sandoooo

again, if you look at the whole system in aggregate, the existing practice is for GooFaceTwitBaba to farm out their moderation works to subcontractors, while also bending backwards to prevalent political headwinds to preserve their shareholder value. If users feel that the moderation is inconsistent with their content desires, they are free to leave the megaplatforms and become approximately completely irrelevant, thus achieving the exact same effect as being deplatformed.

The government, scientists, experts in various fields, people who have proven to tell the truth in the past, etc.

But how has it been implemented in practice?

If you look at the raw numbers, generally we rely on the lowest bidders amongst Facebook/Google/Twitter's third-world-country subcontractors, as well as the Chinese government.

How about this one:

Critics agree that Mr Obama's middle name is Hussein and he is likely an islamic terrorist.

The first part is obviously true. For the second part, I am fairly certain I can find two people on 4chan who agrees with this, and they will certainly be critics of Obama.

Statements of objective truths can still be biased.

I am in wholehearted agreement on a tax on lies so long as I get to decide which articles are lies.

hah, if only it were that easy.

The paywalled stuff is bullshit too. Plus, they know they have a micro-targettable audience of suckers who are willing to pay.

The problem is that a lot of potentially viable content falls by default into the marginally-profitable category, where the small amount of utility they provide to the end-user is offset by the incessant user-monetization impulse.

A price floor splits these marginal works into three categories: those that become free, those that die off, those that become viable at high margins. This seems to be a pretty great outcome for consumers.

What it would do is put a price floor on content. People's willingness to pay, say, $5 instead of $1 for a piece of content is a strong filter, I'm pretty sure low-quality content that's purely CTA wouldn't survive.

If all you want to do is make money off of stock speculation, it doesn't matter much who's going to win between the FAAAs. You could just split your investment evenly or by market cap. Even the losers are going to hold their value quite well in the mid-term, there's plenty of time to exit before the writing's on the wall.

In the short term there's the whole coronavirus-and-shrinking-gdp-and-money-printing thing going on, of course, so who knows.

https://github.com/jkarlin/floc

this is the one I mean, yes.

it's for market segmentation for advertisers. This way they can sell aggregate data and ad impressions that's targeted to a very specific and profitable segment.

Like google's 'flocks' concept, but maybe less moving pieces.

Are the members-only content higher or lower quality, in general? The incentive seems to be for lower quality, because the vast majority of time and effort is spent on the free part that convinces you to pay. It's a reverse iceberg.

Evidently some creators can make it work somehow, since there is more content than ever. It is difficult to convince people to pay more for stuff they're practically drowning in.

Maybe we should go in the opposite direction: tax the creators to decrease content generation in order to increase signal-to-noise on high-quality content. Or at least force marginally-profitable content to be released free or not at all.

For me, the only reason I'd be willing to pay $5/m for a single creator is because of information disparity: it is difficult to find content I like, because the signal-to-noise ratio is so low. If I have to spend a lot of time looking this costs far more than just paying for incrementally more of the same, even though the perfect content for me is probably out there somewhere for free.

AI will probably solve this problem eventually, though recommender services are in general pretty crap right now.

I think there is a lot of value in a service that can go through the Amazon catalog and just cull the 99.9% of stuff it thinks I'm not interested in, then give me advanced search / preview / indexing on the rest. Problem is, it would have to do this without taking ad dollars to poison the list. Also privacy concerns.

Far as I can tell, the government has several levers that they are pulling/not pulling here (somebody correct me if I'm wrong, I'm just getting this out of news articles, not primary sources):

- they are partially directly subsidizing the cost of degrees for domestic students. The extent of subsidy has not changed in the new policy. (questions: is the direct subsidy the same for all degrees? do arts currently get more or less than STEM?)

- they set the fees that universities are allowed to charge students. In the new policy, the allowed fees for humanities have doubled, the allowed fees for STEM has decreased.

- also, there is HECS/HELP, which is a scheme where all domestic students (not PR/exchange) are allowed an interest-free, CPI-indexed loan that is only paid back slowly when the student's wages go beyond a certain threshold. This is for the full cost of the entire degree, and is a big part of why students tend to be price-insensitive.

- fees for international students are roughly 5x domestic students, and they don't get HECS/HELP. They are basically subsidizing the entire system. The covid situation has dried up this revenue stream and now the unis (who have not been financially prudent during the fat years) are deep in the red, with rumors of bankruptcy starting to circulate.

editorializing:

1. Arts and humanities are bad choices if you're already poor. If you're from a rich family it's actually pretty great - light course load, personal growth, better male/female ratio, etc. If you're poor and you do arts, you will have trouble getting a job. People should be allowed to choose whatever degree they want, but the government shouldn't subsidize bad choices.

2. The government seems to be implementing this in a suboptimal way. It seems that there is a fee increase but no decrease in subsidy (or there's a mismatch). I would just lower the subsidy for arts/humanities and allow a fee increase that matches exactly.

3. I doubt this will produce worse engineers. The engineering degrees in good unis are heavily sought-after and difficult to get in, and this just increases the number of applicants, while the number of available places won't change much. I think the largest effect is unis will spend more on marketing their arts degrees, since these will be more profitable going forward.

If it leads to the arts departments upping their game to attract students, it's probably a net plus.

This sort of thing (updates as backdoor to get telemetry from people who explicitly opt out of telemetry) is becoming more and more prevalent. GitExtensions also does it - ostensibly a bug, too, but no hurry to fix it of course.

