Profile cover photo
Profile photo
Adam Liss
Engineer, Puzzle Fanatic, Incurable Punster. Armchair cryptographer, security and privacy advocate. Insatiably curious about the world and its inhabitants.
Engineer, Puzzle Fanatic, Incurable Punster. Armchair cryptographer, security and privacy advocate. Insatiably curious about the world and its inhabitants.


Kid: So apparently I'm a Buddhist.

Me: Oh! Wait! I know this one! It's because, when you do origami, you always fold the paper 8 times! Right?

Post has shared content
"Dad, why do I need to learn math?"

So you'll know better when someone tries to tell you how "unfair" it is to help under-represented groups at the "expense" of the majority.

Thanks for taking the time to write a clear and indisputable analogy, Trey!
The supermarket parking lot metaphor for affirmative action

I wrote this reply to another comment thread, but figured it could do with a few more eyes than the final four still paying attention to an already-dead-horse-beaten comment thread.

To contextualize: when I posted below, I had stipulated, without conceding, my correspondent's claim that biological differences between sexes conferring differences in the ability to be a techie or in tech leadership were real; that men had "greater variance" in ability and interest that resulted in a distribution such that, when "creaming the top" for the "best" regardless of sex, you naturally would get an over-representative proportion of white and Asian men; and that diversity programs were only a drag on non-diversity hires, that there was no advantage that accrued to them as a result.

All of which—to be utterly clear—I vehemently believe are false claims. But I wanted to concentrate on one particular aspect—the idea that the "advantage" conferred on minorities from diversity programs can't be enough to make a difference unless the "disadvantage" conferred on non-minorities is significant, which the parking lot metaphor helps to make much more intuitively understandable.

My probability calculations below are undoubtedly off—I modeled it as discrete "parking space open/parking space closed" events rather than more correctly as durational flows (and totally neglected the ebb/flow cycle that is how parking capacity really works, apparently, as people don't bother with very full parking lots and go elsewhere, or are attracted to a store with lots of easy parking), but I did some floor/ceiling calculations to absurdity to ensure that even if I am off, I'm not off by enough to invalidate the overall argument. (Since it's a metaphor, anyway, I don't care it doesn't exactly model a real parking lot....)

Also, I was working from two different ADA documents, one of which was using a 400-space lot as its model, the other a 200-space lot, so I may have doubled or halved a figure somewhere. So if you spot an arithmetic or modeling error, let me know in comments.

— Original comment: —

The fact that there can be a help in one area for minorities, one that affects the majority so little as not to matter in any real way to any individual "disadvantaged" by that help, yet makes an enormous difference to the minority helped—that this is even a possible thing seems to elude you. You seem to see only a zero-sum game here, but miss that when the numbers are large and the disparities are huge, zero-sum games can be tweaked in ways to make them immensely more fair to all. (It, in fact, isn't a zero-sum game, since a company starved for talent will hire both the man and the woman if both are qualified. But even stipulating it is at some macro level.)

It seems you're unfamiliar with the large-scale statistical economics of affirmative action. Understanding them require some specialized knowledge that it's not possible to quickly get up to speed on. So let me change the framework we're working in entirely, and perhaps it will clear up the mathematical "disparate multiplier" effect a bit.

Let's imagine a supermarket has room for 200 standard parking spaces for customers. As required by law, it sets aside space, close to the entrance, to allow for 4 wider handicapped spaces—one 11′ wide space for a van with wheelchair hoist, the other three the same width as a normal "compact" space—8 feet—but with each pair having a 5′ wide access aisle with ramp between them, equaling 50 feet total; a normal "full-sized" parking space is 10′ wide¹.

Local code differs on whether such parking lots can vary between 8′ compact and 10′ full-sized spaces and in what proportion, but let's assume the most extreme case: the area for the handicapped spaces would otherwise have otherwise gone to five compact 8′ spaces and one full-sized 10′ space, with no access aisles. So, six non-handicapped parking spaces have been replaced with four handicapped spaces.

And so, as constructed, there are 3% less spaces available for drivers who do not have a handicapped permit. This is unquestionably a disadvantage, particularly since the numbers as described were chosen to try to ensure there will almost always be at least one handicapped space open, even when the rest of the parking lot is full. (That, in turn, may be described as an "unfair advantage" of the minority: access to the parking let even when it is "full".)

The chances that the lot will be full when you, without a handicapped permit, arrive, is quite low; and if it is, you have 194 spaces that may open up.

Had those spaces not been set aside, the "reduced disadvantage" to the majority of the additional capacity would be very low. But the impact on the handicapped driver would be enormous. Let's make some rough assumptions, some intentionally incorrect but which serve as convenient ceilings or floors:

• A typical 200-slot lot where there is a single parallel row farthest from the store, 2 double rows perpendicular to the store, plus two half rows at each end for 6 total perpendicular rows, but no "store-adjoining" parallel row
• That handicapped drivers will be able to use the slot at the store-facing end for each perpendicular row of slots where they can use pavement or sidewalk as maneuvering room (note that in most actual lots this will not be the case for one or more of those slots because of entrances/exits, shopping cart storage, greenway requirements, etc—but in the most liberal estimate, there will be 6 slots they might use. (The fact this is the same number as the set-aside formula uses is not coincidental.)
• An, on average, constant churn
• That handicapped drivers take exactly as long to get out of their cars, shop, and get back in their cars (which obviously isn't true; on average, they'll take longer, particularly if they have to deal with using thoroughfare or lot corners as maneuvering room)

Assuming that the average space empties 30 minutes after becoming occupied, when the lot is full, an able-bodied person who can take any space but will be able to reach 20% of the lot before another driver looking for a space beats them to it will be have a > 50:50 chance of being able to find a space in a full lot with such a churn rate within about 90 seconds. Remove six spaces from their available pool in the lot, and this figure doesn't change by more than a couple seconds at most.

