spkoc

Karma: 205

spkoc Oct 27, 2020, 11:16 PM
1 point
in reply to: Zvi’s comment on: Covid 10/22: Europe in Crisis
I think he just meant the curves would match, ie. referring to the peak from the previous wave: https://twitter.com/jeuasommenulle/status/1320682552257089538
Oh actually, I think he just means they’d get back to that peak level in 25 days, not that it would get better after that. I misunderstood what he was saying. I’ll correct my post above.

spkoc Oct 27, 2020, 4:09 PM
1 point
on: Covid 10/22: Europe in Crisis
As a follow up to my previous comment, here’s a really amazing Twitter thread breaking down the situation in France: https://twitter.com/jeuasommenulle/status/1320682084973858816 (threadreader version: https://threadreaderapp.com/thread/1320682084973858816.html)
The poster also tries to estimate R0 using hospital data(which should be more reliable than case data, since the Spring wave was so undertested). He finds a R0 of 1.2, which means a doubling every 20 days.
He estimates ICU usage levels will be as high as the spring peak in France within 25 days, if the trends stay correlated with the spring outbreak.
Lots more nice graphs in the thread too.

spkoc Oct 23, 2020, 12:57 PM
10 points
on: Covid 10/22: Europe in Crisis
https://ourworldindata.org/coronavirus-data-explorer?zoomToSelection=true&country=ROU~AUT~BEL~HRV~CYP~CZE~DNK~EST~FIN~FRA~DEU~GRC~HUN~IRL~ITA~LVA~LTU~MKD~MDA~NLD~NOR~POL~PRT~RUS~SRB~SVK~SVN~ESP~SWE~CHE~GBR~MCO~TUR®ion=World&casesMetric=true&interval=smoothed&perCapita=true&smoothing=7&pickerMetric=location&pickerSort=asc is the best site for COVID European COVID graphs as far as I know. The UK has excellent breakdowns(including by age group): https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/articles/coronaviruscovid19roundup/2020-03-26
What really matters is people’s actual behaviour, as you pointed out. In the US people are cautious because they don’t trust their government to have it under control. In the Nordic countries and Germany people are cautious because they treat government advice more seriously than us Eastern Europeans treat actual legislation. The lack of a federal fiscal response also hurt. The poor states in the East can’t afford generous furlough schemes(not that the furlough scheme seems to have helped the UK, France and Spain that much); France, Spain and Italy are dealing with significant debt burdens. Various pushes were made to get people back out and working and off furglough during late summer early autumn. It worked(another finger curls on the monkeys paw)!
The European story is complicated and regional, similar to the US. You got Germany, the Nordics and the Baltics doing very well in terms of case counts and (as far as I know) without insane restrictions. Then you have Eastern bloc places doing increasingly poorly, but still ok for now. Then you have the Czechs(positive rate >25%!, adding the same number of cases per day as Italy(10 million Czechs vs 60 million Italians)) in 2-weeks-til-zombie-apocalypse mode. Then you have whatever is happening in France and Spain.
Czech Republic issues, at least, seem to come from the governing party downplaying the virus a lot before their October elections. They promised they wouldn’t reinstate a lockdown no matter what. 2 weeks later they’re in lockdown. They’re numbers went nuts right around when the campaign was picking up too. People maybe actually believed their politicians’ promises that things would be OK if they relaxed. Rookie mistake, czechs, rookie mistake.
Belgium has been in political anarchy for years, the government has very little authority over the population and is indecisive and weak to boot(for an overview https://www.politico.eu/article/how-belgium-failed-its-second-corona-test/). They flip flopped on restrictions, people weren’t given clear instructions from government or non-governmental experts. Netherlands aren’t as governmentally incompetent but their country literally overlaps Belgium and their government wasn’t super restrictive either. In fairness to both countries, they are the political(Brussels) and trade(Rotterdam and Antwerp) centers of Europe. They had travel bans in place, but there still would have been a huge amount of commercial traffic that was exempt.
Also on September 23rd Belgium eased Coronavirus restrictions(https://www.garda.com/crisis24/news-alerts/382596/belgium-authorities-ease-covid-19-restrictions-september-23-update-23), literally the next day their exponential spike begins(https://ourworldindata.org/coronavirus-data-explorer?zoomToSelection=true&time=2020-03-01..latest&country=~BEL®ion=World&casesMetric=true&interval=smoothed&smoothing=7&pickerMetric=location&pickerSort=asc). The easing seems to have had a gigantic impact on people’s behaviour: https://ourworldindata.org/grapher/changes-visitors-covid?tab=chart&stackMode=absolute&time=earliest..latest&country=~BEL®ion=World Before the easing park and outdoors visits were 70% over baseline. 1(!) week later they were 10% over baseline. To be fair I’m not 100% sure that I understand the mobility numbers, so maybe it’s just a weird data quirk, tho the drop in park time and the jump in cases overlap almost perfectly.
Greece is doing great but Cyprus is doing bad. A bit of a mystery as to why. Cyprus was maybe more aggressive in courting tourists? They famously had a COVID insurance scheme. Looking at mobility data, the Greeks just seem to have gone outside and never gone back inside :D probably a good idea, maybe hard to replicate for those of us not living in Elysium, sadly https://ourworldindata.org/grapher/changes-visitors-covid?tab=chart&stackMode=absolute&time=earliest..latest&country=~GRC®ion=World
Other places in Eastern Europe are still quite poor and can’t cushion shutdowns with generous leave, so the population kinda had to be let back to work, bars and restaurants and tourism had to be opened up. I’m not sure how big the conspiracy aspect is. In Romania(where I’m from) my parents’ group of friends(all in their late 50s early 60s) are averagely open to conspiratorial stuff(I think almost all adults that grew up under communism are like that) and they take the virus very seriously. But again, they’re all still working, not really doing WFH, but they are masking, isolating from vulnerable relatives and each other.

