absherwin

Increasing payouts should bring more players though it would be offset by eliminating the advertising that currently consumes some of that revenue. Flattening jackpots would have a negative effect but we just need to optimize for being above the point at which there's competition instead of maximizing lottery revenue.

If we assume the main problem with lotteries is that they make some people poorer, we need to solve two things: The payout being less than the cost due to both high overhead and transfers and the payouts tending to concentrate wealth since even in a lifetime of play few win the largest jackpots.

One can't make them too flat because consumers like large jackpots which is why they've grown over time and become harder to win. Presumably there's a level below which consumers would switch to other types of gambling regardless of legality.

There's a simple answer: Offer a lottery with a high payout rate and flatten jackpots to the extent they don't cause a shift into other worse sorts of gambling. The total amount of money transferred to support government programs makes up less than 0.5% of total government revenue; it's trivially replaceable.

These criteria are such a low bar that merely stating them upfront would have changed the discussion last week.

I wonder if these were the original criteria or if this is a partial walk back. The SBA usually defines a small business as having less than 500 employees even if it has significant revenue.

If I had been asked to guess the criteria based on the initial announcement, I would have guessed 500 employees, $100MM in revenue, or $5MM in money raised with some exceptions for top-tier investors or other judgmental criteria.

That’s true for most adjustments. Buffett also suggests investors ignore income statement effects of marking securities to market.

“I must first tell you about a new accounting rule – a generally accepted accounting principle (GAAP) – that in future quarterly and annual reports will severely distort Berkshire’s net income figures and very often mislead commentators and investors. The new rule says that the net change in unrealized investment gains and losses in stocks we hold must be included in all net income figures we report to you. That requirement will produce some truly wild and capricious swings in our GAAP bottom-line.” https://www.berkshirehathaway.com/2017ar/2017ar.pdf

The most significant adjustment was excluding a $3.2B writedown of the value of their stake in Didi

The key challenge with this system is merchant adoption. Since merchants are forced to charge the same prices in the less valuable currency, they'll drop out if its exchange rate falls by more than their profit margin (and perhaps before then).

You will find that management consulting firms (and the companies that hire them) have fairly clear rules about how they approach and talk about pricing and competition to avoid the appearance of impropriety. Most notably, they would never ever: 1. Indicate that their advice always is to raise prices 2. They would never share the pricing advice they gave to a competitor.

I don't know if what Patrick does is problematic. It wouldn't surprise me if he was trying to be glib and his actual advice is more nuanced.

If one interprets it literally: Someone who receives advice from him to raise prices can look down Stripe's customer list and find its competition and feel confident that the competition will receive the same advice.

Again, I don't know what he does. All I know is that what he says if read by the right person would be sufficient to trigger their curiosity to ask for records of his correspondence and find out.

Of course the customer can say that. And nowhere does Patrick suggest that taking his advice is a condition of continuing a relationship with Stripe. That's not the issue.

The issue is that coordinating pricing is illegal in any form. In any industry, each company could make more money if they knew their competitors would raise prices. Even calling a competitor to tell them that you are raising prices is illegal if they raise prices.

If there's even a single incident of two companies being advised to raise prices on products which compete with each other, it's about as close as it comes to an open and shut price fixing case. And this post makes it easier to prove because it suggests that companies know he's giving the same advice to their competitors. Whether that's said explicitly or not won't matter.

"every time I convince a Stripe customer to raise their prices ... we benefit directly" sounds like the exact words of an illegal cartel. While I doubt that that is Patrick's intent, if I were an antitrust attorney, this would convince me that the notion of payment processors as clearinghouses for price-fixing merits investigation.

This is a great example of how one can do something well-meaning at small scale (Advise tiny, tiny businesses that they would be better served having more confidence in their value) that can turn into something illegal at scale (Advise price leaders that they can safely lead by less and so-on).

This preprint explores, in part, cross-reactivity for a similar test and found it to negligible: https://www.medrxiv.org/content/10.1101/2020.03.17.20037713v...

This is an antibody test which will only be positive days after the onset of symptoms.

This is principally useful for several purposes:

1. Figuring out who to isolate in hospitals if the RNA test isn't available in sufficient quantity

2. Understanding who has already recovered from COVID-19 and is thus immune with all that implies in terms of inability to spread the disease and reduced need for PPE

3. Enabling us to confirm continued immunity later this year and understand how long the recovered will remain immune

This is not unique to this company. It's unclear to me whether the price is meaningfully less than competitors.

This thread is helpful for further understanding of the test and its utility: https://twitter.com/NAChristakis/status/1240689953895411714

Two more helpful references: State of testing techniques as as of a week ago: https://sph.nus.edu.sg/wp-content/uploads/2020/03/COVID-19-S...

