The best kittens, technology, and video games blog in the world.

Tuesday, March 29, 2022

Russia is Doomed

I drafted this post last year, with this exact title, but I ended up writing the 100 Programming Languages Speedrun instead, and didn't have time to finish it. I'm feeling a bit silly now. Anyway, nothing here is in any way related to the Russian invasion of Ukraine, so just imagine that this was posted before all that.

Russia is a regional power like Italy, Turkey, Brazil, or Saudi Arabia, LARPing as a superpower. It's baffling that people actually buy that, and think that Russia is some kind of a global superpower like US or China.

If you look at GDP or population, Russia is really unimportant. Projected population growth for Russia looks even more miserable than for the West, and their economic growth potential is very poor if you exclude two sectors.

Now Russia has two important things going for it, but here's why both will end soon.

Fossil fuel exports

Russia's biggest strength is its fossil fuel exports. If you look at export statistics, they constitute about 50% of all of Russia's exports. Exact number varies year to year based on their prices. Russia has been called "gas station masquerading as a country" for a good reason.

This is not sustainable. I despise virtue signaling over climate, but regardless of that, renewables technology is ceaselessly advancing. We already reached the point where renewables, with reasonable subsidies, are almost competitive with fossil fuels for electricity generation during good parts of the day. The main problems are intermittency of renewables, use of fossil fuels for transportation, and cost of switching all the existing infrastructure.

The first two can be solved by cheap batteries, and battery cost fell about 90% over the last decade, with little reason that this will stop. And given enough time and money, existing infrastructure will get replaced.

Full replacement of fossil fuels will take many decades, but this is not a growth sector. And Russia has nothing to replace it with. Other raw natural resources bring a small fraction of that money.

Green useful idiot

One thing Russia succeeded in doing, is funding "environmentalist" group that oppose European energy independence. Europe could have been far closer to energy independence already if it proceeded with fracking, nuclear, clean coal, and other technologies, just as US have done. It took US twenty years from being in just as deep shit as Europe to being a net energy exporter. All it took was political will.

However "environmentalist" groups, some receiving substantial Russian funding, others being Putin's useful idiots, managed to prevent European energy independence, and made Europe more dependent on imports from Russia for the time being.

Also did I mention recently that this guy should hang from a lamp post for treason, and it's a disgrace to his whole country that nobody hanged him from a lamp post or other suitable object yet?

Anyway, while this has been extremely successful Russian subversion of democratic countries, in the long term it won't matter, as it will be possible for just about any country to just buy enough batteries, solar panels, and EVs to gain energy independence.

Weapon exports

The second thing Russia has going for them are their exports of military equipment.

Russia is targetting mid-tier market. Their equipment is much worse than top-tier exports from the US, but it's a lot cheaper, and still a lot better than low-tier weaponry poorer countries can manufacture locally.

This is going to unravel really quickly, much faster than energy exports. The reason is China. China has been trying to build weapon export industry, and there's nothing that could possibly stop them from succeeding.

Chinese industry outproduces Russian industry 20:1 and the gap is growing every year, China has amazing track record at going from low-tier to mid-tier in any key industry it wants. That's what happened to far more competitive low-tier and mid-tier smarphone market over just a few years. The high end is safe - be that Apple and Samsung or US and other NATO military equipment. And countries can do their own low-tier basic stuff like rifles if they really want to. But China is amazing at going after the whole global mid-tier market.

I don't know why nobody talks about this, but it's pretty much guaranteed that the same will happen with Russian weapons exports. China will dominate this. And other mid-tier competitors like Turkey are also trying to break into this market, all at Russia's expense.

Why military exports matter?

Weapons exports are not a huge money maker for Russia. Overall global weapons trade is about $100bln a year, just 0.5% of $20,000bln a year global trade.

This is important to Russia for two reasons. First is that modern weapon development is horrendously expensive. Spending all that R&D money and your own military being the only buyer, translates to enormous unit cost. Even US can just barely get away with it, and non-exportable F-22 was in retrospect a huge mistake, compared with exportable F-35.

Russia is just far too small a country to even attempt creating any modern non-exportable weapon systems.

The second reason is that weapon exports give Russia a lot of political influence among world's poorer countries. It's hard to criticize what Russia is doing, after your minister of defense took Russian bribes to get them to win the contracts, and now your air force is now made of Russian planes, and Russia is the only source of spare parts.

This is not going to last. Oven the next few years Chinese offers will be better, cheaper, and for most countries less politically toxic, than Russia's.

There will be some markets left, like India might be unwilling to buy from China, and too corrupt to buy from the West (which is far less willing to engage in outright corruption necessary to win military contracts), and too poor for high-end gear anyway, so they might still buy worse Russian jets over better Chinese ones. But overall, this part of Russian economy will collapse even faster than their energy exports.

