Posts by DavCrav

"Personally I like the Irish model, i.e. keep corporate taxes low so that companies benefit from setting up local operations with high-value jobs - the kind of jobs that attract high income taxes. Then the money goes into the local economy via payroll taxes which are very hard for anyone to get around. Company profits are very much less than what is paid in staff salaries, so the tax take on a % of profits is less than a % of staffing costs."

I downvoted you because you are essentially saying

"Personally I like the parasite model. Drive direct taxxes to companies as low as possible to they move to low-tax jurisdictions, taking both jobs and tax money from other countries."

Of course, that's a stupid argument because it leads to an arms race and then the only winners are corporations and their shareholders. Everybody else loses. Fuck The Other Guy is a game that works if you only play once, but in the long run it doesn't. The Irish model is despised by all of the countries that it's fucked over.

Re: just a reminder

"If you raise their taxes, they will raise their prices to cover."

No. They will hopefully lose market share to local companies who have to pay their fucking tax.

Re: Damage the hard drive?

"It wouldn't cause physical damage, but it would kill the windows install."

Sure. But since there isn't an uninstall feature for IE, people weren't 'trying to uninstall it', they were 'trying to delete it'. Start deleting random directories that store programs (even ones that do have uininstall features) and see where that gets you on Windows.

Re: An Attack or a Screwup?

"Nope. They're out to batter an Australian brewery into submission???"

They are also attacking all of the things you suggest. It's total cyber-war, by China, likely as punishment for Australian 'transgressions' regarding the international investigation into COVID.

Re: ...unless someone sneezes or coughs directly on you...

"Agreed that wearing a mask while driving is generally dumb, unless there are others in the car who need to be isolated, for instance driving someone to a COVID test site."

I've driven with a mask on. I had been out, somewhere potentially infected (red zone) and I didn't have anywhere to wash my hands and put the (washable) mask somewhere safe, so I wore it in the car on the way back. Then when I entered my house, went into the utility room (yellow zone) and washed and doffed the mask.

Unless you have a lot of equipment in your car to disinfect and store reuseable masks, it's best to treat it as a red zone for PPE reasons.

Re: What this is really going to be used for

"Mostly from care homes and elderly or people who already have chronic illnesses."

Sure. And where did they catch it from, eh?

"I'd say more pure pig headed stubbornness in the face of change, which is a longstanding British tradition."

It's partially stubbornness, but it's also familiarity. I know what a decent fuel economy is for a car in miles per gallon, not in litres per 100km. So I buy my petrol in litres and compute fuel efficiency in miles per gallon.

"According to the official statistics, if you do not have an underlying health condition, you stand little to no chance of dying if you are under 50, slightly increased risk between 50 and 65 but a high risk in the 70+ group... There will be out-lyers but, in terms of risk, you stand more chance of dying in an accident if you are under 50."

This is, unfortunately, not true. It's playing around with statistics to obtain an answer, but it's not the right one. Left unchecked, essentially all people will catch COVID-19. Thus the correct figure is to take the mortality rate from those who have it, not from the whole population. Doing so yields a mortality rate for healthy individuals of the order of magnitude of 0.1%. That is way higher than your chance of dying in an accident. A mortality rate of 0.1% for the roughly 40m healthy under 50s would be about 40k dead.

By comparison, the all-causes mortality rate for 25-34 year olds in the UK is about 1 in 2000, so 0.05%. COVID-19 isn't the plague, but the fact that it infects many many people very quickly means that lots of people die, even with a very low mortality rate.

Re: The household effect

"And this 60% thing is just plain bullshit. Any bit helps and the more that use it the more will then join. It will start slowly but will keep compounding."

No it won't. See my post earlier about percolation. It's not just the app, of course, self-isolation, contact-tracing, social distancing, etc., all play a role. But they need to add up to effectively 60% (or whatever the precise value is) to prevent a runaway infection.

"They don't even tell the truth about those few that have actually been tested, count multiple tests on one individual as 'tests'."

That's nothing. In Florida they count the number of infections as the number of positive samples. So if you test positive four times, you appear four times in the statistics.

You can only die once though, so it makes their mortality figures look a bit better.

"Actually, percolation collapses if the individual "cells" *resist* percolation with probability p > p_c."

I said:

"For a simplistic version of this, we will assume that the app works perfectly to stop spread, and that all people infect their four neighbours."

I was giving that as an explanation as to why there are counterintuitive sharp boundaries in these sorts of problems. The same applies with assuring randomness in shuffling, and other mixing phenomena.

I think the model is not applicable because it assumes that all humans are standing on a square grid, rather than the app doesn't work very well.

Re: That's the plan

"IIRC if you're contact traced, you'll be offered a test. But. It doesn't get you out of jail if it's negative. You still have to do the quarantine. I guess they're wary of false negatives, though that doesn't exactly fill me full of confidence if the Test part of Test, Track and Trace."

I just read this, and you are correct. The thinking is not so much false negative (although that happens) as delayed positive. At the time of testing you aren't shedding virus, so you will test negative. Then you later go on to shed it far and wide.

I am unsure whether any other countries are trying this approach.

"You should use the app, no matter how low the infection rate is."

It's one of these game theory type problems, a version of the tragedy of the commons. Unless 60% have the app, there's not much point in you downloading it yourself. You have to each take a collective responsibility. Given the ravers/shoppers/protestors/Cummings/beachgoers/general dickheads, I cannot imagine we'll get anywhere near 60%.

"The number "60%" has been spread around like a virus, but it's not true. _Any_ amount of uptake will help. The more the better obviously, but any uptake helps."

Quick introduction to percolation theory.

