On Machine Learning and the Ethics of Computation

I was reading a recent post on machine learning from one of my favourite technical writers (Julia Evans) and was inspired enough to comment that I signed up for Disqus under my own name for a change 🙂
However, it turns out that the post was closed for comments, so I thought that I’d write up my thoughts here.  I’m pretty busy today so I’m just going to paste it as a quote rather than edit it into a more ‘first-person’ writing style.

This is a pretty interesting topic overall – I have largely avoided machine learning b/c of a lot of concerns about how it’s used (and its results assumed to be ‘normal’ or ‘correct’) but recently decided that that’s the wrong approach. Instead, I’m going to learn more about it so I’m in a better position to critique how it’s used and point out implicit assumptions and biases. To that end, I’ve signed up for the Stanford course that just started.

Even in the first lesson in that course, I saw some interesting examples that made my skin crawl. I think the biggest issue I have with it is (and I am JUST starting to learn) is that there’s an implicit assumption that all relevant information is externally observable and that conclusions drawn from objectively measurable data/behaviour will be correct. I’m fine with that when it involves some kinds of events, but I get very uncomfortable when we’re applying it to humans. So much of human motivation is invisible/intuitive that leaning so heavily on machine learning (which necessarily relies on events that are can be observed by others & fed into algorithms) leads to things like, as you say, the Target pregnancy issue. There are many other ‘positive feedback’ effects of assuming that reinforcing/strengthening conclusions based on visibly available data that are detrimental – gender-segregated toys is a primary one. “65% of people buying for girls bought X, so when someone is shopping for a girl, we’ll suggest X. Look! Now 75% of people shopping for girls bought X – we were right! MOAR X for the girls!” [eventually, the ‘for girls’ section is nothing but X in various sizes, shapes, and colours (all colours a variant shade of pink)]

Another ML issue that came up for discussion when I was working at Amazon was: some people consider LGBTQ topics to be inappropriate for some readers, so even if someone had actively searched for LGBTQ fiction, the recommendation engine was instructed to NOT suggest those titles. That has the effect of erasing representation for marginalized people and increasing isolation among those who are already isolated. In fact, one could argue that one of the things that ML does best is erase the margins (obviously, depending on how it’s implemented, but in the rush to solve all problems with it, these types of questions seem to be ignored a lot).

I mentioned positive feedback loops before. The analogy I have in my head is: ML type algorithms (unless you build in randomness & jitter) amplify differences until you end up with a square wave – flat and of two polar opposite values. Negative feedback loops lead to dynamic yet stable equilibriums.

I mean, it’s obviously here to stay, and it clearly has some very significant beneficial use cases. But there are a lot of ethical questions that aren’t getting a lot of attention and I’d love to see more focus on that over the technical ones. Thank you for mentioning them 🙂

The more time I spend in this industry, the more I believe that the one Computational Ethics course I took back in my CS degree wasn’t nearly enough, and that we could really use a much broader conversation in that area. [To that end, I’ve also signed up for some philosophy courses to go with the ML one ;)]

The Last Day

2012 Tulip Festival @ Agassiz, BC
Photo by GoToVan / CC BY

Today, I have lived as many days on this planet as my mother ever did.

[That is, assuming Excel’s date math handles leap years properly. Yes, I made a spreadsheet.]

There are many things I could write about my mother’s life, and our family’s life together, but I don’t feel comfortable writing about people I love without their knowledge and permission. The only thing I will say about that is that I don’t think she spent very much of her life doing what she wanted to do.

I’ve spent a lot of my life being angry. Reactive, hurt, feeling powerless and wild inside.

Last spring I realized that my next birthday would make me the same age as my mother was when she died and I began to really think about death – MY death. How it was a certainty. How I didn’t know when it will happen. Not just for me, but for everyone. And a while after that, I actually started to understand it. Not just rationally (of course we’re all going to die, everyone knows that) but emotionally.

Last spring, I was working in an office at a company I didn’t respect very much and becoming increasingly miserable (and this for a person with a pretty damn high baseline miserableness). I saw friends withdrawing from my constant perspective of disgust and complaint. I heard that someone I used to work with said he wouldn’t want to work with me again because I complained too much (at that time, he wasn’t wrong). The feeling I had at all times is “EVERYTHING IS WRONG, PLEASE MAKE IT STOP BEING WRONG.” I changed roles at the company I didn’t respect very much – hoping that maybe it was just the aspects of the job I was doing that were the problem.

