• Actually inclusive engineering

    I want to talk about ethics, diversity, and inclusion in engineering, how we often miss the mark, the impact that has, and the changes we can make to truly bring change from the inside out. My goal is to explain why this is important, and show you some examples where a simple decision resulted in a barrier for someone.

    Why does this matter? Why is it important to be thinking about ethics when we’re developing software? Because software (apps, websites, etc) is becoming the fabric of society – increasingly it is involved in everything we do, from shopping for groceries to online banking to socialising. There is very little in our lives now that is not touched, in some way, by software.

    As we integrate software into more and more areas of our lives, we are also increasingly turning to automated and predictive solutions to perform tasks that were once manual. We are asking computers to do “human” things, more open-ended “thinking” tasks, but computers aren’t human. Most people, when they think of AI, think of something like Data from Star Trek. The reality of AI however is that we have “narrow” AI – models which are trained to do a specific thing and that thing only. These models are unable to add context to their decisions, to take additional factors into account if they are not in the data model, or even to question their own biases. It takes in data, and returns data.

    Lastly, we often spend a lot of time discussing how we will implement something, but perhaps not as much time discussing whether we should implement something. We know that it is possible to build software which will have in-app purchases, and that it’s possible to incentivise those in-app purchases so that they are very attractive to app users. We have seen that it is possible for people to target this marketing towards children – answering the “can”, but not addressing the “should we?”

    When I say we should consider the “should” rather than the “can”, what do I really mean? I’m going to show some real world examples where decisions made during product design ripple out into the world with negative effects. In each of these examples, there probably wasn’t malicious intent, but the result is the same – a barrier for an individual. Most of these examples are not due to programming errors, but by (poor) design.

    Have you ever accidentally subscribed to Amazon Prime?

    Do you know what a dark UX pattern is? You’ve probably encountered one, even if you’ve never heard the term. Have you ever accidentally opted-in to something you meant to deselect, found an extra charge on a bill that you didn’t even realise you had signed up for? Have you ever tried to cancel a service, only to discover that the button to “cancel” is hidden below confusing text, or the button that looks like a cancel button actually signs you up for even longer? How about accidentally signing up to Amazon Prime when you just wanted to order a book? These are dark UX patterns – design changes that are designed to trick the user. They can be beneficial for the person who implements them, but usually to the detriment of the user. In the image above, we see two buttons to add your tickets to the basked. An optional donation can be added with the tickets, but the option to add without donation is much harder to read. It also points backwards, implying visually that this would bring you back a step. Is the value of this donation worth the confusion? Is this ethical? Should a donation be opt-out or opt-in?

    Have you ever been told that your name is incorrect?

    Your name is one of the first things you say to people you meet, it is how you present yourself to the world. It is personal and special. But what if you are told that your name is incorrect due to lazy or thoughtless programming every time you try to book an airline ticket, access banking, healthcare, or any number of services online? A multitude of online forms fail to support diacritical marks, or declare that names are too short or too long based on simple biases and the incorrect assumption that everyone has a first and last name that looks like our own. Instead, we should be asking – do we need to separate people’s names? Why do you need a “first” and “last” name? Could we simply have a field which would accommodate a user’s name, whatever form that takes, and then another which asks what they prefer to be called?

    Let’s talk about everyday barriers

    We’ve never been more aware of handwashing, and a lot of places are using automatic soap or hand sanitiser dispensers to ensure that people stay safe without having to touch surfaces. But what if they don’t work for you? Many soap dispensers use near-infrared technology, which sends out invisible light from an infrared LED bulb for hands to reflect the light back to a sensor. The reason the soap doesn’t just spill out all day is because the hand acts to bounce back the light and close the circuit, activating the soap dispenser. If your hand has darker skin, and actually absorbs that light instead, then the sensor will never trigger. Who tests these dispensers? Did a diverse team develop these or consider their installation?

    Why don’t zoom backgrounds work for me?

    If you’re like me, you’ve been using meeting backgrounds either to have some fun or to hide untidy mixed working spaces while adapting to working from home during this past year. When a faculty member asked Colin Madland why the fancy Zoom backgrounds didn’t work for him, it didn’t take too long to debug. Zoom’s facial detection model simply failed to detect his face. If you train your facial detection models using data that isn’t diverse, you will release software that doesn’t work for lots of faces. This is a long running problem in tech, and companies are just not addressing it.

