Eye Science Part 1 – How we read and what a human eye can take in.

Since our mention in Wired Magazine, we’ve had a lot of great feedback from users requesting changes to our reading interface. We had the whole gamut of  requests from different colors to different fonts, changing of font sizes, adding words to each cluster, subtracting words from each cluster.

These are all great suggestions and they seem pretty commonplace. One has to ask, why haven’t the guys at Spreed already implemented many of the requests?  To technically implement them is not that hard. What gives?

The answer is not simple.

The easy answer is that we are the first company entirely focused on reading enhancement. It is a new space with new challenges. Our goal is to become the experts in this field. We have developed some significant expertise and we try to use this when building in features. This means that we have a look at the existing sciences to justify a feature’s benefit in terms of reading efficiency.

We look at past research, we try to find new research, we try it out on ourselves and a portion of our community before a change to our reading interface ever makes it to the live site.

Our initial goal when we started Spreed was to let the computer do the heavy lifting of speed reading. We wanted to develop an algorithm and a reading interface that would be effective for most people. No doubt Spreed demands that people challenge themselves to learn how to get through information faster. We remain adamant that with a little (or in some cases a lot) of practice, we can help you read faster. When Spreed eventually catches on and is integrated with other content and technology providers, you’ll be able to let the computer do the ‘heavy lifting’ for all your digital reading.

You might even find an increase in your speed when reading conventionally. Recently, we had a few people say that they are now reading faster on paper since they started using Spreed.

That’s the easy answer to the latency in adding features. I am going to pick a specific part of our research to have our users think about: Have you ever wondered how the human eye picks up words?

When reading traditionally your eyes do not move in a linear fashion across the page. The eye makes many “stops” and occasionally doubles back to words previously read. Even fast readers double back –  only they are a lot faster at it than the average reader. A “stop” is called a saccade and it typically lasts in the range of 200-250 milliseconds.

The Science of Word Recognition – Kevin Larson, July 2004.

http://www.microsoft.com/typography/ctfonts/WordRecognition.aspx

On each stop, the eye will focus on a word, look for first few letters of the next word and also go further ahead to gauge the length of upcoming words and the sentence as a whole. The eye’s ability to look forward might be the reason why single word flashing also know as RSVP might be less efficient than our algorithm.

So what does your eye process in a saccade?

There is a vast array of academic research in this area. I am going to quote Kevin Larson who is the leader in this field. Mr. Larson is a cognitive psychologist working at Microsoft with their advanced reading technologies team. Who knew Microsoft has such a team??? Regardless, we find his work very useful.

During a single fixation, there is a limit to the amount of information that can be recognized. The fovea, which is the clear center point of our vision, can only see three to four letters to the left and right of fixation at normal reading distances. Visual acuity decreases quickly in the parafovea, which extends out as far as 15 to 20 letters to the left and right of the fixation point

The Science of Word Recognition – Kevin Larson, July 2004.

http://www.microsoft.com/typography/ctfonts/WordRecognition.aspx

Visual acuity? Fovea? Parafovea?  What happens when font sizes are increased? when colours or contrasts are changed? when the number of words or characters in a cluster change?  How will this affect how you can absorb information? You need not worry about these questions, but we do!

Parafoveal rules are the basis for the length of clusters in Spreed. The algorithm ensures that characters do not fall outside the average person’s parafoveal field of vision. Depending on the length of words in a cluster, the formation algorithm can produce a cluster with one, two, three and sometimes four words. At times the algorithm does not allow clusters to reach the maximum length because of another set of rules (i.e. grammar rules). The alogorithm tries to encompass speed reading principles, visual perception (eye science), and English grammar / linguistic rules. We’ll leave the other rules for another post.

We err on the side of caution when we make changes to the cluster formation algorithm and reading interface. Our goal was to allow the average (if not daring) person to read faster. There will always be outliers who require a larger font, or different colour scheme etc. Is the algorithm perfect? Certainly not. We feel we are at the inception of a reading revolution and will continue to innovate and test both within our labs and our community at large.

ps. We will be at the Web 2.0 Summit in San Francisco this week November 5-8. If you feel like talking about Spreed, shoot myself or Dave an email (suhail@spreedinc.com or dave@spreedinc.com)

One More Step in the Right Direction

At Spreed we love the Kindle. Any technological innovation that makes electronic reading easier and more accessible is alright with us. But the product has two clear limitations that bother us. First, and foremost, is its size. We get it – Amazon is almost metaphorically replicating the traditional book here. But in this case we think holding on to a form factor that is quickly becoming anachronistic is a mistake. People no longer want “portable”, they want “pocket-size” when it comes to their mobile devices. Why should their mobile reading device be any different?

The second limitation is Amazon’s decision to use e-ink. Again, they’re trying to replicate the traditional reading experience. Less light being emitted from the page means fewer saccades (eye movements) which means a slightly more pleasant experience. But at what cost? E-ink is necessarily a picture of the page. It is not HTML or any dynamic code, and that renders the device little more than a picture window. Obviously, Spreed is all about leveraging the power of the computer to assist and improve the reading experience, so our bias here is pretty transparent. But by opting for e-ink, rather than a traditional browser, the Kindle forces itself into a corner and prevents the user from using the device in so many other ways.

Thankfully, there are others out there who are moving in the right direction. Case in point: Polymer Vision’s new Readius. Check out this article from the New York Times that describes the Readius in more detail.

The Readius is trying to offer pretty much everything that the Kindle does. Only it meets the two criteria above – it’s pocket-sized (therefore truly mobile) and is not limited by e-ink. Take a look at the picture and it’s not hard to imagine the device as a phone. Isn’t that exactly what we really want?

Congratulations to Polymer Vision for taking us one more step in the right direction.