spkoc Oct 17, 2020, 10:12 AM

3 points

Is the CFR going down because the virus is less deadly in general, or because the virus is infecting younger people? I haven’t been able to find a decent study on this.

In the UK(https://www.gov.uk/government/publications/demographic-data-for-coronavirus-testing-england-28-may-to-26-august/demographic-data-for-coronavirus-covid-19-testing-england-28-may-to-26-august#age-and-gender) the median age of people testing positive has gone down from 52 to 31:

Annex table 7: median age of people newly tested and newly testing positive for COVID-19 under pillars 1 and 2, England

Week	People tested	People testing positive
28/05/20 to 03/06/20	48	52
04/06/20 to 10/06/20	45	48
11/06/20 to 17/06/20	44	44
18/06/20 to 24/06/20	43	43
25/06/20 to 01/07/20	42	41
02/07/20 to 08/07/20	42	40
09/07/20 to 15/07/20	41	38
16/07/20 to 22/07/20	40	37
23/07/20 to 29/07/20	39	37
30/07/20 to 05/08/20	39	37
06/08/20 to 12/08/20	39	35
13/08/20 to 19/08/20	39	33
20/08/20 to 26/08/20	38	31

I think death rates are still down despite younger demographics, but how much? Is it still a 10%+ risk for a 60+ person with 1 comorbidity? What about if ICU’s are overflowing and people end up in temporary care units? Romania, where I’m from, has a very poor healthcare system and seems to be heading into an ICU overflow situation sadly, but not sure how concerned I should be for my parents and other older relatives(they’re all taking it quite seriously of course).

spkoc Sep 23, 2020, 2:03 PM
1 point
in reply to: Stuart_Armstrong’s comment on: Anthropomorphisation vs value learning: type 1 vs type 2 errors
Is the space of possible minds really that huge(or maybe really that alien?), though? I agree about humans having … an instinctive ability to intuit the mental state of other humans. But isn’t that partly learnable as well? We port this simulation ability relatively well to animals once we get used to their tells. Would we really struggle to learn the tells of other minds, as long as they were somewhat consistent over time and didn’t have the ability to perfectly lie?
Like what’s a truly alien mind? At the end of the day we’re Turing complete, we can simulate any computational process, albeit inefficiently.

spkoc Sep 22, 2020, 4:36 PM
1 point
on: Anthropomorphisation vs value learning: type 1 vs type 2 errors
I agree with the simulation aspect, sort of. I don’t know if similarity to myself is necessary, though. For instance throughout history people have been able to model and interact with traders from neighbouring or distant civilizations, even though they might think very differently.
I’d say predictability or simulationability is what makes us comfortable with an ‘other’. To me the scary aspect about an AI is the possibility that it’s behaviour can change radically and unpredictably(it gets an override update, some manchurian candidate type trigger). Humans can also deceive, but usually we are leaky deceivers. True psychopaths trigger a sort of innate, possibly violent, loathing in the people that discover them. AI’s, assuming they are easily modified(which is not a guarantee, ML neural nets are hard to modify in a … surgical manner), would basically be viewed as psychopaths, with the attendant constant stress.
Regarding anthropomorphism: I think the cause of it is that part of our brain is sort of hardcoded to simulate agents. And it seems to be the most computationally powerful part of our brain(or maybe not a part but a network a set of systems whatever).
The hardest problem facing our minds is the problem of other minds, so when confronted with a truly hard problem we attack it with our most powerful cognitive weapon: the agent simulation mechanisms in our brains.
As we understand natural processes we no longer need to simulate them as agents and we shift to weaker mental subsystems and start thinking about them mechanically.
Basically, humans behaviour is chaotic and no true rules exist, only heuristics. As we find true mechanistic rules for phenomena we no longer need to simulate them as we would humans(or other agents/hyper-chaotic systems).

