Accessibility improvements!

As you may—or may not—have noticed, we have made some changes to the website. Let’s ignore the elephant in the room for now, and if you haven’t noticed an obvious change, perhaps it will come to you after reading a bit of this post on accessibility.

So instead of discussing the more visible change first, let’s have a look at a recent change you probably haven’t observed, but which you may have experienced if you can’t see. For those who can see but wouldn’t observe this, it’s all going to make sense in the next section.

Vision-impaired

Something I became aware of recently is the importance of meta-descriptions attributed to images in posts. In HTML, this is called alt text, short for alternative text. These alt texts are generally invisible. If an image has alt text attributed to it, if the image can’t be rendered, it will display the alt text instead. But I would argue the alt text has an even more important function, which is that these descriptions are read aloud to blind users on a screen reader, such as Orca.^[1]Image ALT Text | PennState Accessibility

So for those of our readers who are blind, the images in our posts would have had no consequence to you; you wouldn’t even have known they were there, thus missing the information presented in those images. It may be a diagram, for which you were only able to read the caption but didn’t know what the diagram actually showed. Or, in case of decorative images, there would have been no caption, and you might have missed out on a joke, or images that add some warmth to my admittedly sometimes cold and clinical writing style (I am trying to improve on this, to make my writing more accessible as well).

If you are not blind, but are otherwise vision-impaired, you may also have wanted to rely on those alt texts had they been there. So if you are vision-impaired and you are reading this, we apologize for not having met your accessibility requirements until now!

If this is you, then you might have noticed that in the last posts those alt texts are present, and we are working on updating older posts as well. And of course, we will make sure to include alt texts in all future posts. Also, let us know if there is anything else we can do to make our website more accessible for you.

Failing the Turing test

Speaking of accessibility, we have been getting reports now and then from people who wanted to comment on a blog post, but who were blocked by our security system because their on-screen behavior was deemed to be too much like a bot. The issue here is reCAPTCHA v3, which is an online Turing test of sorts we have been using for almost a year now to protect the website from spam. With reCAPTCHA v2, you had to solve a little logic puzzle or find an image to prove to the system that you are not a robot. We upgraded to v3 because it is an automatic system, so now the user no longer has to spend any time proving that they are indeed human, but instead their behavior on the website is tracked and assessed for humanity. That sounded great to us, because that meant we were no longer causing obstructions for our users, which improves accessibility and usability.

But as it turns out, there are issues with the system. I’ve had 4–5 reports of people who couldn’t comment because they were blocked by the system, and Natalie tells me she had about 20 of such reports in total. That makes me quite concerned about how many people were blocked that didn’t report it. Probably many more than that.

What is going on? Well, I read something about how it tracks online behavior, which I had assumed included how you move your cursor on the screen. So I theorized that when you make too many linear movements at a fast rate, the system flags you as a potential bot. I wondered if perhaps there is something about the online behavior of autistic people that gets them flagged as bots more so than is usual. After all, I would think the reCAPTCHA v3 system has been fine-tuned for neurotypical online behavior, with the risk of rendering non-neurotypical behavior non-human.

So is there an issue with reCAPTCHA v3 in general, or has our website exposed a weakness in the system regarding autistic people? Either way, it’s a problem for this website. I also learned that while reCAPTCHA v3 is very secure, it goes at the expense of privacy.^[2]Google’s new reCAPTCHA has a dark side | Slashdot I think the privacy concerns probably have something to do with why so many autistic people inadvertently got blocked.

Our web developer, Quinn Rusnell (the guy above), stated:

You might be right that some users may be blocked more than others because they seem more bot-like in their behaviour, although if they are telling the truth that v3 doesn’t use any user interaction, it’s certainly not looking for “bot-like behaviour” live on screen. Google is very secretive about how it works though, and there are no strength of security settings like for V2. Reverting to V2 might be the best option in this case.

And:

On a side note, it would be interesting to set up a test to check if Autistics get blocked more. I suspect however that they use more IP and browser data to assign a risk score rather than on-site or on-form behaviour.

Consequently, we have now reverted back to v2. That sounds like regress, but as Quinn notes, we have a lot more control over the security settings. Also, we turned off reCAPTCHA for comments, and installed another plugin to defend us from common spam threats. So as long as we are not receiving spam messages, we will keep it this way.

So at this point there should no longer be any false positives when it comes to our Turing test. It’s nice when an autism website is actually accessible to autistic people, isn’t it?

More accessibility improvements

I was reading the page How to Make Your Blog Accessible to Blind Readers on the American Foundation for the Blind, which lists 7 accessibility concerns blogs ought to address. #4 is to avoid the use of top and left-hand navigation bars because the screen reader starts at the top of the page and reads from left to right, so on every page you load it will read the menu again. Adding a link to skip the top menu will be obtrusive to many readers, but if you access our website via mobile, the menu will be on the right-hand side.

Other than that, it seems we comply with every guideline, except for the last one; guideline #7 is to avoid forcing links to open in a new tab. The protocol we have been following is to make external links open in a new tab, while links to other pages on our website open in the same window. As such, we make sure users don’t inadvertently leave our website as they click on a link. Apparently this practice can be very disorienting for a visually impaired user, because unless they have the latest screen readers, they get no indication that a new window has been opened.

So in order to avoid such frustrations for the visually impaired, we will now give the user the option of opening a link in a new tab, regardless of the type of link. To make sure all links adhere to this guideline without having to adjust all links manually, we installed the WP Accessibility plugin, which also corrects a myriad of accessibility issues and makes further accessibility improvements. If there is an accessibility feature missing which you would like to see on our website, let us know in the comments and we will consider implementing it!

Update: We have received reports from people indicating that the fact that external links don’t automatically open a new window is quite obtrusive. Hence we have to make a decision about who we offer the most comfortable experience to; most of our readers, or the minority that uses screen readers. We apologize we can’t offer the best experience to everyone. However, on second reflection I think the screen readers should define how links are opened, rather than urging websites to abandon the target=_blank attribute and thus undermining usability in the process. So relunctantly, we choose not to follow guideline #7.

Typography

So now let’s address that elephant I mentioned earlier. Did you notice anything about the typography? It’s larger now! At least 11.25% larger, to be exact.

• The body text of the desktop version of our website went from 16px to 17.8px, which is a 11.25% increase.
• The body text of the mobile version of our website went from 14px to 17.5px, which is a 25% increase.

Come to think of it, Monster Insights shows that 33% of our website visitors use the desktop version, while a staggering 63% of our visitors access the website via their mobile phone, so I should really have advertized the 25% increase in the typography, as this is what most users experience. Hence despite its size and the in-your-face typography, you might call it a modest advertisement.

Quinn is also working on other improvements on our website, but we will let you discover those on your own. But do let us know what you think of the changes! What other improvements would you like to see?

PS: If you need a website, we highly recommend hiring Quinn. Have a look at his portfolio at Strategic Web Development.

If you like the design of our website, consider hiring Quinn and myself together; he makes whatever I design work as it should!