    Why can’t I complain about it on twitter?

    On twitter, the longer image I shared was cropped….to include only Colin’s face.

    When Colin tweeted about this experience, he noticed something interesting with Twitter’s auto-cropping for mobile. Twitter crops photos when you view them on a phone, because a large image won’t fit on screen. They developed a smart cropping algorithm which attempted to find the “salient” part of an image, so that it would crop to an interesting part that would encourage users to click and expand, instead of cropping to a random corner which may or may not contain the subject. Why did twitter decide that Colin’s face was more “salient” than his colleague’s? It could be down to the training data for their model, once again – they used a dataset of eye tracking information,  training their model to look for the kinds of things in an image that people look at when they look at an image. Were the photos tested diverse? Were the participants diverse? Do people just track to “bright” things on a screen. It certainly seems there was a gap and the end result is insulting. Users tested the algorithm too, placing white and black faces on opposite ends of an image to see how twitter would crop them. The results speak for themselves. Twitter said they tested for bias before shipping the model….but how?

    This impacts more than social media. It could impact your health

    Pulse oximeters measure oxygen saturation. If you’ve ever stayed in a hospital chances are you’ve had one clamped to your finger. They use light penetration to measure oxygen saturation, and they often do not work as well on darker skin. This has come to particular prominence during the pandemic, because hospitals overwhelmed with patients started spotting differences in oxygen levels reported by bloodwork and by the pulse ox. This could impact clinical treatment decisions, as they report higher oxygen saturations than are actually present. This could lead to a delay in necessary clinical treatment when a patient’s o2 level drops below critical thresholds.

    This could change the path your life takes

    COMPAS is an algorithm widely used in the US to guide sentencing by predicting the likelihood of a criminal reoffending. In perhaps the most notorious case of AI prejudice, in May 2016 the US news organisation ProPublica reported that COMPAS is racially biased. While COMPAS did predict reoffending with reasonable accuracy, black people were twice as likely to be rated a higher risk but not actually reoffend. The graphs show that risk scores are very far from normal distribution – they are skewed heavily towards low risk for white defendants. In multiple real life examples from the ProPublica analysis, the black defendant was rated as a higher risk, despite fewer previous offences, and in both cases that individual did not reoffend, although the “lower risk” defendant did. 

    And these are, sadly, just selected examples. There are many, many, many more.

    Clang, clang, clang went the trolley

    As we come to the end of the real world examples, I want to leave you with a hypothetical that is becoming reality just a little bit too fast. Something that many people are excited about is the advent of self-driving cars. These cars will avoid crashes and keep drivers safe and allow us to do other things with our commute. But….

    Have you ever heard of the trolley problem? It’s a well known simple question that is often used to explore different ethical beliefs. In case you aren’t familiar with this yet, the picture above is a fair summary. Imagine you are walking along and you see a trolley, out of control, speeding down the tracks. Ahead, on the tracks, there are five people tied up and unable to move. The trolley is headed straight for them. You are standing some distance off in the train yard, next to a lever. If you pull this lever, the trolley will switch to a different set of tracks. However, you notice that there is one person on the side track. You have two options:

    • Do nothing and allow the trolley to kill the five people on the main track.
    • Pull the lever, diverting the trolley onto the side track where it will kill one person.

    What is the right thing to do?

    The tricky thing is that there are numerous ways to try and decide what is right, and there isn’t really a right answer. As humans we can perhaps use more about the context to aid us in a decision, and we can take in all of the information about the situation even if we have not encountered the situation before, but even then we still can’t always come to a right answer. So how do we expect a smart car to decide?

    While we might not see a real life trolley problem in our lifetimes, the push towards self driving cars will almost certainly see a car presented with variations on this problem – in avoiding an accident, does the car swerve to hit one pedestrian to save five? Does it not swerve at all, to preserve the life of the driver? Given what we know about recognition software as it currently stands, will it accurately recognise every pedestrian?

    How will the car decide? And who is responsible for the decision that it makes? The company? The programmer who implemented the algorithm?

    I don’t have an answer for this one, and I’m not sure that anyone does. But there is a lot that we can do to action inclusive and diverse programming in our jobs, every single day, so that we remove the real barriers that I’ve already shown.

