Interview with - crowdsourced accessibility

A few weeks ago a lot of tweets, emails and instant messages started chatting about a new system that would allow screen reader users to fix issues with web sites in a crowdsourcing manner:

WebVisum is a Firefox3 extension that seems to be too good to be true: it provides screen reader users with a hotkey to report bad markup and missing information and allows the community or the user himself to add the information and thus remove the barrier for the next time. I’ve caught up with Marc Dohnal, the initiator of WebVisum to ask him some questions about the system.

Hi, so can you tell me what is WebVisum, who are you and what is your engagement with it?

WebVisum is a social Firefox extension that is designed to empower the blind and visually impaired community and give them the tools they need to help themselves to a better browsing experience.
I’m responsible for the ideas and initial funding behind the project as well as the ongoing features and guidance for the development.

What are in your opinion the main problems that people with these impairments face these days and how does WebVisum help?

As someone who has had no experience with accessibility issues before I started the project, I guess I came to it with a fresh point of view. The more I read about the difficulties people are facing the more I realized that only time and lots of effort are going to improve the situation on the web today.

What I mean is that most people on the web are like me, they may not even imagine that blind people are surfing among us. And then obviously, they don’t know what tools the blind and visually impaired are using. Not to mention, that most people do not have the knowledge to make web sites and pages that comply to any and all standards - that’s something that even professionals find difficult to achieve.

So the main problems are navigational menus that are built from images only, buttons which are made out of an image and include no alt tag, and missing alt tags or important markup. There are of course more problems such as dynamic web sites (JavaScript, AJAX, Flash, etc.), but these are more advanced topics which we plan to address in the future.

And then there are CAPTCHA images which blind people are simply unable to overcome without sighted help. In the rare cases that an audio CAPTCHA is even provided, many times that CAPTCHA would be very difficult to comprehend.

Having worked as an accessibility consultant I found that people are very happy to find a checkbox to tick and then claim they are accessible. Educating people goes just so far in that. You take a different approach then?

We take a slightly different approach to this. My personal view is that it is very difficult to change the Internet (and the world) for the better. So I think we should provide the tools that allow the users in need to make the change right now, for themselves.

Once more sites and pages become standards compliant, or improve their accessibility, our tools become less relevant to these sites. But the Internet is huge and therefore it would take many years, if ever, for most of it to become accessible to a good degree.

So instead of asking and begging site operators to change their ways, to fix their pages (often trivial tasks), we build tools that allow the community to literally help itself.

How does that work in terms of workflow - especially when it comes to CAPTCHAs?

Once you’ve got the extension installed, you surf the web the way you usually do. When you encounter a page that has a CAPTCHA that must be solved in order to perform an operation such as site registration, forum or blog post, you simply hit the right hotkey, our code then finds the right image on the page and submits the image to our server. Our server will then process the image and generally in under 15 seconds the response would be provided to the user. The screen reader will read the response and the user is told that the result is in their past buffer - all they have to do is paste it into the correct form field.

We worked quite a bit to make this process as simple and as smooth as it possibly can be.

Many people think that our CAPTCHA solving is a wonderful feature, probably our number one feature. But I disagree.

Well it is the flashiest feature for sure (also slightly controversial, but let’s tackle that later). What is your favourite then?

Our most important feature is the one that allows “fixing”, “tagging” or “annotating” broken links, form fields, page titles and other page elements.
For example you go to a new site, or one that you frequent and all your screen reader reads for you is “link, link, link, image, link” and then some content that’s on the page. If it’s a site you frequent, you remember that link number 2 is the login page, for example. and link number 4 is the “contact us” page.

With our tool, you give each link its true meaning, the next time you visit that page, the screen reader will read the text that you’ve entered for every link. In addition, any other user of our service and extension would automatically get this data as well upon their visit to the same site. This means that you didn’t just help yourself by “fixing” a site you frequent, but you also helped other members of the community that may be using the same site. Everyone benefits, you help both yourself and your friends. They help themselves and you. This is what I meant when I said that our product empowers the community.