Postscript

As it turns out, I was late with this post. Europe is now scrambling to reduce their exposure to Russian fossil fuels, and Russian weaponry turned out to be a lot worse than advertised, so both sectors might collapse a lot sooner than I thought.

By my original guesstimated times for these, Russian energy sector would go into serious decline in 2030-2040 kind of range, but their weapon exports likely as soon as 2025-2030, while Chinese weapon exports would skyrocket just like their smartphone exports did.

People are now willing to talk about Russian fossil fuels, so that take isn't all that hot anymore. However, I haven't heard a single other person even entertaining the second point of the inevitably coming Chinese (and surprisingly Turkish) competition crushing Russian weapons exports.

Wednesday, March 23, 2022

Pandemic Retrospective

There won't be anything too controversial here, just a bit of big picture view.

Basics

Covid-19 is not over, and there's no indication that it will ever be eradicated, there are just a lot more important things going on right now.

So far the only established human disease we managed to eradicate was smallpox. Polio eradication efforts have largely stalled, with little progress since year 2000. I still have some hopes that we'll win over polio, but the chance of total elimination of covid-19 in any realistic timeframe is essentially zero.

The most outcome is that covid will be just another seasonal disease like the flu, with us for the long run. This is mostly fine, as going forward, most people will get exposed to mild infection when they're young, and by the time they're old and vulnerable, they'll have some degree of immunity, from some combination of previous infections and vaccines.

Age is the main risk factor, far greater than anything else. Every extra 20 years means 10x higher death rate, or in other words every additional year of your life is extra 12% risk of death if infected.

This is a good way to put things in perspective. For example according to the few RCTs we have, masking (with a real medical mask) may offer about 10%-20% reduction in infection chance, with no reduction in severity of symptoms in case of infection. So you're about as vulnerable masked in 2022 as you'd be unmasked in 2021, just because you aged enough to counter all that benefit. And "masking" with non-medical mask doesn't seem to do anything at all.

Vaccination with regular boosters may reduce risk of death by about 80%-ish, equivalent to being about 15 years younger. So a fully vaccinated and boosted 40 year old is at about as much a risk as completely unvaccinated 25 year old. There doesn't seem to be much reduction in chance of getting infection in the first place, mostly reduced severity.

It's often underestimated just how much covid is a single-risk-factor disease, that factor being age. For example obesity increases risk of covid death by 25%. So 40 year old obese person is at as much risk as 42 year old healthy weight person.

I think it's a useful visualization tools to translate all relative risk factors into equivalent age differences.

Risk of hospitalization and other serious health outcomes roughly follows risk of death, with similar risk increase for every additional year of age.

Anti-Covid Measures

There are three stages at which we could intervene to prevent covid deaths (and other severe outcomes):

  1. prevent pandemic from getting to your country in the first place
  2. reduce spread of infection once there's already community spread
  3. reduce chance of death for an infected person

A lot of measures have been taken, but only two have been very highly effective:

  • total border closures (step 1)
  • vaccinating old people (step 3)

Everything else did either nothing, or had minor effect, or was outright counterproductive. These include such measures as:

  • bank account closures
  • contract tracing
  • firing people for having different opinion
  • hand washing
  • ivermectin
  • lockdowns
  • mask mandates
  • masking
  • mass event closure
  • partial border closures
  • quarantines for travelers
  • riots
  • school closures
  • social media bans
  • stimulus checks
  • testing
  • vaccinating low risk people
  • vaccine mandates
  • voluntary masking
  • work from home
  • etc.

In the end, none of them had a big impact on the death rate. Notably almost all the measures attempted to target stage 2 (reduce infection rate once it's already established), and for covid that was simply the least effective stage to target.

The most unexpected thing about covid is that stage 1 interventions (total border closures) were extremely effective even after covid already got established. The original Wuhan lab leak variant was just not all that infectious, so closing borders and preventing much more infectious subsequent variants from coming in was an extremely successful approach, even in absence of other interventions. Japan is a good example of this combination.

Cargo Culting

A huge problem with global pandemic response is that countries threw away all established science, and mindlessly copied what other countries were doing, even if it was completely ineffective. So instead of 200 different policies we could meaningfully compare, it's largely same ineffective policies tried over and over.

Knowing just country age structure, border closures, and vaccination rates among old people, we can fairly accurately estimate pandemic severity, throwing away all other information.

Arguably the main problem with global pandemic response was that everyone was cargo culting part of Chinese response that was pointless (lockdowns), and in a half-assed way that couldn't have possibly worked anyway; and they did not copy the part of Chinese response that actually worked (total border closures).

