- once private information is leaked, it can never be undone
- being treated differently by some party because that party had knowledge of private information about you is difficult to prove
The ultimate solution is to legally require that important services cannot depend on private information.
An example would be: define a set of properties health insurance cost is allowed to depend on, and then have health insurance providers publish their formula for the premium (that can only depend on that set of properties).
> : define a set of properties health insurance cost is allowed to depend on, and then have health insurance providers publish their formula for the premium (that can only depend on that set of properties).
That's exactly how it works in Italy
Parameters used to establish premiums must be approved and you can't make them in a way that target a small group or a single person
This is likely to increase insurance costs on average. So it’s a tradeoff.
It also has a big issue of adverse selection. Say everyone with some known “private bad health markers” signs up for insurance but the insurance can’t take into account. That’s not going to be great. So we need to be careful to allow enough to be taken into account.
The western economic system cannot deal with lockdowns, and more generally, any kind of stop-the-world scenarios.
What will be most important now is how the west will be able to deal with this crisis in comparison to China.
If the Chinese system is able to deal with the crisis more effectively in economic terms, this may put the western system as a whole in jeopardy, in particular values like freedom and privacy.
A very good point (and I'm not a china shill). China's philosophy is The State Above The Individual, so it is willing to trade off a many more deaths, in theory. The west is The Individual Above The State (or at least to a much greater degree), so in theory we should do worse.
However the west has imposed large scale quarantines so the collective is in fact valued (because most people understand them paying a some cost will benefit others).
Also china is not an ant's nest, and the government can't go mad or their government will become unstable and even overthrowm.
We're not so totally far apart. I don't know. My feeling in good times the west does better, in hard times authoritarianism does better. We're in a hard time. Things will change. Also china and the west are entangled economically (so you can't examine each in isolation), so if the west does worse, china suffers economically. Also china's allegedly been misusing cash - the tide's going out, who's (more) naked? It's a good question.
But that's not what we've seen is it, China has worked hard to minimize the number deaths by going really hard on a heavy lockdown early, while the US with 1/4 the population has dithered and not really mobilized the resources of its state and has 24 times as many people die as China has.
I think it's more that modern China has been prepared to do everything it can to minimize death in this case and by doing a serious lockdown along with serious contact tracing they've come out early and are able to reopen their economy
Some countries in the West have done the same and are in a similar state. Others have fiddled why their countries have burned.
I wouldn't put too much stock in their exact numbers, but it's not realistic to think that China has had 24 or 12 or even 6 times as many deaths as publicly reported, let alone on a per capita basis.
China has 18.7% of the world's population and 0.00018% of the current coronavirus cases [0], if you believe their reported numbers. So yes, I'm willing to believe there are vastly more deaths than reported as well.
Why is it not realistic? The CCP numbers are essentially propaganda and have no basis in reality.
The CCP claims ~4,600 deaths, while there are reports like 5,000 urns being delivered to a single mortuary in Wuhan alone (Wuhan has at least eight mortuaries). According to some reports 40,000 urns were distributed in the span of just 10 days.
In that case it would appear the numbers used for your comments are very susceptible to the widely reported attempts by China to manipulate what is externally reported.
If China had deaths on the same per-capita as the US has had so far they would have had almost half a million - do you really think they are hiding that many?
It doesn’t matter what I think, this is a widely reported possibility with recent articles from both the Times and the Guardian. Serious reporters at least believe there’s evidence of interference and possible data manipulation by China.
Basically the only thing we can conclude is that the numbers from China are suspect & not comparable with those reported by western nations until more investigation is done. Maybe it will turn out China was right all along, we just don’t have reason to think that right now.
While data from China is difficult to find (and transparency always questionable), a key issues in comparing the two populations is:
- General health. Covid-19 rarely acts alone, at least in The West. Preexisting conditions play a key role. For example, obesity and its "side effect" diseases pair well with C19.
- How it cares for its elderly. People in nursing homes in the US have been significant contributors to the death count. Is that true in China?
It helps to get into the details a bit. While the proof has vet to be studied, anecdotally, cultural differences appear to contribute to the success (or failure) of a given government's decisions.
And that's sort of my point, countries that are willing to act and act quickly (be they communist or capitalist) are faring well and are restarting their economies, ones who's governments have tried to ignore the problem and hope it would go away are doing poorly.