That’s a wonderful feature, but doesn’t it just mean we fix a symptom and not the cause? Are there any plans to give “fixed” sites access or maybe collaborative messages about their issues? Something like “our users fixed n problems on your site last week, can you look into that?”

There are plans to showcase these changes to site owners and have our community possibly create a presentation and maybe even a list of signatures of site users who want the change to happen. We may possibly also provide a list of sites which are popular with our users yet are very inaccessible and hopefully this way push site operators to improve their sites faster. So yes, we’re curing symptoms but quite possibly with the right tools and the community to back it up, we might be able to introduce change as well.

As we’re currenty a very small and self funded team, we’re first concentrating on what we consider as the critical issues. We’ll then branch into advocacy and related fields.

That is a great idea, as most of the time site maintainers (especially on large sites) are blissfully unaware of the issues. Another good source of information to provide to people on the web are problems that have to be constantly fixed.

Right, many site operators, just like average Internet users, are simply unaware that only little changes would make the lives of the blind and visually impaired users much easier. We can help and educate them.

This is all very obvious, but there is quite a massive amount of paranoia going on, too. Providing a means to circumvent CAPTCHA could be seen as an attempt to hack sites. One of the first proofs that CAPTCHA are a glass shield was someone hijacking a CAPTCHA and showing it on a third party site luring real users to enter the CAPTCHA (enter for free porn/downloads). This might come your way sooner or later. How can you de-fuse this scare/argument?

Yeah, there will probably be some paranoia and fear. First and foremost CAPTCHA images are not a perfect protection and that is already well known. CAPTCHAs on specific and high profile sites are routinely broken.

While our solution does solve just about any CAPTCHA that we throw at it, we restrict the use of this feature to our extension users only. Each user can solve a few CAPTCHAs per day - it is not unlimited. We employ several protections and safeguards from preventing the massive use of our API for this purpose.

We have a few tricks up our sleeves that would deter even the most dedicated hijackers from using our service for these purposes. Among other things, we plan to build a “trusted” user base, meaning, not everyone would just be able to come along, sign up and start solving CAPTCHAs. It is quite open right now, but we have plans to make this a members only club, so to speak - just not as snobbish :-)

Does members mean that people will have to pay for the service?

By members only I do not mean that we ask people to pay for these features. In fact, because we believe that these features are fundamental to the community, we do not plan to ever charge for them.

Throttling the CAPTCHA solving API and creating a trusted group is a very important step, as it would be a shame if the system would be forced to shut down because of hacking accusations. Monetization is a good question. Right now you pre-fund the service. Are you planning maybe to partner with trusted companies to possibly host the data with them?

We’ve received very positive feedback about what we’re doing right now. We’re very dedicated to maintaining this project in a trouble free manner for the entire Internet community and we will work with any site operators should any problems arise.

We’re open to partnership offers; however, we just launched two weeks ago, it’s a bit early to talk about that.

As for funding, we plan to be fund further development of features and services on top of the free core that we’ve got now. Our plan is to charge a very low monthly fee from users who wish to either support us or have access to advanced features such as PDF OCRing, and similar things.

To make it perfectly clear, we will not charge for our basic features, not now and not in the future.

There is not much information about the people behind WebVisum on the site, is that on purpose? It can spook out people a bit.

It’s not on purpose. We’ve been so busy with the extension, our server code and our site that we just didn’t have time to open up just yet. We plan to have a very open development process, too.

To our surprise, many people were saying that this is too good to be true. We’re extremely happy that they think so. We’ve been getting lots of positive feedback so far. So back to your question, we will open up very soon.

How many users and downloads have you got so far?

Sadly these numbers are way below my initial expectations. I always aim very high, though.We’ve had just around a thousand registrations and a similar amount of downloads so far. More people are signing up all the time and we see the usage increasing daily, so that’s positive.