Imagine an infinite square grid, with a single person on each intersection of the grid. They know each of the four neighbours. They have the app with probability p, between 0 and 1. For a simplistic version of this, we will assume that the app works perfectly to stop spread, and that all people infect their four neighbours. Thus if we place a few bits of virus down, we will expect to get bubbles of infection appear, and will be stopped if they are surrounded by people with the app.

The question is, is there an infinite lump of infected people, all connected? Notice that if I let finitely many people download the app, this does not change whether there is an infinite lump of infected people or not! Thus a theorem known as Kolmogorov's Zero-One law comes in. It says that for p=0 (no app) the answer is obviously yes, and for p=1 (everybody has the app) the answer is obviously no.

Therefore, and here is the weird thing, there exists some probability, the critical probability, say p_c, such that if p<p_c then the infection spreads uncontrollably through much of the population, and if p>p_c then it always collapses. For the model I gave, the percolation threshold is about 0.592746.

This is the infinite case, but it works pretty well for the finite case as well, as 70 million is essentially infinite for this type of model.

What the critical threshold is depends on the model used, but there is a critical threshold for blocking percolation through a network, below which it is essentially useless, and above which it is great.

Re: Unfair

"That is most unfair and a rather bitchy thing to write. Dido Harding is, among other things, a remarkably talented jockey and owns Cool Dawn - a Cheltenham Gold Cup winner. Fine: a complete disaster when it comes to business but lay off the things that she does really well."

I'm a bit confused. The author was criticizing her judgment in letting the meet go ahead, not her ability in the saddle.

Re: Tow the party line

Must. Resist. Urge. To. Correct. Spelling. Of. Toe.

No, it's no good.

"I ordered a test on Monday (as part of a study, I've not got symptoms, just hay fever).

It was delivered on Tuesday.

I did it on Tuesday night and a courier collected it on Wednesday morning.

Still not had a result."

On the other hand: My son was sent home from nursery with a fever on Tuesday evening. Signed up for a drive-in test on Tuesday night. Turned up 8:30 am Wednesday morning, one other car in the whole centre. Out by 9am Wednesday. Three negative test results e-mailed Thursday 1pm.

"In one week? Highly unlikely. I’d estimate 2 weeks iOS, 2 weeks android, two weeks server software to make sure any idiot can’t enter themselves as “infected”, two weeks QA. So 4 weeks total."

Hmm. I got 2+2+2+2=8, not 4. (He said he was on his own, so unless he is able to do three things at once, it'll be six weeks developing.)

And two weeks' QA is a joke. You need to test it in the real world, so a full test like the Isle of Wight. That has to last at least a month just to get enough installs and check it's working.

Re: Have a go at the policy...

"Fake news. That is a misquote due to bad English translation. It was 'brioche'."

More Fake News. The quotation was around for decades before Antoinette's birth.

"Those key workers never got a lock down anyway so that point is moot,"

Not it isn't, it's explicitly not moot. Why didn't they have a lockdown? Because everyone would starve. But how do you pay them to farm when nobody else is working? Money isn't some magical nonsense, it represents, at a basic level, the labour required to produce goods and services. It's more complicated than that, of course, but fundamentally, no work = no food.

"The hotels option is perfect however 14 days worth of travellers would fill those hotels 10 times over."

It would at normal levels, but not at the current nobody travels levels. And I would also mandate no non-essential travel right now. It's a health emergency, no you can't go to fucking Benidorm.

"Got ya, money more important than lives."

Oh for the lover of God. It is not the safe thing to do to force everyone to stay at home. We all starve. So we let some workers out like farmers, supermarket staff, delivery drivers, out to work. But how are we able to afford the food? If it's grown in the UK then we can essentially force the farmers to growe it for free (pay them money, then debase the currency). But that won't cut ice with foreigners, so since about half of our food is imported, it would be a crash diet.

Lockdown kills. The virus kills. Deciding which to do and for how long is a balancing act, and the extremes of 'no change to anything' and 'shut down life' will kill the most people.

"Why not put coaches on with the driver isolated on it? I'll tell you why because it costs money and money is more important than you or I."

I can think of a couple of other reasons why we won't do that.

1) Where are you going to get all the coaches from to individually drop people off at home?

2) Logistically, this would be a complete nightmare.

I would just mandate quarantine near the airport. We have hotels at airports anyway.

Re: This!

"So, write it in C, using a hand-tweeked threaded algorithm, and inline assembly for the innermost parts. That'll do nicely!"

As I mentioned in a previous comment, C code, hand optimized, is about 100x or more too slow compared with a big brain (not mine) and assembler code that is optimized to hell, using each level of cache perfectly and only calling from RAM as and when needed.

Re: So what's the point other that Python & Java suck at calculations?

A very clever guy called Richard parker has spent a significant amount of time trying to optimize a matrix muliplication algorithm, but over finite fields rather than using doubles. (Think {0,1} if you don't know what a finite field is.)

He has managed to improve the current best algorithm, as implemented, by orders of magnitude. This is used for multiplying matrices in the hundreds of thousands of dimensions, rather than a few thousand.

The first thing is you don't need to do as many rounds of Strassen as you might expect to do, only half a dozen at most. And then the problem becomes chopping your matrices up into the correct sized blocks so you work just inside each level of cache. You really have to worry about the distance between the processor and the memory when you are doing things like this, and trying to optimize it. But you also need code that works on lots of different computers. Thus he has to make conservative estimates on the amount of L1/2/3 cache, etc. so that it always works.

He has to work in assembler, because he can't trust even a C compiler not to stuff up his code.

It's fascinating stuff, but a little too close to the coalface for my taste.