I live in NZ - we've had no new cases nationwide for 2 weeks, we have (AFAWK) 1 active case in the whole country. Our borders are largely closed, anyone coming in is subject to mandatory quarantine. When we went into lockdown it was fast and deep, and people honored it, we started quarantining visitors from China early, as a result many of our cases came from the US. Now we're coming out, carefully step by step, I went for a drink at the pub on Friday night, then went out to dinner - there's still mandatory social distancing and collection of contact tracing data - that may all go away next week.
My hardware hacker friends in Shenzhen describe similar careful steps - apparently street markets may be back this week. But when they arrive they have similar mandatory quarantines as we have in NZ - does the US do any?
As far as nursing homes I think it's traditional for older people to live with their kids they don't get warehoused as much as we do in the West. As I understand it one of the big worries in China was that it was CNY, the whole country was on the move visiting family (those grandparents) they locked everyone down before they could return and then moved people back, directly into lockdown
I think what you might have is correlation. On the other hand, cause does exist for preexisting conditions. I'll go out on a limb a bit and theorize that elderly parents at home is safer than a bunch of them together in (under funded?) nursing homes.
Acting quicker might help but the fwct remains that underlying cultural issue have a significant impact on outcomes.
Side note: In NY state - and perhaps elsewhere - a nursing home patient with C19 that transfered to hospital and died is _not_ a nursing home death. The percentage of nursing home deaths is high, and - at least in NY - undercounted.
Would we have locked down the whole country is NY and NJ (and Philadelphia) done better with nursing homes?
> And that's sort of my point, countries that are willing to act and act quickly (be they communist or capitalist) are faring well and are restarting their economies, ones who's governments have tried to ignore the problem and hope it would go away are doing poorly.
Shocking stuff indeed. You've set the world of epidemiology on fire. What's your point? That in managing a dangerous epidemic communism/authoritarianism does better, or relatively liberal western government styles do better, or that governmental styles are not relevant? Because that was the original question.
I think the point is that the attribute most closely aligned with Covid outcome is not liberal-authoritarian, but functional - crippled by political division.
No, my point was that decisive governments willing to move quickly who are also able to motivate their populations to move with them (be they capitalist or communist) are doing far better here than governments whose leaders just put their fingers in their ears and tried to ignore the problem.
I lived in the US for 20 years, half my adult life, people there pride themselves in having a national government that is structurally deadlocked and unable to make sudden changes ... while this may be a useful attribute much of the time there are times, and a pandemic is one of them, when the government has to move really quickly - if you've fired all the experts and are more worried that your party's investors will lose money than you are that your voters will die then you get just the wrong sorts of results
Original question asked if the cap or com governments were intrinsically better at this. You're saying governments that don't screw up are better at it. Yes, but it doesn't answer the original question.
It's not so much governments that screw up, it's governments who trust the science, who are decisive and move quickly, and who's populations are willing to follow them - my argument is that it doesn't seem to matter whether they are capitalist of communist - some communist countries have done well (Vietnam, China), as have some capitalist ones (NZ, South Korea, Taiwan, etc).
I'm talking from a societal philosophy. If you absolutely insist on dragging stats into it, well, here goes:
China is incredibly monolithic in time (50 centuries of similar style of government, at least). The west is far more variable. The current US gov't acted like fuckwits, so deaths are unnecessarily high. Speaking as a brit, we've been even more fuckwitted.
I don't believe china's done as well as you claim. The fucked up early, downplayed covid, twisted the WHOs arm and plain lied about their infection and death stats.
Whether they reopen the economy early is of less-then-100% benefit if the rest of the world is not buying. If their economy isn't as healthy as they like to claim, this will hurt them.
But the bottom line is western governments can and do change. How china will fare I don't know. I remain where I started, I don't know which is better in these situations. The west may collapse, china may collapse, I can't see the future.
Quarantine doesn't necessarily benefit the state more than the individual. It benefits the collective more than the individual but the state and the collective aren't the same thing.
How do they differ? I considered them interchangeable. BTW, polite request, if you say something like this please give a link to something that explains where I'm wrong, it means I don't have to ask for useful stuff. Thanks.
I'm not a philosopher nor a political scientist, I'm not making a technical argument.
Are you asking about the inter change ability of the state and collective?
Did George Floyd benefit as much from the apparatus of government as Donald Trump? That's the difference between the state and the collective in my mind.
Or are you asking about inter change ability of individual and collective? I think it's pretty obvious...
The state is composed of the institutions, elected and appointed officials bureaucrats, etc. The collective includes everything else. I'm a part of the collective. I'm not a part of the state.
