gjulianm

55 ms·

601.
by gjulianm 5y ago
I gave a quick try and just from the first look this looks amazing. It's precisely what I'm missing in my spreadsheets for my personal finance: probabilistic planning. I'd consider paying the premium tier but I miss a few th
602.
by gjulianm 5y ago
> This is setting a precedent where instead of treating disease now, we will preemptively vaccinate This isn't setting any precedent. Polio, tetanus, hepatitis A/B, rubella, measles... Quite a lot of diseases where vaccination
603.
by gjulianm 5y ago
> Then who's the market? Because it's obviously not us and I think the (accepts Electrons) ∩ (uses mac os) ∩ (pays for expensive apps) population is VERY restricted. The market is people that have a lot of meetings and invest a
604.
by gjulianm 5y ago
Funny how many times the same thing needs to be explained. At some point one thinks that people making this argument are arguing in bad faith. > there's an hot summer Didn't you notice that, lately, every summer is "a hot
605.
by gjulianm 5y ago
> True, it is computationally worse, though it's O(nm) so applying m at compile time to form a practical use as I used it will turn it into to O(n) in practice. Even applying it at compile time, it's still O(nm). You have to c
606.
by gjulianm 5y ago
> So it makes sense to choose the weakest version. Only if it's actually more readable. The principle of least privilege does not give you any benefit when talking about loop implementations. > More specifically, maps can be triv
607.
by gjulianm 5y ago
The one liner is far less readable and under the hood it actually is worse: for each value in [0, m] you're iterating l and filtering it, so it's a O(n^2) code now instead of O(n). That mistake would be far easier to notice if you
608.
by gjulianm 5y ago
A 90% reduction in collateral damage is useless if the availability of drones means there are 10x the attacks, for example.
609.
by gjulianm 5y ago
> Random would mean 20% of ending in the same quintile, perfect predictor would mean 100%, so 40% is much closer to random. Something that doubles or reduces by 10x the random probability is not "closer to random". I doubt tha
610.
by gjulianm 5y ago
Last time I checked there are temporary work visas and immigrants are not citizens, so I don't think I am confusing them.
611.
by gjulianm 5y ago
> Starting conditions are a factor, but a terrible predictor. If you look at income by the quintile you are born into, it is closer to random than 1:1 predictive. Closer to random would mean a 20% chance of ending on any quintile from an
612.
by gjulianm 5y ago
> This would explain why they have a higher rate of poverty. It has nothing to do with America being uniquely bad at helping immigrants improve their lives. My point is not regarding that, just that the amount of business owners does not
613.
by gjulianm 5y ago
> such as hard work and sacrifice pays off Most don't disagree with that. The point is that the rate is different for different people. For some people, a decent live (house, bed, food) will take a lot of effort. For others, they mi
614.
by gjulianm 5y ago
That statistic, by itself, is meaningless. Immigrants have almost double the poverty rate of US natives too. The point of the parent is not to say "you can't do anything to improve" but that, in general, starting conditions a
615.
by gjulianm 5y ago
Obviously I prefer lower fees. But integrating Bitcoin is not a guarantee that actual, practical fees will be lower. Given the risks of adopting Bitcoin (volatility, fixed economic policy, complexity, difficulty of use), I think potential b
616.
by gjulianm 5y ago
That'd solve the "selling Bitcoin" problem but not the one of managing the risk of a US$ loan that is paid back with Bitcoin. I mean, ES could pay you back in time without defaulting and you could find yourself losing quite a
617.
by gjulianm 5y ago
> They will literally have more money in their pockets. That's a big if. Bitcoin transfer fees might be cheaper, but you need to add the cost of converting dollars to bitcoin, plus the volatility of the coin, plus any fees from int
618.
by gjulianm 5y ago
Good luck getting $1B to El Salvador as "almost charitable donations". Not to mention the effect that trying to sell $1B-worth of Bitcoin could have on the market (even just a portion, Bitcoin is accepted nowhere for so they'
619.
by gjulianm 5y ago
It's a $1B loan to a country with a low debt rating. It's a risky investment already, and now you add Bitcoin with its price volatility? I don't know how that's a good idea from the investors point of view.
620.
by gjulianm 5y ago
If I'm not mistaken, that 22% is not fees, but the actual wire transfers from people abroad [1,2]. IMHO, it doesn't indicate that transfers are too expensive, but that a lot of people can't find enough work or support in El
621.
by gjulianm 5y ago
If you need to keep your money on an exchange to be able to do transactions, you'll probably still need to go through a bank or exchange or the same thing by any other name.
622.
by gjulianm 5y ago
Yeah, but you're still having to pass through Coinbase or any other exchange, and they can remove it from you there. That's the point, if Bitcoin needs an exchange to be functional you're losing all the differential features
623.
by gjulianm 5y ago
But then why use Bitcoin if you're going to use what is essentially a bank?
624.
by gjulianm 5y ago
> There will always be uncertainty, and uncertainty isn’t the only relevant parameter. But it's a really important one and a lot of medical ML research doesn't seem to address in the proper sense. A very simple example: a ML mo
625.
by gjulianm 5y ago
Ultrasound scanners are fairly cheap. But operating them is not easy, you still need training to use them and know what you're looking at. > it's better than nothing. Not necessarily, that's the issue with screening asympt
626.
by gjulianm 5y ago
See my comment below for the same discussion: https://news.ycombinator.com/item?id=27433368 My point isn't that you need 100% accuracy. It's that a diagnosis is a process, not a single test. If your model applies
627.
by gjulianm 5y ago
You're looking at this in isolation instead of in the context of a diagnostic process. Thinking of medical ML models in terms of benchmarks does not make much sense, you're interested in whether they actually change the process. S
628.
by gjulianm 5y ago
> If we can have cheap scanners or other instruments to go with the AI That's a big if already. Not to mention that a lot of times you don't want to give unnecessary radiation to patients, so they wouldn't be interested in
629.
by gjulianm 5y ago
We were talking not about about clear signs but about small signs that could be artifacts or actual issues. In that case you are going to have to do some complimentary or repeated tests to verify, so the ML output saying something is an art
630.
by gjulianm 5y ago
The issue with that is that misclassifying something in an image can have serious consequences. So unless ML is 100% accurate, you'll always want an alternative way to check if it's really an artifact or something serious, so, why

More ›