Continuing our conversation with Sina Bahram. He discusses what’s coming in the next 3 to 15 years in terms of automation, the fact that we don’t really have an accessibility problem, we have an awareness problem, and new avant-garde tooks must be accessible.
Nicolas: Welcome to the A11y Rules Podcast. You’re listening to episode 8. This episode has been sponsored by patrons like you. I really do appreciate your support.
I’m Nic Steenhout and I talk with people involved one way or another with web accessibility. This week we’re continuing our conversation with Sina Bahram. And I invite you to listen to the first part of the conversation if you haven’t already done so.
Hi again Sina. Shall we continue our conversation where we left off?
Nicolas: Wonderful. We finished last week in talking about your greatest achievement in terms of web accessibility. I’m going to kick off with something to make us think a little bit. What controversial statement about web accessibility would you share with us?
Sina: Whenever you specifically ask someone to be controversial, I feel it’s always less controversial than the accidental things people say. Let me think about this for a second. I think that for me the field… We talked a little bit about that during our previous conversation. It needs to not exist. But that’s going to take a while to happen. But what I do think is going to start happening is, I think a lot of people in the field these days underestimate the real power of deep learning and machine learning that’s being brought to bear to solve problems in accessibility. A lot of them have been very used to be, and they’re right about this opinion, they are absolutely correct, at least to date, that they and folks like them, accessibility professionals, are the only ones that can solve certain problems. That’s starting to go away. Not only a little bit at a time, but at an accelerating pace.
I would say that the one statement I might make is: “Accessibility is in no way automation proof”. I think that not too many people share that opinion.
Nicolas: I’m seeing more and more automation actually getting ready for prime time. How long do you think it’ll be before we can automate “all the things”?
Sina: All the things? Automate all the things. I think if we’re going to talk about automating all the things we might want to have a 30 to 40 year timespan. But if we want to talk about maybe solving… Let’s maybe look backwards as a metric of looking forward for a second. Over the past 3 to 4 year we’ve seen automatic image description go from “that’s really cute and adorable – you generated a single word and 50% of the time it might relate to the image”, to entire sentences are being generated that are anywhere from 80% or 90% accurate. “Man standing outdoors with sky in the background”. And you’re seeing projects like AI out of Microsoft, or Facebook automated description stuff. You name it that are looking at applying and blending together things like natural language generation, computer vision. It’s going to be very soon before you start seeing that applied to video. If you can apply it to an image, then logically you can do a little bit more to apply it to whole video. Now, I think that in the next 3 to 5 years you’re going to begin seeing projects with automatically generated audio description. Sure, it’s not going to be as good as the human who has sat down and written the script and read the original script to do it, just like YouTube auto captions are not as good as an actual transcriber writing them right now. But again, that’s an engineering problem all of a sudden. It’s almost no longer a computer science problem. There’s a lot of complex computer science that goes into it but that’s just iterating and making those solutions better and better. I think that’s another avenue that’s going to be subject to automation. You’re going to start seeing self-driving cars and so forth get better and better. Slowly as you think of the different challenges that you can drop into a spreadsheet and align against various disabilities, whether it’s cognitive, mobility impaired, visual, hearing… I think you’re going to start seeing automation really putting a lot more agency back in the hands of individuals with that disability. And if it even doesn’t solve all their problems, or even their one problem 100% of the way they want, what it does do is give them a set of tools with which to then design better solutions. I’m really excited by that. Short term time frame, 3 to 5 years I think more automation in terms of description, natural language generation, that kind of thing. 3 to 13 years, by 2030 for example, you’re going to really start seeing the ability to use machine learning and things like screen reading technologies and other interface-based things, where you can come to an interface that maybe has not been coded with accessibility in mind and start inferring all the semantics: “oh, those are buttons on a toolbar”, “oh that’s a file list”, whatever the case may be, “oh touch is available and here are the gestures that I can programatically query”. It’s almost like thinking of our devices as assistants, much like a Siri or a Google Now or an Alexa as opposed to a dumb program, if that makes sense?
Nicolas: Yeah that does make sense. Let’s schedule an appointment right now for 5 years from today and we can revisit. See if your predictions come true.
Sina: Hey, I’m down for that!
Nicolas: What would you say is the number 1 reason people fail to succeed with accessibility?
Sina: Just to recast your question, I do think you mean “why they don’t end up making things that are accessible?”, right?
Sina: A lot of it comes down to a perception that… Well, there are 2 things. One is lack of knowledge, just straight up. We have a horrible pipeline problem in the world right now, that if you are graduating with a technical degree, whether it’s just a programming certificate, computer science, it doesn’t matter. You most likely have not, 9 times out of 10, 99 times out of 100, have not been exposed to accessibility or inclusive design or any of the concepts with talked about on our previous episode or this episode. And that’s really important. Because if you don’t even have exposure to it, we have a fundamental problems. And we do, because most things are not created accessibly. So that’s the first one.