Oh, ok. I treated them as synonymous. I can see how "the state" could mean its formal structures only, but I included everyone in it. I'll mind the distinction more carefully now.
> The western economic system cannot deal with lockdowns, and more generally, any kind of stop-the-world scenarios.
A strong statement, but lacking some proof. My guess is that just like mold, the economy will grow back, adapted to the new constraints. And while the players might be gone, with whatever pain that might incur, the game is still played by the ones who adapt.
> The western economic system cannot deal with lockdowns, and more generally, any kind of stop-the-world scenarios.
I kind of disagree, though it may just be a diction issue.
I think it's the western political system and freedoms that can't enforce a lock down rather than the economic system. Countries that have aggressively locked down early, enforced that lockdown, and instituted privacy invasive tracing apps have been able to virtually eliminate the virus from within their borders.
Compare that to the US, for example, where people are insisting on their rights to not wear a mask for purposes of protecting others. It's less about economy and more about individual freedoms and lack of state power to enforce lockdowns.
==I think it's the western political system and freedoms that can't enforce a lock down rather than the economic system.==
Is this true? Even within the US we have plenty of states respecting stay-at-home orders and making things like masks mandatory. Other Western democracies like Germany, Australia, New Zealand have fared pretty well.
I feel uncomfortable at the ubiquitous, silent assumption that what is marketed as AI is a computer implementation of a brain.
I see how the term neuronal network reinforces this believe, but we (especially the researchers among us) should allow for the possibility that we are missing something.
Neural networks also have no ability to create new information based on their own mistakes. What is a mistake? When does something look "off" but still very interesting?
For example, you can feed a neural net all the recipes of burgers to create a perfect burger. Great. But how does the same net invent the burger?
The burger, like many foods or accidental art, was invented as a result of scarcity, circumstance, experimentation, or just fortunate error. That sort of imperfection is very hard to achieve with AI, because it is designed to be either perfect or fail.
GAN can do that. For example, AlphaZero invented strategies for the game of go from nothing but a random number generator and the rules of the game. As for perfection neither go nor chess AIs play perfectly, and they can still beat the best human players.
Of course, an AI intended to play go isn't going to invent the burger. But I see no reason why, given a list of ingredients, their properties and a model of what human enjoy eating, a neural network couldn't invent the burger.
Creating a new recipe is just an optimization problem at its core.
>>For example, you can feed a neural net all the recipes of burgers to create a perfect burger. Great. But how does the same net invent the burger?
Wait....but it just....did? It took the information about all possible burger recipes and invented a new one out of these. Like, a human could only invent a new burger if they knew anything about burgers in the first place, at the very least that it's a bun with some filling in between, otherwise you'd have no context to invent anything.
Not OP, but I think they're not talking about inventing a _new_ burger, but inventing _the_ burger, as in the first one ever.
As in, the neural net in this example is able to improvise a new burger recipe solely because it was given existing recipes to burgers as input; it did not come up with the notion of a burger and then produce a recipe that outputs something fulfilling that notion when followed.
Personally, I would argue that this distinction is not as clear-cut as the tone of the original comment seems to suggest. Humans didn't invent the burger from nothing either. We've been grilling meat and making bread for millennia, and sandwiches have been a thing for over a century.
A 'burger' is just another iteration of our biological neural nets' attempts to make food from ingredients already present in our physical reality. Given that we flow in a single direction through time, any food we make is in turn added to our list of ingredients for making food "the next time". One could argue it is only a matter of time once meat can be ground into patties and grains turned into bread that burgers start being made - given the relative benefits humans gain from consuming both.
This comes back to what others have expressed elsewhere in this thread, that the probable [most] important distinctions aren't between software vs hardware, or organic life vs silicon processors, but the environment & capacity to interact with said environment. Some sense of "innate tendency to experiment" (i.e. curiosity) is probably either equal in importance or a direct runner-up.
The burger was invented because a hungry traveler walked into a restaurant in Connecticut that was closing, and the owner had nothing but some beef and bread left. So he improvised - cooked the beef patty and squeezed it between two bread slices.
To this day they serve their burgers between two bread slices - not buns.
If you want to look it up, it's called LOUIS’ LUNCH.
I agree. I think its very widely known that our ANN’s are only very rough approximations of how the brain actually works, I think the people who say its a computer implementation of the brain are either laypeople who don’t know much about machine learning or the brain, are people marketing the hype for personal gain or people without neuroscience knowledge who have bought into the hype.