    What can we do?

    First and foremost, diversity starts from the very bottom up. We need to be really inclusive in our design – think about everyone who will use what you make and how they will use it, and really think beyond your own experience.

    Make decisions thoughtfully – many of the examples I’ve shown weren’t created with malicious intent, but they still hurt, dehumanised, or impaired people. Sometimes there isn’t going to be a simple answer, sometimes you will need to have a form with “first name” and “last name”, but we can make these decisions thoughtfully. We can choose to not “go with the default” and consider the impact of our decisions beyond our own office.

    Garbage in, garbage out – if you are using a dataset, consider where it came from. Is it a good representative set? Is your data building bias into the system, or is it representative of all of our customers?

    Inclusive hiring – when many diverse voices can speak, we spot more of these problems, and some of them won’t make it out the door. Diverse teams bring diverse life experiences to the table, and show us the different ways our “defaults” may be leaving people out in the cold.

    Learn more – In the coming days and weeks, I’ll be sharing more links and some deep dives into the topics I’ve raised above, because there is so much more to say on each of them. I’m going to try and share as many resources and expert voices as I can on these topics, so that we can all try to make what we make better.

  • Lots of things make you angry

    Picture the scene: a typical lunch room in an office. There’s people from all teams gathered around tables, people are joining in and out of conversations as they finish their lunch or as the topic changes.

    A coworker mentions that he fasts all the time. I joke to someone else at the table that I could never do that because I get “hangry” (angry because I am hungry).

    My male coworker immediately says “lots of things make you angry, don’t they?”

    This coworker has a history of pushing boundaries that I set, including my one hard and fast meeting rule – unless I have asked you to do so, or you are presenting at this meeting, do not bring your laptop to my meeting (or at the very least do not have it open). I have instituted this rule because in my experience, when running certain types of meeting, if people have their laptops open they will browse, work, or otherwise not pay attention, and require me to have to repeat things multiple times. It’s inefficient and disrespectful to whoever is presenting. I make this rule clear and I enforce it equally. This coworker repeatedly brought his laptop anyway, refusing to close it, only closing it halfway then peeping in through the partly open gap, etc.

    Back to the lunch room, I am surprised that the coworker has decided to make such a statement in the lunch room, in front of all my colleagues, when it is actually untrue. I, knowing that I had just reminded him the other day about my laptop policy, try to brush it off with another joke like “ha, yes, I do get grumpy when people bring laptops to my meetings”.

    “Lots of things make you angry, don’t they?”

    And then he starts to list times he feels like I have been angry in the office.

    I interrupt him to ask “are you keeping a list?!” and he says “yes”. I say “that’s actually pretty creepy” and then attempt to just start a conversation with another person.

    I don’t need a rundown of times I’ve been “angry” from a man who threw a tantrum one day because I asked that developers set their Jira stories to “in progress” when they were “in progress”.

    Notable is that, actually, my coworkers haven’t seen me angry at all. The one time I have been genuinely angry at a work situation, I left the office for a few minutes for a brief walk to manage the feeling. My coworkers have seen me express disappointment that a commitment wasn’t met. They have seen me gently and then more firmly enforce a policy that they agreed to and that I have made clear. They’ve even seen me make clear at the start of a meeting that any less-than-cheerful tone they may sense from me is a holdover from a previous difficult meeting and that it is not a reflection of them, and that I’ll be working to ensure it doesn’t impact their meeting. But angry? No. This man has not seen me angry.

    That day he didn’t even see me angry. But I was. And I was disappointed that everyone else just sat there and let a coworker announce to the room that he kept a list of times he felt I was angry.

    I am 100% certain that he would not describe these behaviours as angry if they had been exhibited by a male coworker. When women assert themselves in the professional space we are bossy, we are talking too much, we are shrill, we are angry. We are never assertive or firm or powerful. But we aren’t doing anything different. Too many people think that anything other than sitting quietly, smiling prettily, and being agreeable are the only acceptable behaviours for women in the workplace and are affronted when you do anything else.

  • Women just don’t ask for raises and promotions

    In one of my jobs, I found myself pretty unhappy with my situation. Some of the other stories I’ve already shared had happened, I was feeling underappreciated, and passed over for things in favour for some of my male colleagues.