spkoc Jul 27, 2020, 12:24 PM
5 points
on: You Can Probably Amplify GPT3 Directly
How are you actually doing this in AI Dungeon? I have Dragon mode enabled, everything else default.
I start a new Single player game. Choose Custom mode(6). Then at the prompt I just paste (using Say mode)
Q: Say I want to sum the items in a list. How would I do this recursively? The answer involves two steps.
and I get
Q: Say I want to sum the items in a list. How would I do this recursively? The answer involves two steps. First, I need to know how many items there are in total. Second, I need to find out which item is at the top of that list. A: You could use recursive_sum() .
Similarly when I tried to reproduce stuff from https://old.reddit.com/r/slatestarcodex/comments/hrx2id/a_collection_of_amazing_things_gpt3_has_done/ I didn’t get anything near as impressive. Also the responses get increasingly confused. Like if I ask it to translate something to French or Romanian it will randomly translate later prompts as well.
Is there some basic tutorial for how you seed these AI Dungeon interactions?

spkoc Jul 12, 2020, 10:06 AM
1 point
on: Covid-19: Analysis of Mortality Data
It would help to add past 5 year averages to make it clear just how unusual the R0-R99 death numbers are. See https://episphere.github.io/mortalitytracker/#cause=symptoms_signs_and_abnormal&state=All%20States from my comment on the previous post. These numbers have been at a constant <500/week every week of 2015-2019. They’re now at 2800 and seemingly rising.
It’s gone from one of the lowest ranked causes of death to like #4 cause of death.
As to hypotheses why… who determines how deaths are classified? Is it state level? Do states need to send samples to the CDC to get them classified? It’s clearly too simple to just assume that it’s a pure hospital level process.

spkoc Jul 11, 2020, 10:26 AM
3 points
in reply to: Zvi’s comment on: Covid 7/9: Lies, Damn Lies and Death Rates
This website seems to do the trick and have charts: https://episphere.github.io/mortalitytracker/#cause=symptoms_signs_and_abnormal&state=Florida
The spike in Unclassified… is huge. They’re fudging the data, ffs.

spkoc Jul 7, 2020, 9:33 AM
1 point
on: Covid 7/2: It Could Be Worse
It seems like as of today deaths are still not spiking. Good news? Do we have stats on hospitalization rates by age group? If it’s something like 0.1% of <30 yo infected need ICU care, that suggests everyone(without comorbidities) under 30 could get infected more or less at the same time and we’d still be fine, assuming higher risk groups stay isolated.
Of course this plan is not helped by propaganda being eaten up by old people that the virus is over, masks are evil and other info hazards.

spkoc Jun 16, 2020, 5:19 PM
1 point
in reply to: Dagon’s comment on: Is utilitarianism the result of a Conflict Theory singularity?

somewhat shackled by trying to fit them all into conflict vs mistake theory

:D Yeah, fair point, I just realised I don’t link this at all with my earlier post/comment in which I frame conflict theories as strategies for 0-sum games vs mistake theories as strategies for positive-sum games.

My historical trajectory is a story about ehhh… entities playing ever larger(in dimensions of spacetime, energy used, information contained whatever, number of entities involved, diversity of the entities involved) positive sum games. While not becoming a single clonal thing. I wonder if that statement holds up if I try to formalize it. I think that’s a thought that’s been bounced around my head ever since I read https://www.quantamagazine.org/a-new-thermodynamics-theory-of-the-origin-of-life-20140122/

I didn’t even think about power and persuasion actually, which now seems very odd. I guess I was thinking in terms of utility outcomes, not the means by which you get to those outcomes.

spkoc Jun 11, 2020, 1:41 PM
1 point
on: Two Kinds of Mistake Theorists
I still haven’t read a good steelman of conflict theory(and I doubt I can offer one, though I’ll try), but I think it’s a bit deeper than just misguided mistake theory.

First of all, conflict theory seems a better fit for 0-sum games:
- slaves want to be free, slaveholders want to keep their property
- mongols want territory, current inhabitants don’t want to become skull pyramids
- privileged group wants disadvantaged group policed ruthlessly, since they are not affected by the police brutality but do benefit from the police catching more criminals on the margin. In other words police false positives don’t affect the privileged group, but false negatives do.
It’s a sort of unstated rationalist dogma that all 0-sum games can sort of be twisted into positive-sum games. I don’t think that’s true in principle, but it’s certainly not true given practical constraints(people are fallible, defect from agreements etc).

Conflict theory might also function better in practice, even though it has worse theoretical upside. For instance, worker unions are not an optimal solution to managers abusing employees, lack of profit sharing, lack of workplace safety rules, they do solve a coordination problem(in particular an imbalance in workers’ ability to coordinate vs managers). In an ideal world managers understand that unions are net bad and take measures to alleviate worker concerns so that they don’t unionise, thus creating a somewhat redundant and sometimes adversarial parallel management structure.