I also recently heard an argument for why our ANN models won’t spontaneously become sentient: human brains don’t learn from just observation, but also interaction. A young child doesn’t learn abouthow blocks are stacked by looking at images of stacked boxes, they learn through experimentation, by stacking boxes and seeinghow their actions affect the world around them. For an AI, that means we either need to also work on robotics so the AI can interact with its environment, not just sense it, or we need to simulate an interactive virtual environment. Some people are working on this and making great strides, but your average toy ANN won’t exhibit human intelligence in isolation, in my opinion.
Combine those two things and we’re still quite a ways away from human-like intelligence or implementing a human (or animal)-like brain.
Interestingly, there are some studies that imply that intense thinking about doing an activity (such as a gym workout[1] or hitting a baseball) can improve your physical skills than if you didn't think about it. So this is supporting the notation that you can rewire your brain by thinking, as well as tactile input.
That’s not really what I’m referring to (or at least, only a little). Once you have a mental model of something, you can for sure think on it or build on it without interaction, but to initially set up our mental models (as children or whatever), I believe it takes interaction. Once we have a base, we can think abstractly about it and learn, but building that base..
Or, put another way, its my belief that you can “_improve_ your physical skills” by thinking, but to buildthe skill in the first place, interaction is necessary.
But even if its not true and interaction isn’t strictly necessary, I think (wrongly oerhaps) that few people would disagree that usually learning by doing is far superior that only learning by thinking/reading/listening/watching. So even if not neccesary, its at least more efficient (doing both together is probably most efficient).
Absolutely. I think what AI has highlighted is that the problem set is now looking more similar to a human experience. For example, how you train based on input and learn from failure and how limited information can confuse even a human brain (think image recognition). That said, because the problem looks the same, doesn't imply the method of processing is the same.
I am definitely not an expert on this topic but my impression is that the research is not really focusing on structured abstractions of sensory input, or making these abstractions stateful. Shapes, colours, music, and whatnot are clearly stored and retrieved in our brains, which is something NN research is not looking at (enough).
I've followed Bitcoin for a while. What seemed to happens there was the deliberate manipulation of the price using all means available (ads, etc.) The result was a few winning big.
I have no reason to believe the same doesn't happen on the real stock market.
Few enough people can be bothered to leave reviews for free, charging for them is going to drive that number even lower. So people paying for fake reviews need to buy fewer to get results.
It’s a highly technical problem that doesn’t have much practical use. Essentially it says that a weak (polynomial time) classical computer (called the “verifier”) with the ability to query (polynomial number of times) a small number of all-powerful computers (called the “provers”) with unknown/unbounded computing power is able to be convinced that the prover is indeed all powerful (convinces the verifier that it knows the answer to some problem that is known to be difficult). These specific provers can not communicate with each other (collude) but can share some entangled bits (the quantum part). The result here is (basically) that the verifier can be convinced that the prover knows the answer to the halting problem.
The result is more interesting with some context though. One of the biggest discoveries of the 21st century is that PSPACE=IP. That’s where a single all powerful prover can convince a weak verifier of any solution in PSPACE (an extension of P that we believe to be more powerful than P). Then more recently we found MIP=NEXP which is a bunch of classic all powerful provers (with no collusion) can convince a single weak verifier of any problem in NEXP (which is like NP but with much larger proof sizes, thought to be more powerful than NP). The surprising fact is that multiple all-powerful provers can convince a verifier of harder problems. Intuitively this doesn’t make sense because we know each provers are all powerful and are permitted to do anything, even physically impossible tasks, so the fact that a single prover cannot convince a verifier of any problems harder than PSPACE is surprising. More surprising is that multiple all powerful provers can somehow prove harder problems when it feels like we’re not adding any extra power (they’re already “all powerful”).
So in that context we find the also surprising fact that if we weaken the “no collusion” requirement to “only quantum entanglement” and suddenly they can solve the halting problem (but still no more!).
This line of proving both upper and lower bounds about complexity classes is exactly what we want to see in the P vs NP world (which is in many ways less “powerful” of classes but more “practical”). But we don’t even know if P ?= NEXP.
NEXP is nondeterministic exponential time, right? In that case we know that P /= EXPTIME by the time hierarchy theorem and EXPTIME is contained in NEXP. We also have a direct result that NP /= NEXP (reference given in https://complexityzoo.uwaterloo.ca/Complexity_Zoo:N#nexp).