    Things came to a head when I was told that I was expected to take on yet another product and team, and that the rest of the team would travel to onboard with the new product, but that there simply wasn’t budget to send me too. I would have to arrange my own calls with the product owner that I was taking over for, and figure it out myself.

    So I took the weekend to gather my thoughts. I sat down and made a list of the things that I was unhappy with, instances where I felt like the wrong decision had been made, where I had been left out, etc. And I scheduled a meeting with my manager to discuss.

    When people are this unhappy, often their managers only hear about it when they submit their resignation, but I wanted to give my manager a chance to fix things, so I sat down and told him what was going wrong, and what needed to change. The meeting elicited a number of promises that things would change, and that certain opportunities would be offered to me.

    At this point, my manager could not have been more clear that I was unhappy, that I had a list of grievances, and that many of them were within his power to change. I had done the hard thing, I had laid out the problems.

    It’s important that you know that this discussion happened shortly before our annual performance review cycle, where people would be offered raises and promotions based on performance. It’s also important that you know that while in this role, I never received less than an outstanding performance review. That’s not a brag, that’s literally the title of the rating I received, year after year. At the time of this conversation with my manager, despite my performance reviews and having grown my area of responsibility from one team to three, I was still on the same grade level.

    It is a commonly repeated (disputed) fact that women simply don’t ask for raises and promotions, and that’s why we don’t get them. Well, performance review time came around, and I got a standard increase, and no promotion. So I asked. I asked my manager what about promotion, and pointed out that I was still on the same level and had been for 3 years.

    His response: “Oh, I didn’t realise”.

    After I had specifically sat down with him and outlined my feelings about my role. After I had made it clear that I wasn’t happy with the lack of support for my role and the way I was not being appreciated. He didn’t realise that I hadn’t been promoted at all in 3 years.

    He didn’t realise because when he sat down to decide which employees to promote that year, I hadn’t even been on his radar.

    He followed up with some indications that maybe next year he could look at it, for the next cycle, and that it would maybe possible to promote me two minor steps up the ladder (which would have been necessary for a more major promotion). I left the meeting dejected. I had asked. But been told “not possible til next year”.

    A short time later, I found out that a male colleague also had not been promoted in that cycle, and had also expressed his displeasure at this.

    So my manager went to HR to talk about whether or not budget and scope could be found for promotions…. for my male colleague.

    My male colleague was promoted. And my manager received my resignation letter.

    Women do ask. We just don’t get.

  • I’m still speaking

    A theme of International Women’s Day 2021 is “choose to challenge”. They want to encourage people to challenge biases where they see them, to call it out. In my career, I haven’t experienced very many instances where someone called out sexism or bias. Usually I was doing the calling out, and others sat silently and let it happen.

    I worked with a guy who was infamous for his long, rambling questions, and his explanations which were complex, went on too long, and somehow left you more confused than when you started. He was also infamous, at least among the few female employees, for constantly interrupting and speaking over people.

    Once I was in a meeting, while acting as the product owner, and someone asked me a question. The question specifically related to product, there was no reason for anyone else to answer.

    As usual, I got about two words into my sentence, and Mr. Interrupter decided it was his time to shine. This time, however, I had had enough. I interrupted him back, stopping my explanation to tell him that I was speaking, and that it was a product question directed to me, and that if he wanted to add something he should at least have the decency to let me finish speaking first.

    No one in the meeting said a word.

    I finished my answer to my other colleague. Part way through, Mr. Interrupter just got up and left the room without a word.

    I wasn’t rude, I wasn’t angry, I just stood up for myself. I chose to challenge his constant interruptions of me, and I did so without any backup from any of the other men in the room. And what still stings to this day is that afterwards, I felt not empowered, but sorry.

    Because no one backed me up, I worried for hours that I had done something wrong, and every bit of social conditioning in me told me I should apologise to this interrupting guy. I challenged, and in the end I felt bad and alone.

    I didn’t apologise. I wasn’t wrong. He shouldn’t have kept interrupting me.

  • What about the boys though?

    Something I was quite proud of at one time in my career is an initiative I was involved in to encourage young girls to stay in science and technology. It was based on a program that had been run in a US branch, but I spearheaded it in Ireland. I developed a whole curriculum for it, including different printable resources for teams to use, learning goals for the day, different demonstrations that could be run, etc. I was proud of the “behind the scenes” parts too – I ran training courses so that other locations could run the events, and when I scheduled volunteers to run the events, I made sure that it was predominantly women who taught the girls, and shared their work experiences. I also made sure that there were a variety of experiences shared, not just coders, to try and emphasise that technology was broad, and a lot of jobs fall under the tech umbrella.