It wasn’t. And when I realized that, I knew I was never going to be happy there. So last spring, I quit my job.

Let me just step out of this story for a minute to note a few things – I know that not everyone has that option. I’m well educated and have a marketable set of skills, I have a very supportive partner. I don’t have children that I need to support. I am financially secure. Not everyone could do what I did and I am grateful for the things that gave me that option, although I wasn’t particularly grateful for them at the time. I didn’t really understand that I was doing much better than a LOT of people are because I was so wrapped up in pain and anger and self-hatred. I only began to recognize how lucky I am to have those supports when I relaxed and let myself use them.

That’s part of why I quit: I felt like I was trapped, like I HAD to stay in the career that I was in. That’s the rule, right? If you’re what other people would describe as successful, even if you don’t feel it yourself, then it would be ungrateful to not value and cling onto those things that ‘they’ all want. [A contrasting perspective is that means you’re ‘taking up’ a job that someone else might actually want & be happy with, so as a conservation of resources dimension, quitting is a net happiness gain]. Although I had made some choices in my life that had led me to where I was, there were a lot more situations where I just went along with whatever the next easiest step was. I certainly wouldn’t say I was where I was because I planned to be there.

Three realizations are what got me moving. Like I said above, I really understood that I am going to die. Not (necessarily) imminently, but at a time most likely not of my knowing or choosing. Secondly, I realized didn’t need to be making as much money as I was. That based on my own situation, I have a lot of freedom to explore new alternatives and still be ok. And finally, I realized that I didn’t want the company I was working for to be successful. That I had enough concerns about its values and business practices, and the impact it was having on society, humanity, and the planet that I wasn’t rooting for it.

[The day you realize that you don’t want the company paying you to be successful is the day you think to yourself “If I keep accepting money from them, how is that different from stealing?” Or it was for me, anyway.]

So I quit.

I’ve spent the last 9 months working on changing my perspective and attitude towards this short life of mine. I have been trying new things, learning, discarding, keeping, adapting. Often I meditate. Sometimes I exercise. I meet new people. I reconnect with people I’ve lost touch with. I’m gradually learning what is good for me and what I need to avoid. Sometimes I make mistakes and do the things that make me feel angry or hopeless, but increasingly I’m recognizing when that happens and more importantly, I’m starting to react with a sense of humour and compassion towards myself when that happens. It’s not just an immediate descent into self-loathing and shame anymore. That change in myself is more than enough to pay for this past year of not working.

So today I am going to a meet-up to learn how to come up with a talk to give at conferences. I’m going to a friend’s place for a birthday dinner. I’m going to meditate and, weather co-operating, I’m going to go for two 45 minute walks. I’m also going to think about the last day my mom was alive and remember that I’m going to die and know that that’s ok. I have a lot of choice in how I live in the meantime.

We need another dimension for bra sizes

I’m guessing that most people are familiar with the standard components of a bra size (at least in Canada & the US – I don’t know international sizes). There’s band size – a number, ostensibly in inches (e.g. 38) that represents the circumference of the ribs below the breasts, and there’s cup size – a letter or string of letters (AA, A, B, C, D, DD, etc*) which is based on the difference between the band size and the circumference around the broadest part of the breast (generally right over the nipples).

I needed some new bras and since it had been a long time since I had bought any, I figured I’d get sized. I went to Victoria’s Secret (since I assumed they’d know what they were doing) and they did the standard measuring. I picked out some bras I like (I prefer “T-shirt style” – no visible seams + underwire).

 

They’re torture.

 

The problem is that those two degrees of freedom, band & ‘cup’ don’t adequately cover all possible shapes, especially where there’s an underwire.
We also need ‘radius’.

Radius

Let’s say we have two people with the same circumference under the breasts & across them. Let’s also say they aren’t clones, and actually are shaped fairly differently:

Note that the underwire curve of each person is rather different.

Different underwire curves
These are not the same

So guess what happens when we don’t take that into account when building bras?

It’s 2015. We can do better.

*interestingly, some schemes stop adding new letters at D and just keep adding more Ds, (i.e. DDD=E, DDDD=F)