What Was Not Tried

In retrospect was can safely say that nothing targetting stage 2 would have worked. Any response strict enough to eradicate transmission after it was well established would just completely crash the economy, and with borders even partially open, covid would get back in anyway. And if you closed the borders, then you didn't need to be that strict in the first place (see Japan again).

At least none of such interventions would work if they targetted general population. We could have absolutely done crazy strict lockdowns of nursing homes specifically. About a third of all covid deaths in US was in long-term-care facilities, and these are not economically significant. We could have completely isolated them from the rest of the population (food delivery in full body ppe suit kind of isolation), and likely avoided 1/3 of all covid deaths. By moving more old and vulnerable people into similar protected environments, deaths could have been reduced even more.

Nobody really tried that, and there's been overwhelming lack of interest in any interventions that segregated people by risk level. If we did something more along the lines of:

  • no restrictions of any kind on under-30s
  • some modest risk reduction recommendations for 30-60s (no mass events, work from home recommendations, masking on public transport recommendation etc.)
  • total lockdown for 60+s

That would be far more effective and far less destructive than what actually happened. To repeat the basic fact, 20 years of age difference means 10x risk of death difference. There's no sane way to treat kids and elderly the same, and that's what all countries did.

More idiotically, students were often restricted far more than adults, just because schools have far too much power over students' personal lives. It was cruel, and it did nothing to help with the pandemic.

There's some stage 3 interventions that could have had a meaningful effect, but weren't attempted.

The most notable one that would be mass vitamin D supplementation, as pre-infection vitamin D levels correlate very strongly with risk of death, and vitamin D insufficiency has been widespread in most modern populations. For people with most severe deficiency, effect size looks about as big as vaccines, based on observational studies. It would be much smaller averaged over entire population, as not everyone is as severely deficient. Unless you know that your vitamin D levels are adequate, you should absolutely get daily supplement, especially during winter, and not just for covid.

There's also some new antiviral drugs that could reduce risk of death. They happened late in the pandemic, but I guess if someone writes similar retrospective in 2025 they might get added to the list of interventions that made a big difference.

There's also an interesting idea for stage 1, to totally close not just countries but each city and region, and wait for covid to disappear there, before reopening only borders between covid-free areas. But this would only work if external country borders were closed as well, and it would be extremely disruptive. China sort of did that initially, but nobody else really tried that at all.

Who Predicted It Correctly?

Nobody at all.

Pre-pandemic expert consensus was very strongly against lockdowns, border closures, masking, and most other measures that ended up happening. Pretty much none of that was followed.

Epidemiologists' models were all universally worthless, and the whole field of epidemiology turned out to be about as reliable as social psychology.

Governments were both extremely incompetent and openly lying. The media were even dumber than usual. Social media companies were banning people for saying true things.

I don't think even any individuals got it right. As far as I know, nobody predicted either of these:

  • the initial Wuhan Lab variant was no big deal, the variants were a big deal, so total border closures would be super effective even after Wuhan lab leak variant was already there
  • vaccines would be developed very quickly, but they'd only offer big redution in severity and risk of death, and wouldn't meaningfully reduce infection rate, so mass vaccinating old people was really important, but any mass mandates are pointless

I don't think anything useful came out of prediction markets either.

Who Did Well?

The private sector did amazingly well. Rapid transformation of big parts of the economy from office-based to remote-first was amazing, and we should absolutely embrace the remote-first world, even if it made little difference for covid.

Similarly supermarkets, restaurants and so on switched very effectively from location-based services to delivery-based services. Restaurants have a lot more reason to exist than offices, but this shift is likely permanent, and a much bigger share of the economy will be online or delivery-based going forward.

Big Pharma did quite well. Multiple vaccines were developed in record time, of kind not used ever before, for kind of a virus that never had an effective vaccine before. Tests, antivirals, and so on were also developed very quickly.

What It All Means For The Next Pandemic?

Very little actually. Covid-19 had so many unusual features, there's no reason to expect the next one to be like that.

Wednesday, March 09, 2022

crystal-z3

A small announcement. Mostly as a way to play with Crystal, I created Crystal bindings for Z3 library, and I think it might be quite usable as a MVP.

Check included examples and specs for how to use it.

Monday, March 07, 2022

New series - Open Source Adventures

I started a new series last week, Open Source Adventures. It's not terribly focused, it will be some random unrelated Open Source things I wanted to do. The first mini-project is Z3 support for Crystal programming language, but I'll jump to some very different subjects.

Just like with previous two series, you can read it on dev.to or hashnode.

The code for the series, unless it has specific reason to go elsewhere, will go to this repo.