The second one is even when you do find out about it, the perception is that it’s ok to let that be the thing that is not done before the deadline. You’re going to get the UI to look exactly correct because the boss is going to ask you about it. You’re going to get that functionality that the one stakeholder is really vying for worked out because they are writing the check, etc. But you’re not going to make sure that your form has labels. You remember that accessibility talk you went to 6 months ago and you really should have done that but the deadline is TOMORROW, and it’s just going to fall by the wayside. It’s that perception of that being ok, and the perception of being ok is allowed to flourish because there’s no pressure on the other side to remind you of that. When the UI is broken and the button doesn’t work when you click on it with your mouse when you’re a sighted user then you’ve a very strong impetus to fix that button. It’s very obviously broken. When things are not obviously broken in the case of for example, an unlabelled form, then it becomes very easy to ignore it because it’s not in your face, it’s not preventing you from getting other work done, and other people are not asking you about it. So it’s very much out of sight out of mind.
Nicolas: How can we as accessibility professionals help these people succeed and not fail to implement?
Sina: A lot of it has to do with process. A lot of folks will go in, they’ll do a WCAG, the Web Content Accessibility Guidelines, they’ll do a WCAG audit and things of this nature and they’ll provide suggestions for things to fix. “You should put alt text on this, and label this and put these in an unordered list, etc. Fix your headings.” But learning doesn’t necessarily occur there. Some learning occurs and some exposure occurs. “Oh my god I had no idea that that was what it was for”. You get that all the time. But you don’t necessarily then affect those individuals, that team or that organization to do those things on the next project. One of the things that, for example, we do, is we try to weave accessibility into the fabric of the organization. We don’t just go in and solve an accessibility problem for them. We do that of course. And we like doing that, it’s fun. It’s really really cool work. But we then say “And by the way, what have we learned here that can bubble up in the policy?”, for example an inclusive design strategy, or accessibility policy for the whole organization. And what have we learned here that can influence training? So when you onboard any new developers and new program managers and folks like that. And also what have we learned here that could influence QA or QC, Quality Assurance or Quality Control? So that when you go through testing, the testers are not only testing, for example, the password field hides your text as you’re writing in the password, but making sure that it has a programmatically available label of password. Running it through accessibility checkers, or even using it with assistive technologies occasionally. Those types of things I think are the types of strategies that can be used to change things systematically as opposed to a series of one-off solutions.
Nicolas: I approach it that way with clients as well, where I get them to think about processes rather than just the nitty-gritty of “this item passes, this item fails”.
Nicolas: Getting away from a checklist approach to accessibility I think. What would be the greatest challenges for the field of accessibility moving forward?
Sina: We need to be woven in to other fields. Accessibility needs to be something that everyone does. Whether it is programers, program managers, things of this nature, going back to that idea of becoming irrelevant as a field in a way. It’s not really becoming irrelevant. It’s almost not needing to have it be a separate concern. But having it be part of all discussions and moving it forward incrementally and iteratively. We need to do a better job of disseminating the knowledge and expertise and core competencies that we bring to, whether it’s usability or programmatic ways of facilitating access or semantic programming, these kinds of things. We need to bring that expertise to the technical field at large. And there are ways of doing that. You can do conference talks and training and workshops and things of this nature. And also that pipeline problem that I talked about earlier. Making sure that those next generations of computer science students all have some exposure to this. But I think it’s almost a challenge of dissolving ourselves into all the other fields. As opposed to being a separate one. I do think that specialization matters, don’t get me wrong. There’s going to be a need, I won’t say always because I don’t believe that, but for some time to have specialized accessibility folks, for sure. But we’ve really got to take this problem seriously of “we can’t have most of our conversations involve ‘oh my goodness I never thought about that before'”. And if that’s the response, we have an awareness problem, we don’t have an accessibility problem. That’s easy, we can tell anyone how to make buttons accessible. It’s not hard. What’s hard is the fact that they didn’t do this in the first place. Or at least that they didn’t know about doing this. Wilful ignorance is one thing. But this is legit, they really didn’t know about that. They never thought about that before. And that is not only an accessibility problem. That touches on things like media, social media, how folks with disabilities are portrayed in terms of how they use technology, etc. There’s a larger problem there, but one that’s infinitely more worth solving in my opinion because it really moves society forward as opposed to these one-off solutions.
Nicolas: Yes. I’ve long been thinking that I would like to see more curriculum about accessibility in computer science, multimedia, graphic design, all these kind of programs. When I was in Savannah I was invited to do a 2 hour session at the Savannah College of Arts and Design (SCAD) on that topic to web accessibility students. And that was great. But when you look at it, 2 hours over the course of 2 or 3 years, it’s not enough.
Sina: That’s exactly right. And by the way there are some projects that aim to fix this. Teach Access is one that I would definitely recommend that you look into. Really trying to solve this pipeline problem. And introduce accessibility principles into higher edu curricula so if you graduate with a computer science degree for example, you will have been exposed to at least, and I don’t know what the current metric is but I think it’s one semester, or maybe even one quarter of accessibility stuff and inclusive design. So there are some things that are gaining traction now to try and solve this problem.
