Hey if you haven’t read it yet take a look at this new blog, called Time for One More.
Gorgeous writing, and an inspiration for my new project on thinking about the elderly and technology.
One thing I call for in the essay is the teaching of ethics to aspiring data scientists, and yesterday some very cool professors from the Berkeley School of Information wrote to me and told me about their two classes on data science and ethics, one for undergrads and the other for graduate students. I seriously wish I could enroll in them!
Please tell me of other efforts in this direction if you know of them.
People who make their living writing and deploying algorithms like to boast that they are fair and objective simply because they are algorithmic and mathematical. That’s bullshit, of course.
For example, there’s this recent Washington Post story about an algorithm trained to detect “resting bitch face,” or RBF, which contains the following line (hat tip Simon Rose):
FaceReader, being a piece of software and therefore immune to gender bias, proved to be the great equalizer: It detected RBF in male and female faces in equal measure. Which means that the idea of RBF as a predominantly female phenomenon has little to do with facial physiology and more to do with social norms.
While I agree that social norms have created the questions RBF phenomenon, no algorithm is going to prove that without further inquiry. For that matter, I don’t even understand how the algorithm can claim to understand neutrality of faces at all; what is their ground truth if some people look non-neutral when they are, by definition, neutral? The answer entirely depends on how the modeler creates the model, and those choices could easily contain gender bias.
So, algorithms are not by their nature fair. But sometimes their specific brand of unfairness might still be an improvement, because it’s at least measurable. Let me explain.
Take, for example, this recent Bloomberg piece on the wildly random nature of bankruptcy courts (hat tip Tom Adams). The story centers on Heritage, a Texas LLC, which bought up defaulted mortgages and sued 210 homeowners in court, winning about half. Basically that was their business plan, a bet that they’d be able to get lucky with some judges and the litigation courts because they knew how to work the system, even though in at least one case it was decided they didn’t even have standing. Here’s the breakdown:
Now imagine that this entire process was embedded in an algorithm. I’m not saying it would be automatically fair, but it would be much more auditable than what we currently have. It would be a black box that we could play with. We could push through a case and see what happens, and if we did that we might create a system that made more sense, or at least became more consistent. If we found that one case didn’t have standing, we might be able to dismiss all similar cases.
I’m not claiming we want everything to become an algorithm; we already have algorithmized too many things too quickly, and it’s brought us into a world where “big data blacklisting” is a thing (one big reason: the current generation of algorithms often work for people in power).
Algorithms represent decision processes that are vulnerable to inspection more than most human-led processes are. And although we are not taking advantage of this yet, we could and should do so soon. We need to start auditing our algorithms, at least the ones that are widespread and high impact.
I’m excited to be traveling to Harvard next Monday to give a talk at the Data Privacy Lab. The projects going on at the Data Privacy Lab are privacy-related: re-identification, discrimination in online ads, privacy-enhanced linking, fingerprint capture, genomic privacy, and complex-care patients.
My talk will not be entirely focused on privacy – it will basically be a somewhat technical version of my book followed by my proposals for technological tools that could address the problems associated with opaque, widespread, and destructive algorithms (my definition of a “Weapon of Math Destruction”. Specifically, I want to examine the question of how we understand a black-box algorithm in terms of measuring its outputs (as opposed to scrutinizing the source code).
The Data Privacy Lab is run by Latanya Sweeney, a hero of mine who did great work in detecting online discrimination in Google ads among other things. I’m hoping to meet her first because it’s always nice to meet your hero but also because, as the chief technologist at the Federal Trade Commission, she can give me perspective on the kind of technological tools that regulators such as the FTC and the CFPB might actually adopt (or develop).
In other words, I don’t want to spend 4 years developing tools that nobody would use. On the other hand, I have the impression that they generally speaking don’t know what kind of tools are possible.
I’ve been unimpressed with the recent coverage of the Flint water crisis. The overall message is that there’s been a “run of bad luck” but that certain generous people and corporations are coming to the rescue. If you believe the reports, we should be grateful for all the water bottles being flown in from Nestle and Walmart, and we should rest assured that water filters are being handed out and installed, even though they are inadequate.
In many of the articles on Flint, the switch from Detroit to the Flint River is mentioned, as is the concept of water as a human right, but not much more is explained. Specifically, there are two important questions left unanswered. First, how did this happen? And second, where else is it going to happen?
When you think about how Flint residents got into this situation, it’s critical to remember it was directly caused by a suspension in democracy. It was an emergency manager appointed by Michigan Governor Rick Snyder that made the switch to the Flint River as a water source. I’ve talked a bit about which municipalities get their democratic powers taken away; turns out that process often involves poor people of color. The entire point of emergency management is to remove accountability from the actors who put people’s lives in danger under the guise of saving money. Rick Snyder is, unbelievably, still in office.
