote Toyota has announced an EV. Sure it’s bit after others but not by that much.

Also they’re increasing the amount of PHEV’s. As they already have a good hybrid tech one just needs to slap a bigger battery and it’s done. And as batteries weigh a lot the weight of PHEV with it’s ICE and whatnot is around the same as BEV. See Rav4 PHEV as a good example.

Right now energy density of them is enough for decent passenger car experience. But it’s not competitive in big trucks.

> Your statement is entirely unclear. Could you restate it? For Steelman purposes, I'll ignore the non-sequitur about cheating.

By cheating I meant making a sidechain that would become longer than the mainchain, thus enabling a double spending attack. If the amount of money to be made in double spending attack is higher than honest mining someone will make it, the only thing protecting against that is that the cost of mining must be large enough compared to value being transferred.

>I'm not sure what you mean. In general, energy costs seem to be dropping over time

Back when it was CPU mined. Before even the first GPU mining. It was just about proving that you have an actual HW dedicated into the task.

Exactly as mentioned in the quote that you said. He didn't have a clue that it would become dedicated HW that has only a single function, burning energy to calculate SHA256 and nothing more.

Also do note that the whole concept of non mining node was foreign. Instead of the original design of nodes that mine we're in situation where nodes don't mine and most mining happens in centralized farms.

They include air travel and whatnot. Basically the whole energy consumption of the company. Also including building stuff etc.

Visa doesn't share that in order to poke at Bitcoin. They share that because enviromentalism is hip nowadays. So they want to tell how they're improving their CO2 emissions as a company.

Yeah it _might_ grow in the future. But until it does it doesn't work as an argument.

Also in order to get into Visa levels of efficiency one would have to have (assuming the 741kWh in the main network) 500k TX per second in the lightning network with 0 energy cost. Visa handles around 1700 per second.

So basically if we'd have all the worlds economy in Lightning network we might get close to those numbers (assuming 8 billion people each doing one transaction per 4 hours, that's super high but ah well, whatever). And that's assuming the mining energy consumption doesn't change at all which it definitely would not do in case of the whole world relying on Bitcoin.

No information on actual transactions performed. But in practice the lightning network is tiny and stagnant.

If it would actually be used it would perhaps be more relevant, but it really doesn't seem to be. Except as an excuse for the bad efficiency.

And those limits come up to the theoretical maximum of was it 7 or so and in practice around 4

https://www.blockchain.com/charts/transactions-per-second

Also PoW is hilariously inefficient. Because the amount of energy one needs to spend is basically related to the amount of value one could grab by cheating, so if all economy would be based on Bitcoin we'd be talking about basically cooking the planet due to PoW.

When Satoshi originally devised PoW it was more about owning the hardware and using that as basis for voting. Energy costs were miniscule. He simply didn't foresee the, now obvious, result that it will became a function of burned energy rather than amount of hardware out there.

That's 350km/220 miles of range per charge. In total 1050km or 660 miles of driving.

So basically if you want to transfer money and the recipient is closer than 660 miles from you it's more efficient to just get a bunch of bills, drive there with a Tesla rather than send a Bitcoin transaction. Or alternatively 330 miles roundtrip. For each single transaction.

It is hilariously inefficient system. It is literally better to drive a full car and give things physically than use Bitcoin.

A small addenum: Visa apparently uses 146kWh for 100000 transactions. Coming at 1.46wH per transaction. That can drive a Tesla for 9 meters (0.4 seconds of driving at 80km/h) or alternatively keep a nice led lamp lit for an hour. And that includes everything the company does, not just the actual transaction but the upkeep of their offices etc divided by the amount of transactions they perform per year.

EDIT:

Apparently on 2020 it was 741kWh per transaction in practice and/or I made a tiny error in my calculation. Either way it's even better!

https://www.statista.com/statistics/881541/bitcoin-energy-co...

That simply increases the ranges so that it becomes a 3600 mile trip. Or 1800 mile roundtrip.

21% are long term homeless. And that's what people generally mean when they collegially use the term homeless. That's defined as homeless that has lasted for more than a year and has significant social or health component, such as substance abuse or mental illness. The thing about mental illness is that there is also treatment for them. The solution is different for them as for the substance abusers.

If you ever see a Finn begging it's pretty much always an alcoholic who wants more beer. That's because the mentally ill cannot really beg as they're either receiving treatment or if they have unfortunately slipped trough the cracks of the system they're also unable to beg for any extended period.