Nicolas: That’s good to hear. It’s similar to, a pet peeve of mine, architects. They have a 4-year program at university and they get maybe 4 hours of tuition on making buildings accessible for people with mobility impairments so there’s an imbalance there. Hearing that there’s something that’s happening in terms of digital accessibility is great news.
Around that, what would you say is your greatest frustration in terms of accessibility?
Sina: My greatest frustration is the lack of accessibility in new things. We spend a lot of time making existing things and older things accessible. Whether it’s a product, or an app or a website that’s already been coded. And again we get to chose a lot of the folks we work with and I’m very fortunate and work on things where they are considering accessibility upfront. But still that is a small fraction. When you look at someone’s who has a technical background and is interested in programming technologies ant that sort of things, I just assume whenever I hear about a new IDE (Integrated Development Environment) or a new wonderful 3D modelling platform, a new machine learning toolkit, that has a UI around it, that it’s not going to be accessible. It’s one of those things. Always predict the worst and you shall be hailed as a prophet. The idea there is that it’s true unfortunately. It’s just a very big problem that it’s not being backed in these new avant-garde tools. And so that has a double whammy of a negative impact.
The first is obviously new things that are advanced and are things you should be hip to in the field – they’re not accessible to you. But the second is that by doing that it prevents you from designing into the eco-system of where the next tools after that come from. Because these things are all iterative. They build upon one another. That cycle needs to be seriously broken. The R&D technologies all the way down to latest version fo Slack, or whatever the case may be, that needs to be accessible from day 1. Not day 1000 after everyone complained on social media. Now at least we’re there, so we’re not on “we don’t care”, or “our company has no plans of working on that” I think a lot of Silicon Valley companies are hiring accessibility professional. They’re really spooling up their game in that sense. But still that new stuff coming down the pipe that’s not accessible, that’s my biggest frustration.
Nicolas: I think this concept of iteration and making sure you have accessibility from the get go is an important one, and one that no one that I spoke with so far have mentioned that. Thanks for that, it’s really thought provoking.
Imagine a world where you have managed to do such a good job that you put yourself out of a job. What profession would you take? What would you do for a living?
Sina: I’m on a yacht, man. It’s like a nice island, life is good. Let me think. What would I be doing? I really enjoy making things that delight people. I think it’s fun. It’s fun to have that “WOW factor”, whether it’s in learning, whether it’s just “oh that was really cool”, whether it’s in getting folks to do things that they previously thought impossible or highly improbable, those are fun experiences for me. I think it’s one reasons why I gravitated so much towards museums. Because well done interactive exhibits sometimes elicit and invoke those feelings in people. I think something along those lines. But also I love scale. It’s been pretty hard resisting going and working for one of these Silicon Valley companies because it’s one of those things where it is cool to be able to think about things and say “and now what if it needs to happen a trillion times a day?”, or “what if it needs to actually work with multi-second delay because it’s going to be traveling in outer space and we need to communicate with it”. Those kinds of challenges to me are really fun. I like any thing that is very high impact and very large scale that has an ability to impact billions of people. That to me is a lot of fun. So something along those lines most likely is where I would see myself.
Nicolas: Thank you. Who inspires you?
Sina: Lot of folks. Growing up I thought I was going to be an astrophysicist. So I would read books by Stephen Hawking for example, and folks like that. That was really inspirational both conceptually and intellectually because it’s like “wow, the universe is complex”. All the way to “this guy wrote that book using a switched based input device”, and that’s just… We’re talking tensor maths and everything. That’s unreal, the level of frustration that that takes. He was an impatient guy. You read about Stephen, he was very smart and impatient as a young man. To be able to be as prolific through that limitation of bandwidth, that was a big inspiration to me.
I think others are educators. I had a chemistry teacher in college. She would say “come to my office after class”, it was like a Monday Thursday Friday thing. I would go to her office 3 hours a week. She would bring in rocks from her garden, She’s like “poli-exclusion principles, let’s go”. Here’s the electrons, here’s the nucleus, and she would just lay them out on the table and made her own stuff because the disability office was unable to provide that and she came up with the solution in 5 seconds. That was brilliant. Absolutely brilliant. She would give me exams that way. People like that that just take very limited resources and are able to turn them into things that are so in excess of the sum of the parts. That’s a big deal to me. I like that stuff. Those are the kinds of things and people that inspire me.
Nicolas: Wonderful, thank you. Sina I think we’re going to wrap up for now. Thanks again for taking the time to talk today. I really enjoyed it. Hopefully you didn’t find it too painful.
Sina: I loved it, it was great. Thank you for having me.
Nicolas: Folks on the other end, thanks for listening. And until next week, that’s all. Remember that if you need a hand in ensuring your website is accessible, I’m available. Contact me on my website on http://incl.ca