All in all, our project is extremely young. We have had almost no beta testing before our launch and we’re delighted that any problems found so far are relatively minor and that people are happy with what they are seeing.

We’re not well known just yet, but I have no doubt that we will be.

I’ve had some feedback on the “too good to be true” part and people are a bit wary to use the system as there is a security scare about the system showing up from out of nowhere and not much being known about the people behind it. Probably some backing up by Mozilla and coverage on the Mozilla site would make that less of an issue. Is the extension available from there?

I’m actually in a way quite happy we didn’t see a huge explosion in use. In the two weeks since the launch we’ve found a good number of bugs which we’re addressing with a new release, hopefully in the next few days. After that, I would be very comfortable with more publicity.

I would say to those concerned that we would not spend thousands of dollars and countless hours of research and development just to scam, defraud of spam a relatively small Internet community such as the blind and visually impaired. With spending a similar amount of money, time and resources we could have a bunch bigger return on investment by targetting the entire Internet.

What about the software behind it? This would be a pretty cool tool for internal QA of companies, too. Visually impaired testers could report bugs without having to write up issues and explaining but the system could tell exactly which element has a problem. Are you planning to open source it or offer companies installers for money? Test servers cannot (or well should not) be accessible to the outside.

We haven’t given any thought to that kind of use just yet. We’ll address any opportunity and idea as they come along. Thanks for the idea!

I am still a bit confused about the real fixing of issues. I understand that screen reader users can report an issue with a hot key. Who provides that information that is missing then and how is that moderated? Is there a wiki of latest changes where other community members can verify fixes?

Users provide the information themselves. For example, a user can figure out a good and descriptive page title or link name, just by reading through the content of that page.

Any user can then come and modify the tags further, refine them, adjust them, etc.

We plan to export all of the tagging history for all sites through our web site, to our users. They would be able to see who edited which field, when, what they wrote, etc. Users would then be able to vote which description fits the purpose best, etc. The same would go for moderation.

So if I follow a link that is empty (because of a lacking alt attribute on the image nav for example) I can retroactively say that this link really links to that page with this information without having to go back? Is there a link cache or do I have to follow the link, read, and go back?

Exactly! In fact we have a special key sequence that says “Label last visited link”. Now that you’ve established the purpose of the page you’re at, you can simply hit that key and label the link which brought you here.

That could be mixed with social bookmarking sites. I found for example that people find better tags and descriptions for my sites in that I could have ever come up with.

Agreed. We’re open to mix and match various sources for the data. Again, our product and offer is very young and we have a long task list ahead of us. We hope to create a very open environment, wikipedia style.

The other good idea is to use Yahoo’s site explorer or similar systems to learn about sites linking to the current site and automatically tag those. That way you could enhance links automatically without having to visit them

We do have various ideas and ambitions about automatic tagging and fixing of links, etc. But first we must be sure that our core is stable, easy to use and bug free.

Which brings me to the last question for now. What do you need now to take this further? How can the community or companies help you?

First and foremost we will see how far we can go with the amount that we have invested so far. There are possibilities of grants and similar help. Possibly revenue that we may be able to generate from added value services would suffice, at least in the short term.

We’re fresh out of the oven, a lot will happen in the next few months that will shape our look into the future.

Well, thanks for the interview. I think it is nice that you came into the whole accessibility arena from a different angle and found something to have fun with. Any last words or greetings you want to give out?

I’d like to thank Marco Zehe from Mozilla for being the first to do an in-depth review of what we’re doing and Aaron Leventhal of IBM for his support. Both of these guys have been very supportive, provided lots of positive feedback and everyone appears to love what we did so far. We’re having fun working on this and we couldn’t be happier with the results so far.

If you’re a user, or a potential user and need help in getting this going, please don’t hesitate to contact us, we’re happy to help! We’ll be setting up a mailing list, possibly a forum and definitely a Wiki so we can better document the tool and how to use its full potential.