As for the scale of this problem. 2019 there were 961 persons listed in the long term homeless category. That's out of a population of 5.518 million in 2019. That's 0.017% of the whole population. Also, even out of them 584 live with someone, just not with an official address. They wouldn't be counted as homeless in US.

The actual amount of people in street on that category was 177. That's 0.003% of the population.

You do understand that everyone gets social security as money if they need. The problems happen if they are unable to use that money to pay the rent. And yes indeed the main reason is some sort of intoxicant use, as they rather get more stuff than use the money for rent.

Legal debts are also not a reason not to pay rent, as one is protected from repaying them when it's about essentials. So that doesn't count either. As long as you're able to push the pay button in your online bank you use the default system. Only when that's not possible do you fall into the provided housing system. And not surprisingly drug use is a major reason for not pushing that pay button but rather taking the money and using it elsewhere. Does that honestly surprise you? What else could it even be?

That's why we have the second option with food stamps and provided housing. It's not perfect as people elsewhere have stated the obvious "Hey want to buy 20e foodstamp for 10e?". But still they get it.

Sure it's not always perfect, errors in bureaucracy happens sometimes but getting social security is the default. If you accept it you get a house.

It has 200kWh battery so that should give even more juice to power a house.

I wouldn't be surprised if some would buy ridiculously oversized EVs just to alleviate that fear.

That's what makes the arguments from USA so amusing.

I mean they basically nicked, against British law, the designs how to mass manufacture textiles. Is US willing to repay, with interest, all the economic damage that did to British industry? Same with all the later patents. US basically started respecting patents of others when they got to the point that they had more of their own. So only when it was profitable.

It's basically "It's fine when I do it but now that I'm the big dominant power everyone else must stay in line and let me stay dominant"

Don't get me wrong. I'm happy that US did what it did. I'm also happy that the other countries did what they did, because in the end we all benefit more from the countries being more developed than being less developed by "following the rules". Once China gets bit more R&D of their own they will start following the rules simply because it's mostly reciprocal.

Now let’s say that someone would make a chip that would use only 10% of the power of the current ones. What do you think would happen: People would keep current hashrates and just consume 10% of energy.

Or: People would scramble to get these chips and keep on burning the same amount. As your rewards are basically due to your ratio of the total hashrate. Not any absolute amount.

The fact that it was said makes the amusement value even greater.

Having those is pretty orthogonal to actually getting results in science. No need to mix them in.

Universities should actually fund (so indirectly by the goverment from taxes) long term research. That’s kinda the whole point of Universities, in addition to education. Long term research that doesn’t have immediate short them commercial interest. Not make the scientists be basically startup founders that must constantly pitch to get money.

And the reason we want to support that is because on long term there often is very useful commercial benefits that then improve everyones lives.

I’m not complaining about the situation. Supply and demand determines IT wages quite well. For some reason US based devs are simply more in demand.

If Mac APIs actually work as advertised then the difference is that windows programs work weirdly (bad sync etc) or have had to spend tons of time making custom code to workaround the badness of their ”professional” audio API, costing way more.

And another example is Android. It’s definitely not placebo that it tends to have horrendous latency. Originally due to design and nowadays due to bad vendor implementations with only Pixel phones reaching iPhone levels of latency. We’re talking about latency in iPhones being only few ms and tens of ms or even more in average android phone. That does actually matter in music production.

I have a small story to tell about this. Windows WASAPI has a way for it to tell you when, supposedly, the endpoint device started the recoding of a buffer. This is important if one wants to have a sync with precision of sub millisecond range.

Well. It doesn't quite work like that in reality. I noticed the hard way that if one gets a 2ms long packet. Queries the high performance timestamp, then queries the packet recording time, then queries the timestamp again. The packet was supposedly recorded after the first timestamp but before the next. And as the packet was 2ms long and the previous timestamp was way way way less than 2ms from the query it means that it was impossible for it to start the capture at that point.

So instead of returning the start of the recording as per docs state WASAPI just returned the current high precision timestamp. It's basically you asking someone "Hey, this 10 hour video you gave me, when did you actually start recording it?" "Now. I started the recording right at this moment".

This story has a happy ending though. I grabbed portaudio, hacked it's windows low level implementation to return the actual timestamp based on the register of the soundcard that points to the FIFO buffer and managed to get sub ms precision.

So if macOS audio API's are as good compared to Windows (which is a dumpster fire unless one uses some custom stuff dependent on manufacturer) as iOS api's are compared to Android (where only Pixel phones are not dumpster fires) I can easily see why Macs are the devices of choice.