Speaking of money, what’s the larger story here? It’s that, as a country, we can’t seem to pony up the resources to keep up our infrastructure, especially when it comes to water. A 2012 report by USA Today found that water prices had doubled in a quarter of the cities surveyed since 2000. This is because federal funding for water and waste systems have been reduced by 80% since 1977. And that would make sense if our water infrastructure were robust, but it’s not. In fact it’s in crisis, and we’d need $1 trillion to update it. The result is widespread crappy water, expensive water, and privatized water system disasters. We just let it rot at the local level, in other words, and deal with it – or not – in the most expensive ways, when it’s already an urgent situation.
Guess where the pipes are the oldest and most decrepit? You guessed it, where poor people live. When we ignore infrastructure we are inviting yet another punitive tax on the poor, and as it happens, a life-long debilitating level of lead poisoning.
So, let’s answer the second question: where else is this going to happen? The answer is pretty much everywhere unless we get our priorities straight. And I’m not talking about water bottles.
Hey there’s finally been a New York Times column that agrees with me about how to raise kids, so I’m totally going to blog about it.
Seriously, I know that I’m 100% biased, as is anyone who tells you how to raise your kids, but I think Adam Grant has hit upon the perfect explanation of how I think about things in his recent column, How to Raise a Creative Child. Step One: Back Off.
The dumbed down version goes like this: yes, we all know it take a huge amount of practice to get good at the violin. But that doesn’t mean you should force your kids to practice all the time so they’ll become musicians. That’s confusing causation with correlation, the most common of all parental crimes. Instead, ask your kids to be ethical and trust them to find their passion.
The idea is if you give them a strong education in ethics, and then set them free within that framework, they might just decide they love the violin. If they do, then as long as you support their passion, they might just practice all the time and become musicians.
I’ve written a bunch about this exact issue over the years, because although I played the piano as a child, I don’t encourage my kids to play instruments. Because they aren’t begging for it like I did.
To be fair, this isn’t because I’m nervously trying to construct creative kids and want the conditions to be perfect. Mostly it’s common sense. Said plainly, why would I pay for expensive lessons that they don’t want? Why would I set myself up to remind them to practice when they could care less? It sounds like torture for everyone involved, and I honestly don’t understand parents who do it.
I grew up in Lexington, Massachusetts, a hotbed of striving upperly-mobile parenthood, and I was absolutely surrounded by kids – especially second-generation Asian kids – who were being forced to display precocity in all kinds of ways. These kids were miserable, and they hated their violins and cellos. Not all the time, and not in every way, but let me say it like this: very few of them still play music. (Whereas I do, and by the way my bluegrass band has a gig, stay tuned.)
I know, it’s not a lot of evidence, but I still think I’m right, because it’s parenting and people are totally irrational when it comes to this kind of thing, so bear with me, and read the references in Adam Grant’s piece as well, maybe they’re scientific-y.
Of course, it all depends on the definition of creative, which is of course not obvious and I could easily imagine the result changing depending on how you do it. Not to mention that “creativity” isn’t the only thing you’d want from your children. In fact, it’s not my personal goal for my kids to be creative. If I had to choose, I’d say I want my kids to be generous and ethical.
Here’s a bit more background on this very question. a Harvard Education School report called THE CHILDREN WE MEAN TO RAISE: The Real Messages Adults Are Sending About Values that found the following:
About 80% of the youth in our survey report that their parents are more concerned about achievement or happiness than caring for others. A similar percentage of youth perceive teachers as prioritizing students’ achievements over their caring. Youth were also 3 times more likely to agree than disagree with this statement: “My parents are prouder if I get good grades in my classes than if I’m a caring community member in class and school.” Our conversations with and observations of parents also suggest that the power and frequency of parents’ daily messages about achievement and happiness are drowning out their messages about concern for others.
When I read this report I performed an exceptionally biased poll in my own household and made sure my kids knew what’s up. And they all do, most probably because I am not forcing them to practice the piano.
Brussels is a pretty nice place for a hellhole (according to Trump). I got here early yesterday and walked around; obviously I bought a bunch (technically an asston) of chocolate and took pictures of impudent statues.
I know this sounds entirely unhistorical and arrogant, but I can’t help thinking that Brussels was created out of some indulgent American fantasy of Europe that confused Paris and Amsterdam and added a bunch of chocolate stores, beer, and waffles. Oh, and gold leaf.
It’s a great city; possibly it’s replaced Amsterdam as the place I’d like to live if I moved away from New York (which will never happen). It’s pedestrian dominated, there are plenty of sex shops, and the recycling bins are covered with graffiti. In other words, it’s got the right values and it’s not overly sanitized. Trump’s got it wrong once again.
I’m here for an annual conference called CPDP, which stands for Computers, Privacy, and Data Protection. This morning I attended a super interesting panel on privacy and the world’s poor. In that panel I learned about an algorithm being used to sort unemployed people in Poland. As is typical of many of the algorithms I’m interested, it’s both entirely opaque and high impact; the open information laws also don’t apply for inscrutable reasons.
Later today I’ll be on a panel in which we’ll discuss software tools that investigate privacy and data protection in the real world. Besides me, the people on the panel are working within the context of European privacy and protection laws, which are both very different and much more protective than we have in the states (although the UK is an exception). I will surely learn a lot, both about how people think about data and privacy over here and what the obstacles are to enforcing the strong laws.