The paper on which this test is based: https://onlinelibrary.wiley.com/doi/pdf/10.1002/jmv.25727

A version of what you suggest has been the law since ERISA was passed in 1974.

The problem is what to do if investment performance doesn’t meet expectations, lifespan increases, or future assumed yield decreases. Generally companies had to pay enough to be back to even within seven years. Newspapers sought the right to have thirty years to fully fund. McClatchy was larger than Congress was comfortable with and found itself unable to pay the required fraction of the difference between the NPV and finding amount. Hence, it declared bankruptcy.

The report is misleading because AP tends to be used far more on controlled access roads which have fewer crashes per mile. There are other biases too but this alone could easily explain more than the observed 1/3 risk reduction.

Unbiased means that if I draw infinitely many random samples from a population and average a statistic (in this case standard deviation) across all the samples, the answer will be the statistic computed from the population itself. If one divides by n instead of n-1, the estimate for standard deviation will be be (n-1)/n too small. One reading this might think, "Wait! We're going to infinity so the ratio converges to 1." That's true if the size of each sample also goes to infinity but not if we draw millions of ten item samples.

As for using up a degree of freedom, the easiest way to build intuition for why this is a useful concept is to think about very small samples. Let's say I draw a sample of 1 item. By definition the item is equal to the mean so I receive no information about the standard deviation. Conversely, if someone had told me the mean in advance, I could learn a bit about the standard deviation with a single sample. This carries on beyond one in diminishing amounts. Imagine I draw two items. There's some probability that they're both on the same side of the mean, in that case, I'll estimate my sample mean as being between those number and underestimate the standard deviation. Note that I'd still underestimate it even with the bias correction, it's just that that factor compensates just enough that it balances out over all cases.

A simple, concrete way to convince yourself that this is real is to consider the standard deviation of a variable that has an equal probability of being 1 or 0. The standard deviation is 0.5. But if we randomly sample two items, 50% of the time they'll be the same and we'll estimate the standard deviation as zero. The other 50% of the time, we'll get the right answer. Hence, our average is half the right answer (n/(n-1)=2/1). The correction makes the standard deviation double what it should be half the same while remaining zero in the other cases. This also suggests why dividing by n is referred to as a the maximum likelihood estimator.

That's a zero sum game. Let's consider two possibilities that aren't zero sum.

Late payments: A credit card company might make slightly more money when people pay late than it costs them in servicing and default costs while the cost of the fees alone may exceed that benefit even ignoring the cost of a diminished credit score

Increasing spending and borrowing: A company does marketing that convinces people to spend an amount (either through marketing a new card, a line increase or straight up marketing spend), that nets the issuer $100 in value and creates $30 a month in interest for the customer for six years.

The argument is the same as any other case of information asymmetries that create negative externalities. How should we regulate marketing cigarettes? How should we regulate marketing cotton candy? A purist would argue we ought to tax those activities.

The other bit of evidence she cites but doesn't quantify is that marketing new cards causes incremental debt.

That said, I largely agree with your sentiment that the article doesn't provide sufficient quantification. You also might want to consider that the author might have experience seeing that quantification and making decisions based on it that can't be shared in The New Republic without violating confidentiality agreements.

One lens through which to view this: Is a culture that tends to view its customers as mathematical entities likely to make the best decisions either for their customers or themselves? That's the real tragedy and is half of the explanation for the paradox that such genuinely intelligent and kind people do things that hurt others (and in some cases themselves).

Disclosure: I also worked at Capital One and briefly encountered Elena there

One bit of additional intuition: Since the square of a number goes up from addition by slightly more than it goes down from subtraction, perturbations increase the average of the squares. This is why the difference between the two quantities Feynman mentions is used to measure the variance of a set of numbers.

Since that’s still not precise, let’s compare the square of the mean to the mean square for two numbers a and b.

The square of the mean is ((a+b)/2)^2=(a^2+2ab+b^2)/4

The mean of the squares is (a^2+b^2)/2

Feynman’s claim is that the second is always bigger if the numbers deviate around an average (a and b aren’t equal).

So let’s subtract the first from the second. We get (a^2-2ab+b^2)/4. The numerator is equivalent to (a-b)^2. Since a square of a real can’t be negative, when a and b are unequal the mean of the squares is always larger.

Economic growth is the key to resolving the paradox you highlight of his seeming wealth then being inflation adjusted to only modestly high net worth today.

Using Maddison's estimate for UK per capita GDP in 1820, suggests that the average UK resident generates more than 10.7x as much wealth now than back then.

Of course, all these comparisons are fanciful and dependent on the utility one assigns to various things. In terms of the number of servants he could employ, a man of Darcy's wealth far outpaces all but the wealthiest in the first world whereas in terms of his ability to travel to Rome expeditiously, he can't measure up to a nearly broke student.