This is where I always wish they demonstrated some of the mathematical notation with some variant of quantum Lisp (even if it couldn't be executed in any practical way). The Regetti guys are always around here, so maybe one of them can chime in.
I skimmed the PDF, and it's more than a tad heavier than my usual fodder.
What I think I understood from the document, keeping in mind that given enough time a complete Turing Machine by definition should be able to emulate anything a quantum computer can accomplish, is that a quantum computer should be able to efficiently identify whether a problem run on a classical computer will exit without achieving a complete result. Not that you'll actually get to the result itself, but whether you should expect to be able to, given enough time and compute power.
Because of the density of the paper, maybe I only picked up on a tiny piece of the puzzle. And even then I probably botched it.
It's comparatively difficult to do applied research with grant funding, which is what most "publicly funded" universities tend to rely on. For this sort of high-value research, what you want is prizes so that success can be rewarded after the fact, even when the usual "market" mechanism of selling drugs is not working very well for whatever reason. Current policy approaches provide 'development incentives' of some sort as OP mentions, but clearly these are not enough.
One thing at a time, first uncover more promising avenues to antibiotics. Maybe they get private investment, maybe not, if not then go ahead and publicly fund the next stage.
None of this kind of thing stops say defense spending. We say hey our defense experts say we need x vehicle to defend our nation. Let's build x.
Our medical experts say we will need antibiotics in the future. It's a nice optimization perhaps to use crossover private profitability to fund the path to more drugs, but if the path isn't there privately, no one says, well we needed a new Aircraft carrier to defend the nation but none is available on the private market, oh well, guess we just live without.
Not in the current job climate, but if (say) successfully shepherding something through FDA approval were treated like a Nature paper, I bet you'd see a lot more researchers trying it.
The problem with the prize idea is that this stuff is expensive. Most academic institutions don't have a mechanism for doing something "on credit" and hoping to win a prize that covers the costs and then some. Some companies can, but the optics of giving prizes to BigPharma also aren't great.
> Most academic institutions don't have a mechanism for doing something "on credit" and hoping to win a prize that covers the costs and then some.
Some do, actually. It's called an "endowment". Unless by mechanism you meant an internal policy to allow for this - but that's the sort of stuff that can be changed with relative ease.
I chuckled a little at "relative ease" (nothing involving university legal stuff is easy) but ran some numbers to see how feasible that would be.
The median Phase II trial costs about $10M and has a 30% success rate; Phase III costs a lot more (say 2x that), but has better odds (58% advance to approval, and 85% of those get approved). Neglecting Phase I and everything before it, the expected cost is therefore about $16M (over ~3 years) and gets you an approved drug 15% of the time.
There are only 11 individual universities (plus three university systems) with endowments over $10B, but a surprising number with endowments in the $1B range. The draw rate on endowments is usually capped at 5% so that the principal remains intact, giving you around $50M for the entire university's endowment income.
A single trial would therefore consume about 10% of the endowment income per year--and with a 15% success rate, it's probably not going to win anything. There's no way a university president is going to let a group of researchers make a gamble like that.
That's precisely where the difference between grants and prizes would be clearest. If there is a big fat prize for, e.g. getting approval of a new antibiotic, people will be incented to push it through the pipeline.
That's the thing about applied research, in biotech/med and elsewhere: it doesn't get much interest from researchers, but the results are comparatively easy to evaluate for an uninvolved third party.
A University can never develop a drug. Pharma companies are very large organisations with a great deal of embedded expertise, and have discipline born out of having to sell a highly regulated product in competitive markets.
> Approval incentives were not the only policy included in the GAIN Act. There were also measures designed to promote stewardship, or appropriate use, of new antibiotics. In short, when a new antibiotic becomes available, it should only be used as a last resort to prevent new resistance from arising. This kind of responsible use is a good thing! But stewardship severely limits the number of patients who will receive a new antibiotic and, correspondingly, the potential sales volume.
Privately funded research is unable to overcome the well intentioned regulations that are preventing use of the antibiotics that are developed.
Instead of hoarding the few antibiotics we have, we should be moving faster and keeping ahead of the problem, but the current regulations essentially prohibit that.
The other side of high health care costs in America is that the wealth from overleveraged and dead Americans is absolutely needed to cover drug research and development for the rest of the world
The German voter is against coal (the young climate concious), against nuclear (the old green party supporters), against wind (everyone when it appears in their backyard), and the government killed solar in 2015 which resulted in approximately 80000 lost jobs.
What the German voter doesn't want is to consume less and/or save energy.
