TTPrograms

If Google compiled the extensions that would work. Seems like it would take a lot of standardization to make that possible. But in principle that would certainly be better than the current situation.

Nice - I went to the blog page on your site and this wasn't there. Good stuff.

You both seem to have really confident opinions given that you admit you have no expertise in the field.

Legitimately, if you used ex. Ethereum to compile your code could actually prove that a given executable was compiled from given source. Otherwise the only way to verify a build is to build it yourself and check the hashes. Signing just verifies that the build came from a given person, not that they used the source code they say they did.

This was one of the interesting use cases I thought about when I dug into crypto.

You all should really play that up more in your messaging - "provenance" is one of the hardest and least-addressed components of building AI/ML/data science systems that actually have measurable impact (rather than analysts making plots and speculating). In general having a structured, centralized representation of business processes is super valuable I'm sure.

If you write a blog post describing how critical that is to practical data science efficacy with some examples I bet you'll end up in a bunch of VP inboxes.

Bitcoin does not preclude the use of the legal system or other channels of recourse.

In the example they give they predict the future of a simulation. If you have a perfect simulation, with perfect observation, why not just run the simulation forwards? Well, the goal is to apply this to the real world, where you possibly have only an approximate model and observations - which are noisy and imprecise. So, instead, try predicting the future based on noisy observations. Due to exponential divergence, it seems unlikely this would work. Looking through the paper, it looks like they do not analyze the performance under noisy observation - they just analyze their ability to estimate the Lyapunov exponents under noise, which is much easier.

So the real world application (in terms of forward forecasting) seems like it's limited to cases where chaotic divergence between simulation and real world is due to simulation model errors rather than observation error - the latter is still a fundamental limit. Otherwise, this demonstrates that DNNs can be trained to solve diff eqs well, which is fairly well-trodden work.

As a result, claiming that this can enhance ex. weather prediction is highly misleading at this point, as nothing has demonstrated any prediction performance improvements under noisy observation, which is really the fundamental (and practical) limit in chaotic systems. While it still may be possible to use DNNs to do that (by ex. learning how to most accurately estimate the components of state with the largest Lyapunov constants) I don't think this work demonstrates the feasibility of that yet. The language in the paper is more reserved, in that they primarily claim that they can use the DNN to learn a dynamic model even with observation noise and then that model can be used to estimate Lyapunov constants accurately. That's much more reasonable.

Esketamine is the patentable isomer they're trying to use to enable this process of patenting + FDA so the drug companies can make a return. So the end result is a very expensive drug that insurance would pay for vs a cheap drug with 90% of the benefits that insurance won't pay for.

Racemic ketamine is also dirt cheap, while esketamine might be absurdly expensive after being patented.

If you're an American, every time you read a new article on Ketamine for depression your thought should be "Jeese, healthcare is screwed up". Pharma companies don't benefit from getting Ketamine approved (since it's no longer patent protected), so no one is putting it through FDA for depression. As a result insurance will never cover it (under the current model).

The reason this treatment isn't mainstream isn't because "the research is still out" (though that may still be). It's because there are no financial interests to get it over the hoops into the highly convoluted state that enables affordable medical care.

The important question is if you think it's possible to create impervious functions of ~100 lines of code. I tend to think that with modern formal verification systems this is a feasible goal. Otherwise you have to stick with closed-source security-through-obscurity and rely on legal reversal. This seems pretty weak to me, though - ex, if a stock exchange was hacked, sure you could reverse it legally, but the market disruption would take years to recover. Regardless of the specific nature of crypto, I think formally verifiable open systems are both achievable and a worthwhile goal.

By this logic so is every financial software system, though.

But let's make sure that the definition of AI doesn't encompass this because...?

It seems disingenuous to refer to that as proof-of-stake.

It was so nice of them to, out of the goodness of their hearts, provide such a competitively priced service. It sure seems unlikely that it was at all motivated by competition with alternative financial systems.

I would argue drones are mostly commodities at this point.

Does your insurance cover it?

Keep quiet about condoms too, otherwise people might think you're suggesting they have sex!

It's not really even clear to me that meth is that much of a killer. The article quotes 6000 deaths per year for all stimulants in the US. Alcohol related deaths per year in the US seems to be around 88,000. Obviously alcohol use is more common, but you also have confounding issues of increased risky behavior taking in general for meth users.

I think the effects you're describing are mostly all the result of developed behaviors / personal neglect from taking large does regularly. Taking high doses of, ex. Adderall for two years would likely have similar effects.

If we had mechanisms to identify people who were perfectly trustworthy, ensure these people are perpetually and exclusively in power, and construct perfect security for all involved computer systems then this reasoning would make sense.

In the same way that ex. Bitcoin has intrinsic value as means to write data to an immutable public ledger, both of these applications are insufficient to justify the value each of these things hold.

The fact of the matter is that intrinsic value is not a real thing, and as a model it makes bad predictions for assets primarily intended as mediums for exchange and value storing.

You could definitely reflect ex. infrared without affecting photosynthesis. I think your response is a little quick to brush off these ideas. There are many possible venues for geoengineering that could have fantastic benefits, even beyond affecting climate change. Brushing them off as ideas akin to "playing god" is both unprincipled and potentially harmful to both mankind and the planet.

Alleviating suffering over the long term, which includes fixing institutions when it's appropriate.

Bill Gates is also applying his money where he thinks it will have the most impact - their foundation often invests in new technologies and startups whose products will have the greatest impact per dollar. If Gates and Musk are both working to maximize impact their difference is who they focus on. Gates primarily focuses on the most impoverished on the planet, while Musk focuses on the first world (and indirectly the most impoverished due to the effects on climate change).

Saying that the Gates Foundation does not reinvest into new ideas that maximize impact does a great disservice to what is one of most effective and effectiveness-oriented philanthropic organizations on the planet.

> Smart contracts are being suggested as the building block of "autonomous corporations"

By who? I'm not aware of any major developers suggesting this. As far as I can tell this is just a strawman promoted on HN. The Dao was a single instantiation of this sort of idea promoted by a small number of non-core developers - it doesn't define the idea of smart contracts.

Why did this get pushed so far off the front page? Currently it's at 51 points after being up for 2 hours and it's on page 3.

Where do you get your numbers?

https://en.wikipedia.org/wiki/List_of_U.S._cities_with_high_...

Consider me sniped. My first thought was hashing mod k. Then when you create your hashmap you just check for collisions as you go and then test that they're k apart as opposed to 2k,3k etc. Then you don't have to scan through the hashmap again. It seems like if the parameters described in the problem are uniformly distributed, ie k is generally pretty big and you don't have a huge list of numbers then this initial filter would get you really close.

Let's be clear to differentiate when we're talking about fundamental laws like conservation of momentum or unitarity versus "how much energy can be stored in a fiber", which was really just the derivation someone did on a specific class of fibers and then assumed that every other fiber would be the same thing (some homogenous material, boundary conditions, etc.).

Calling this particular result "breaking a fundamental law" is just typical hyperbolic science PR.