    It was hard work and it was brilliant, and we got amazing feedback from the girls every year.

    One day, someone asked what about the boys? Couldn’t we run it for them? I pointed out that the point of the day wasn’t just “fun day out of school” it was “girls tech is also for you”.

    So when they went ahead and decided it wasn’t fair to run it only for girls, they tried to do so without me. They scheduled a meeting, pulled in volunteers, and starting setting up a version of my day for boys. Using my curriculum. Expecting to use my personal equipment that I volunteered for the days. But without including me in the meeting or the event.

    Because it’s only fair, right?

    But hey, at least the boys got to see that technology is for them too…

  • That’s just how it worked out

    There are often reshuffling of teams when you work in tech. Projects come and go, budget changes, and you may find yourself with a new scrum master, team mate, or manager without very much say in the matter.

    I had moved departments internally but ran into an old scrum master at a coffee machine. He told me about a team reshuffle that had just happened (which I was already aware of) and how it “just happened” that all the female engineers were on one team together, and all the male engineers made up the remaining teams.

    He told me about how he was pretty interested to see how it worked out, if the all-female team did better than the guys, etc. I think he told me this because he thought it sounded progressive. It did not.

    I already knew about this team reshuffle because, separately, two of the women had sent me messages, wondering if they were alone in thinking it felt wrong or weird or sexist. How does it “just happen” when teams are reshuffled by all the managers talking?

  • You can’t see what you can’t be

    I graduated in 2007, and went straight into work. Since then I have worked as a coder at several companies, been promoted a few times, worked at different levels, and even transitioned into product management.

    My current manager is the first female manager I have ever had.

    In previous roles, there were no female managers in the engineering groups.

  • A bit more experience

    At one stage, a company I worked for had a major transition of version control versions, and internal tooling. A number of teams had to be trained in Git (a version control tool), and also an internal UI development framework.

    At the time, I was something of a Git guru. I had been working in tech for several years and already had a few development jobs under my belt. When people on the floor had problems with their commits, it was my desk they came to, and I could almost always help them resolve their issue. I was pretty fluid in Git command line commands and I could usually get the commit back into shape without people having to abandon the commit and start again.

    I volunteered my time to help with the training effort. I’d never travelled for work before and I thought it might be fun and interesting, good for the CV.

    My scrum master pulled me aside to let me know that I probably wouldn’t be going, they were looking for someone with a bit more experience. They sent a guy who had just been hired as full time after completing his internship with us.

  • Why I am sharing these anecdotes

    I wanted to start sharing some of my experiences of sexism in technology for IWD because I think there is a fundamental misunderstanding about what we need to do to try and address the gender balance issue in technology.

    In these anecdotes, people didn’t set out to be mean or sexist. And sure, they didn’t necessarily say anything sexist to me. But that doesn’t mean that some internalised or unconscious bias didn’t creep into their actions. And the end result is the same – I was left out in the cold, and questioning why.

    If we ever want to address the imbalance in tech, it has to be about more than just not actively shouting sexist remarks at your female colleagues. You need to look at all of your behaviours and really, genuinely examine them for unconscious bias. Did you leave that person out of the meeting because they are not as skilled as you? Or because they don’t look like you?

    Did you perceive her complaining as somehow worse than his?

    And what are you doing to fix it?

  • I thought you didn’t like emergency projects

    I worked in a department that frequently got pulled into short lived projects, and we did a lot of “fire-fighting” (i.e. getting pulled onto a project that was in trouble, working on it for two weeks to turn it around, then leaving). We had been through a particularly tough run of this, working nights and weekends, and all of us had vocalised our displeasure at this to managers and in team meetings.

    Then another project came in. And the first I heard of it was when I arrived into work in the morning to find everyone in the big meeting room, in a kickoff meeting about it. Well, almost everyone. They had pulled in every engineer on the floor except for me, one other female engineer, and the interns.

    When I asked our lead about this, I was told it was because we had complained in the past about the firefighting work, and he assumed we wouldn’t have wanted to be included.