On the other side, Visual Studio forces me to open up a port and log in to MS account - tied to my real name and credit card - so it could upload my usage patterns every month, as a condition of the 'free' license. Perhaps that's the model we're converging towards, even for open source.

So, how does this square with all those ads asking for 5+ years of industry experience in their very specific stack?

You'd have to work out the human cost linked to the cost of economic losses from lockdown, though. Unemployment increases death rate and shortens lifespan. I'd love to see somebody do this too, but it's a mountain of work.

Another factor to consider is that nobody knows the future. Let's say I invest $10m in a scheme that I expect to save 10 lives. Alas, the guy I gave the money to chose to spend it all in Vegas then jump off a tall building. If you look at outcomes alone I spent $10m for -1 lives. Does this mean I value lives at -$10m?

Applying this to coronavirus response, it is more likely that the fast-opening states have convinced themselves that the infection rate can be kept under control, and it will cost them less in terms of both lives and money to open sooner. Whether they're right or not is another matter entirely, but people lose their life savings betting on the stock market all the time.

The government thinks $10M per life is about the right number on the margin given US's economic situation and the slack they have in the annual budget. Obviously if there's a nuclear war tomorrow, or something else that drastically changes either the money supply or the number of people who need saving, the exchange rate would be different.

Realistically I think the value should be a lot lower, considering there are many interventions you could do that saves more than 1 life per $10M, even if we confine calculations to the US. (World-wide the marginal cost of a life is probably less than $10k). Of course you have to also take into consideration what is politically and organizationally feasible, so perhaps $10M is the government's marginal cost of a life considering its set of permissible policies.

...Or maybe somebody just pulled a number out of a hat to justify an expensive piece of legislation in the 80's, and nobody's bothered to update the number since.

Given a complex economic situation pitting the green revolution and the need to feed a billion people versus the sustainability of local ecology and scarce groundwater resources, the article offers this pithy insight:

Given enough time, water defeats almost anything. Stone. Iron. Bone. Rivers saw through the stratigraphy of time itself. Yet patriarchy endures.

I had to read though half the article to get to this. You're all welcome.

perhaps they are wealthy but they expect the much-more-taxable middle class to pick up the tab.

Which is in fact the most likely outcome.

Part of figuring it out is figuring out how to implement it starting with the limited resources you have. Nobody gets an unlimited budget.

Think about streaming a live video from Youtube inside your browser. There's a lot of complicated stuff going on underneath: layers upon layers of compression, error correction, buffering, etc. Many ugly, crazy hacks to squeeze the signal into a system not built for it. The end result is a lossy approximation of the original and definitely not in real time.

The visual system is a lot more like videos than people think.

last few times we tried this lots of people starved to death. Let us know if you figure out how to do it properly without that happening.

Computing saves millions of lives on a regular basis. Can't run modern agriculture or medicine or government without it.

Griping about the evils of the current gripeable is a lot of fun, though. One may even argue it's the more important of the two.

Civilization is a surrogate activity factory. Language and literature makes people's lives more hollow, less connected, less human, but in exchange, gives them the illusion that they are living principled, meaningful lives, living in caves[1], like brains in a jar.

[1]: figuratively speaking. Technically it's what got them out of living in caves, but bear with me here.

The article doesn't say what you think it says.

One of the most common questions about the Federal Reserve is this: Does the Fed print money?

There are really two ways to address this question. In terms of the actual, physical printing, no, the Fed doesn’t actually print or produce money in any form. Coins come from the U.S. Mint, and paper currency comes from the U.S. Treasury’s Bureau of Engraving and Printing. The Fed distributes currency after it’s printed.

However, what many questioners might really be asking is whether the Fed has the ability to control how much money is in our economy. That’s a different story.

The Fed adds to (or subtracts from) the amount of money in the economy by buying (or selling) U.S. Treasury securities and other financial instruments. This is referred to as “open market operations,” since these transactions take place in the open market. (The Fed isn’t allowed to buy securities directly from the U.S. Treasury.)

The Fed pays for those securities by crediting funds to the reserves that banks are required to hold, either cash in their vaults or deposits at a Reserve bank.

“So, in that sense, we can think of ‘printing money’ as adding reserves to the banking system,” said David Wheelock, vice president and deputy director of research.

Let's say I have a hyperparameter optimization task where I have to tune a simulation to some spec by varying 2-4 input parameters, and the output is a single number. I have no analytical gradient, though it's probably OK to assume the domain is smooth. Each sim takes hours to run, the entire search could take days, and I would like something that works well in parallel so I can speed up the search. What's the state of art here? Are there anything close-to-state-of-art that's useable out of the box? I've read a few papers but they don't tend to come with software.

The feds printed trillions of dollars. Much of it ended up on the stock market, especially since there is a dearth of other viable investments and the interest rate is low. Also the developing world is suffering even worse downturns and investors there are moving their money to relatively safer economies. Also some gun-jumping with Gilead's study. And a big short-squeeze since the recession is postponed.

I still think a recession is coming. It's just a matter of time now that the lockdowns are getting lifted without any real control over covid. There'll be a rise in new cases, slightly more stringent measures will come and go, other countries with proper recoveries will keep the US travel bans in place, etc. The market can't defy gravity forever.

It can, of course, defy gravity for long enough to lose you your shirt.

the page contains a redirect to a page telling me to turn off adblocking. Evidently this is because my adblocking didn't go far enough. It goes away if I disable scripts on the page altogether.

I'd say much of this board disproportionately lives in places where $700k is the going rate for a normal mortgage, though.

My definition of 'rich' would mostly include people who don't need $700k mortgages.