I agree wholeheartedly with the larger point that we have a higher education bubble.

I’m skeptical both that: Lower tuition would make taxing endowments more likely and that universities believe that that’s the case.

In a world with lower tuition, endowments grow somewhat more slowly and are easier to justify because they are the thing that enables low tuition.

For a more direct set of examples of how much inertia we have, consider the way wealthy individuals use foundations to avoid taxes. I haven’t heard people crying out to tax the Gates Foundation despite its endowment growing over time because it can’t keep up with its contributions. Nor is there a massive uproar about donor advised funds which (particularly in CA) can be 80%+ taxpayer funded while effectively lacking minimum disbursement requirements as part of a larger organization.

The point you’re raising is that an endowment could simply exist to make money for its own sake. The IRS combats this by requiring a minimum distribution of 2/3 of the endowment’s minimum return (which tends to be lower than actual).

See https://www.irs.gov/charities-non-profits/private-foundation... for details.

Two thoughts: Funds are explicitly mentioned as acceptable entities in 501(c)(3). For what non-educational purpose are endowments used? While some uses may seem tenuously connected to a given observer, all are determined to further the university’s mission in some way.

The underlying assertion is that universities must provide financial assistance to the majority of their students to be considered tax exempt. No citation is provided.

26 USC 501(c)(3) explicitly lists an organization operated exclusively for educational purposes as being tax exempt. Charitable organization is listed as a separate type.

Also worth considering is the example of Cooper Union: It charged no tuition and therefore offered no financial aid for most of its history. While one could argue that that is a form of aid so too would any other tuition reduction enabled by an endowment. This further suggests that universities do not set tuition in order to maintain tax exempt status.

Unsupervised refers to whether or not the dataset is being trained against anything. Think about the difference between: How many people will view this webpage? Divide these pages into 20 clusters? The first is supervised. The second isn't.

Deep learning refers to a particular type of a particular learning technique: Specifically a neural network that has many hidden (intermediate) layers. Deep learning can be used for either supervised or unsupervised learning.

Given that the company had been incorporated prior to the YC application, there are two possibilities: The original incorporation paperwork shows a 50:50 split or it contradicts the YC application. In the latter case, assuming that one party handled incorporation but made the other believe it said something different, wouldn't that constitute fraud?

Of course, it's also possible that they discussed resolving the equity when they split but never put it into writing...

Minor crashes are even more frequent than the article estimates. Thus, the real human accident rate is even higher. Probably between 1 in every 24000 and 87000 miles.

The VTI driving study[1] equipped 100 cars with sensors and was therefore able to measure all crashes experienced. It directly measured 1 crash per 24000 miles. If we extrapolate based on the 17.4% police report rate, that suggests 1 per 87000 miles.

[1]http://www.nhtsa.gov/DOT/NHTSA/NRD/Multimedia/PDFs/Crash%20A...

This completely ignores the perspective of those who bailed-out AIG. Whether their ultimate assessment of the ramifications of AIG's bankruptcy was correct (Which is important to understand for future crises) they feared the collapse of the financial system.

Bair's question: "Were the others really in danger of failing?" seemed obviously true to everyone involved in the crisis at the time. When AIG failed, Lehman had just declared bankruptcy and overnight lending rates between banks had sky-rocketed. Washington Mutual and Wachovia were on the brink of failure. Goldman and Morgan were both sufficiently dependent on the short-term funding markets that they could have been rendered illiquid next.

Looking in from the outside, it's easy to see this as insanity. All they needed to do was trust each other and most of the institutions would have been fine. But they lacked sufficient transparency to know which wouldn't be. While we can all imagine better solutions, making those things happen takes time. Put yourself in Bernanke's, Geithner's and Paulson's shoes. You can do something that will lead to be pilloried in the press or risk watching a repeat of the great depression with some probability you can't estimate.

The data has been available; acquiring it would have cost six figures until recently.

Details: The loan-servicers have the same data and many have provided it to Black-Knight (LPS/McDash) for years. http://www.bkfs.com/Data-and-Analytics/CorporateInformation/...

While you mention that Freddie provided data from 2006, the loan-level performance data is even more recent. The original releases were just origination info and thus worthless by themselves for risk assessment.

Areas where children are disproportionately less likely to be married than their same-income peers at age 26 will be reported as having lower income mobility in this analysis since it uses household income.

Average starting salaries:

   Stanford engineering: $74,467[1]
   Olin: $78,074[2]

[1]http://www.nerdwallet.com/nerdscholar/grad_surveys/top-salar... [2]http://www.olin.edu/content/results/