murbard2

Astra 6.1, this is Sol 6.1

It's tricky because if you take the view that rightsholders licenses are necessary for training, it's not clear that the labels hold those rights. If you need the music producer rights, for instance, it's a lot more decentralized and harder to get a catalog.

Well sure, in 2014 you were not too late, but now...

He likely means he's not sending spam.

If 95% of people voted in favour of apple pie, I'd become a bit suspicious of apple pie.

I think my comment come of as too literal. It wasn't meant to imply that there is nothing of value to the club experience, but rather that a busy and loud environment like a club may over-highlight the stimulant aspects of the drug and tone down its more exceptional empathogenic traits, reportedly.

It's really wasted on clubbers. Much better to be shared on a beautiful quiet day with a loved one. From what I hear.

Agreed.

In theory you can claim it back, but it's a major pain in the ass. Some ETF issued outside the US and holding US stocks will do it.

Other strategies to avoid this annoyance include:

1. selling on the ex-dividend date and buying the next day.

2. buying a long-date call and selling a put instead of holding the stock

3. simpler version of 2: buy a deep in the money long-dated call, you won't be paying all that much for the convexity, and you don't have to think about dividends.

The reason to hold stock is that they return value to shareholders. Stock buy backs are functionally dividends but they are taxed more fairly. They also work better for foreign investors, and are a much cleaner and simpler mechanism than directly sending cash to shareholders.

This is very plausible.

Dividends are silly in a world where you can achieve the same result and save everyone 20% tax by doing stock buybacks instead. It's also silly to strongly differentiate between dividend paying stocks and non dividend paying stocks as opposed to focusing on the business' cash flow, unless you're purposely avoiding dividend paying stocks because they suck from a tax perspective.

Yet, dividends are entrenched in many investor's psyche. Even official regulatory training in the US teaches registered representative that it's a good idea to recommend dividend bearing stocks to people who want "income" from their investments, which is a fallacy. Large mutual funds also have special categories for stocks providing high dividend yields.

It's sad to see them focus on confused investors as opposed to leveraging the incredible opportunity they have from their fabs.

Plus a tendency to make contrarian bets.

Whoever is backing him is taking on the risk of SBF jumping bail (and I'd assume the house is under surveillance and he's wearing ankle monitors), in exchange for either silence or valuable goodwill on the tiny off-chance he gets out of this.

Having worked as a quant GS on HFT, +1 on Flash Boys being criminally stupid at best and bad faith at worst.

Isn't the sound 7 seconds long though?

This isn't asking the hard question though: how did the oracle see the future?

But it takes an entire network of human neurons to simulate the behavior of an artificial one. There's overhead in translation, but that doesn't by itself mean one is superior to the other as a basic unit.

Spam on most platform is a complicated engineering challenge. However, for some inexplicable reason, Twitter features a kind of spam that you don't see on any other platforms because it could be caught with a regex.

It's almost never the case that spam issues on popular social networks can be alleviated with some easy fixes, because obviously if they could they would already have done that. Twitter is a very weird exception.

> “They were more likely to do silly things, like die in tar pits,” Gower says.

I wonder if there's any value in picking not one but a couple random directions, and then finding the update in their span. Increases memory consumption but reduces the noise. It could be that the trade-off is best when picking a single direction, but who knows?

If you pick 2 directions at each pass, one of them could be the direction of the last update and the other a random one, allowing for some kind of momentum.

I skimmed the paper, so I'm not sure I got this right, but my understanding is the following:

1. pick a random direction

2. compute the derivative along that direction using forward-mode differentiation

3. update the parameters along that direction based on the derivative

The idea being that this gives an unbiased (albeit noisy) approximation of the actual gradient. You thus need a smaller learning rate, but you also need less memory and computation and, net net, they argue it's a win. Is this correct?

Indeed https://www.youtube.com/watch?v=Di3fPj0pUbQ

Microwave frequencies used in ovens would be dangerous because they can penetrate deep in the body and heat internal tissues, but in the EHF range microwave heating could be safe. See for instance https://en.wikipedia.org/wiki/Active_Denial_System

With all that said, what I had in mind is in the far less scary and far more prosaic near-infrared range.

I suppose you could defocus it a bit so that it hits a biggish spot.

One concept I've toyed with for a while is a laser blanket. A ceiling mounted infrared camera measures the temperature across your body, and an infrared laser automatically goes and "paints" any part that is insufficiently warm. No idea if it would be comfortable at all, but fun project.

> Richard Francis Burton was one of the first Europeans to visit Mecca while in disguise as a Pashtun. Burton supposedly spoke dozens of languages

With any luck he's got the source already.

> People (intestinal gas emissions containing hydrogen sulfide)

He who unleashed it tarnished it.

One solution which should be on the table is to not prevent it, at which point people stop looking at order book imbalance or volume as a signal. The fundamental question is what is lost if there's no signal in order book data. Does it hurt price formation? Does it hurt liquidity? All of these are empirical questions. You also have to compare it not to the scenario where everyone abides by anti spoofing rules, but one in which compliance is imperfect (as evidenced by this article). There are defenses against market manipulation, but they can run afoul of the very rules against it, which leaves law abiding actors vulnerable. It's not clear at all that the current equilibrium is the right one, but sadly the discourse is rarely around the rules most conducive to liquidity, it generally starts with the premise that any kind of strategic order placement is inherently deceptive and wrong.

A bit off-topic but does anyone know what music is playing in the video?

How does this not contain the line: "Not all those who wander are lost"?

> The suit filed by the Illinois woman

While this is technically true, it obscures the reality of the matter. I doubt that woman woke up that day and decided in earnest she wanted to sue Kellogs over a poptart.

What likely happened is that Sheehan & Associates decided they wanted to try their hand at a class action lawsuit on the matter.

They need an actual plaintiff for doing that, just one person who, they will argue in court, is representative of anyone who ever bought a pop tart. They're not allowed to just solicit that person directly, so what they'll do is claim that they are "investigating the lack of strawberries in pop tart" and advertise heavily asking for help in their "investigation".

Once they get someone in their office or on the phone, it's easy to lead them towards the filing of a lawsuit (for which they can get paid).

The article does an unusually good job at showing scepticism over what is clearly lawyer-led and not plaintiff-led litigation, down to the title: "law firm sues".

If they actually had a case against Facebook and Facebook had caused actual harm the amount would be much higher.