On the other hand, a handicapped person who must take one of those six spaces it's possible for them to use will have to wait, on average, 12 minutes. And that's only if they have the ability to get to any of the six and don't have to compete with any other drivers! If they could also only reach 20% of the lot before another driver got the slot, their likely wait time increases to hours.

Only your deranged drunk uncle would begrudge the creation of these spaces. The reduced advantage to the majority is so small compared to the great increased access by the minority, that the endeavor is obviously worthwhile on balance. Both kinds of customer's money spends equally well, so the company's cost is a few signs, a little paint and labor for markings and ramps, and, on average, 1.5% less spaces for customers.²

To put it back into the realm we were talking about before: even if hiring were a zero-sum game (and, again, at a company like Google, it's not—they'd ideally hire anyone they can get who can pass their high standards), the chance that any single white or Asian man was negatively impacted by diversity efforts can be tiny, even when the impact to any single "diversity hire" is huge. "Both kinds of customer's money spends equally well": both the new hires helped by these efforts, and those not, will be roughly equally valuable to Google.

Informally, your "variance" argument actually helps a great deal here: if you are right, and the variance between sexes is different such that there are proportionally more "brilliant" (very top of the top in ability/interest) men than women, then the men who are likely to be hurt are usually going to be the ones who would have been less likely to have been hired anyway, because they'd be "good enough" but not "top of the outliers". Adding diversity hires—giving help to get the person hired if and only if they are good enough — not with the aim to have "stupider" or less-capable diversity hires replace "smarter" or more-capable white or Asian men—will typically replace a man from the "bottom" half of the ability window with a woman who is, on average, going to be right in the middle.

If you assume the effect is "frictional" rather than "coincidental"—informally, that without affirmative action efforts, minority hires are not just generally more unlikely to be hired than non-minority hires, but merely that the "difficulty level" has been ramped up such that the cutoff level of ability/interest is higher—then the women hired won't average right in the middle—they may average above or below the middle depending on variables not being considered—but they'll still be well over the bar for men. (Unless you're also necessarily claiming that either your allegedly more-variant male curve or allegedly less-variant female curve, or both, also have exotic distributions—and what could possibly be evidence for that?)

¹ This was formerly the actual calculation done for big-box stores under the ADA and most state laws in the U.S. The current calculation is more complex, but stores built before 2012 can continue to use these, so I'll use them too since it's easier to understand.

² Some of these calculations have been cribbed from The 0.75% is assuming that, in fact, one handicapped space will be empty at all times, so the six possible spaces have been replaced not with four, but effectively with three spaces.

Post has attachment
"If you can read the sign at the front of the bus, it's too late."

Post has attachment
This is how to combat the embarrassment that is the ignorant, lying, totally unprepared Trump organization. This is how journalism should work.

The entire interview is priceless, but the last 10 seconds speak volumes.

Post has attachment

Post has attachment
Because of course that's what Trump's lawyer would do.

Post has shared content
I wonder if the +Chicago Tribune​ realizes that "Trump retweets cause" stands alone as a perfectly good sentence, making for headlines that are hard to parse, and making forehead lines across my skull..

But that's not likely to be a cause Trump retweets, 'cause Trump retweets are just silly.
Trump retweets cause stir as he attempts to move past Charlottesville uproar

President Donald Trump appears to have mistakenly retweeted a message from one of his critics saying "he's a fascist."

Trump deleted his retweet Tuesday after about five minutes, but not before the message sent to his 35 million followers racked up a big response.

Post has attachment
We should take a lesson from this German town that decided to turn the annual neo-Nazi march into a charity event.

"Without the marchers’ knowledge, local residents and businesses sponsored the 250 participants of the march on 15 November in what was dubbed Germany’s “most involuntary walkathon”. For every metre they walked, €10 went to a programme called EXIT Deutschland, which helps people escape extremist groups."

Post has attachment
Y'know three things that made the Charlottesville protest different from previous hate rallies?

1. Torches make for good lighting.

2. It's almost impossible to attend an event without your image being recorded, especially when people are making a point of documenting it.

3. Twitter.

@YesYoureRacist is determined to make white supremacists afraid again. It's interesting to see how quickly the racists who are accustomed to making death threats start to yell foul when the tables are turned. Reminds me of a certain POTUS who seems to support them.

Post has attachment
I think it's about time to share this again. We could all use the humor and perspective that the Rev. Wade Watts inflicted on Johnny Lee Clary and the KKK.

I never get tired of hearing an ex-Klan leader tell his story about being utterly defeated by one gentle black man.
Wait while more posts are being loaded