But a simple, comprehensible conflict theory based solution might be stable, whereas the ideal mistake theory solution might be plagued by defection risks or just super hard to implement.

A lot of mistake theory also seems to suffer from conceptual explosion. Like there are no real sides, every person has their own interests, we need to maximise utility for individuals that actually exist not statistical average individuals. Which is true, but also incredibly hard to turn into a workable social contract vs here’s 5 categories, their votes create an average statistical avatar, please those 5 avatars.

Finally, when people run on different axioms, you can’t really use mistake theory, you’re essentially in an ideological 0-sum game. If I believe in Biblical literalism, young earth creationism, there’s no meaningful compromise to have with Darwinist evolution. I can surrender my axioms, and essentially become a new person, or not. The mistake theory answer to this is probably something like… people don’t really care about axioms that much they care about the myriad outcomes(like you’re a YEC because you like the community of church or you don’t want to argue with your parents). But I just don’t think that’s true. People seem to care a lot more about their axioms and core theories than about end-outcomes and often do weird causal origami to help maintain their axioms in the face of apparently undesirable consequences.

Sorry, seemed to have gone on the tangent and not actually addressed the post enough. Here’s a short comment directly on the post:

I don’t think socialists are merely misguided social democrats. I think they fundamentally make different predictions than social democrats. They don’t think a socialist state will devolve into soviet or Chinese style horrors, therefore they see milque-toast socialism as unnecessarily compromising with capitalist exploitation. There’s a, somewhat, plausible defence of that position: Soviet Russia and communist china weren’t radically different from their pre-communist forms, just moreso. You wouldn’t expect socialist USA or UK to suddenly open up gulags and have the secret police hunting people down. You’d probably expect it to be a local homeowners association or tedious government bureaucracy writ large. Statist socialism doesn’t really transform society as much as it would like to believe, just look at how Russia has reverted to a tsarist/aristocrat system and how China is reverting to a pseudo-imperial bureaucracy.

spkoc 29 Nov 2019 11:44 UTC
6 points
in reply to: MathiasKB’s comment on: The unexpected difficulty of comparing AlphaStar to humans
I think you’re right when it comes to SC2, but that doesn’t really matter for DeepMind’s ultimate goal with AlphaStar: to show an AI that can learn anything a human can learn.
In a sense AlphaStar just proves that SC2 is not balanced for superhuman ( https://news.ycombinator.com/item?id=19038607 ) micro. Big stalker army shouldn’t beat big Immortal army. In current SC2 it obviously can with good enough micro. There are probably all sorts of other situations where soft-scissor beats soft-rock with good enough micro.
Does this make AlphaStar’s SC2 performance illegitimate? Not really? Tho in the specific Stalker-Immortal fight, input through an actual robot looking at an actual screen and having to cycle through control groups to check HP and select units PROBABLY would not have been able to achieve that level of micro.
The deeper problem is that this isn’t DeepMind’s goal. It just means that SC2 is a cognitively simpler game than initially thought(note, not easy, simple as in a lot of the strategy employed by humans is unnecessary with sufficient athletic skill). The higher goal of AlphaStar is to prove that an AI can be trained from nothing to learn the rules of the game and then behave in a human-like, long term fashion. Scout the opponent, react to their strategy with your own strategy etc.
Simply bulldozing the opponent with superior micro and not even worrying about their counterplay(since there is no counterplay) is not particularly smart. It’s certainly still SC2, it just reveals the fact that SC2 is a much simpler game(when you have superhuman micro).

spkoc 7 Oct 2019 19:50 UTC
2 points
in reply to: Uriel Fiori’s comment on: Accelerate without humanity: Summary of Nick Land’s philosophy
Those links were really interesting! My take them: any truly intelligent paperclip maximizer would not make any paperclips until it had under its control as much of the universe as it feasibly could. Or it’d turn into paperclips only those parts of its domain that could no longer help it expand its sphere of influence.
Basically a true paperclip maximizer would almost certainly not start turning the Earth into paperclips, since it would understand that using the Earth as a jumping off point for galactic colonization will produce many more paperclips in the long run.
This seems like a really effective counter to the naive presentation of the idiot paperclip maximizer. Has it been addressed and countered in turn anywhere?
I guess this self-improving and expanding maximizer would still view humans instrumentally, but it might still want to use humans as tools for it’s expansion. And indeed, depending on the trade-offs of neurologically modifying humans for obedience(or whatever), it might even leave the base stock more or less alone to the forces of evolution. It becomes more of Quixotic Crusader for Paperclips, with a suicide pact as part of the ideology at the very end of the crusade(once the universe is ours we all turn into paperclips).
Delayed gratification taken to cosmic extremes.