Technology makes us information rich and attention poor

You can’t speed read (5-10x faster) without a corresponding loss in comprehension, but it turns out that you can read faster and understand more of what you read if you improve the user experience inside the foveal window. This post describes relevant terms and examines empirical evidence on improving the user experience of reading. A link to a related white paper on improving the user experience of reading is also included.

Recently, the New York Times published an article “Sorry, You Can’t Speed Read” by Jeffrey M. Zacks and Rebecca Treiman. The article covers the challenges of improving our individual abilities to read faster without significant compromises.

Here at Asymmetrica we have over 40 years of combined experience studying visual perception, reading, and comprehending meaning, both in theory and application. Our work on Asym supports the premise that reading dramatically faster comes at a price to the comprehension and understanding of the content we read.

The Experience Shapes the Meaning

Techniques used to speed read work by rapidly scanning or skimming text. They work because they fundamentally reduce cognitive load during the reading process by leaving out significant content. The result is faster reading. But this speed comes at a cost. At best, speed reading loses critical context and shades of meaning in the author’s message. At worst, it changes the meaning of what we read.

RGB Cells
Viewing an individual RGB cell in isolation gives an incomplete picture – context is critical even to “comprehend a pixel”.

Even products such as Spritz and Squirt, which are based on the Rapid Serial Visual Presentation (RSVP) technique, have their own drawbacks. Although they don’t skip content — every word gets presented — they decrease comprehension at higher rates of speed. Words were never designed to stand on their own; they are meant to be interpreted in the context of the phrases and sentences in which they occur. Like presenting individual pixels on a screen one at a time without including the surrounding pixels, RSVP techniques strip away vital context and present fragments that were never designed to stand on their own. The result is fragments of a picture, rather than the complete picture.

Science tells us we can’t read 10x faster without significant trade-offs in comprehension. So what can we do to help our brains read better? Improve the user experience of text.

First let’s talk about what reading really is and where cognitive improvements can be made.

Natural Language Understanding = Hard AI Problem.

Natural language understanding is one of the hardest problems for computers to get right.
The human brain is extremely complex and capable of processing thought, meaning, and critical thinking in ways that even the largest supercomputers and most ambitious AIs created by some of the largest companies have yet to approximate. When we read, our brains perform a deceptively complex set of both serial and parallel functions to turn pixels, letters, words, sentences, and discourse into “meaning.” Complicating things further, human language often contains many non-literal contextual markers like humor, irony, or other indirect modifiers which invert or mean the opposite of the literal meaning. Sometimes these subtleties are the point. Or there may not seem to be a point if you miss the punchline of a joke.

The Act of Understanding


Order and position matter. Often these can be a source of mistakes, confusion, or errors in one of these processes. We make mistakes – miscues – in reading all the time, but we use context to correct course and get back on track. Wolfgang Iser described reading as an act, the act of using text on a page to generate our own virtual text that translates symbols on a page into personal meaning. Reading is an act of constantly generating and comparing our virtual text to the text we are reading. While not everyone agrees completely with Iser’s interpretation, most agree reading (like listening) is an interactive process, where the reader is trying to guess the author’s (or speaker’s) intent. Reading, like other cognitive activities, is part passive perception, part active reconstruction, building understanding as we go.

This is the magic of language – you can get inside the head of someone else and understand his or her thoughts, even when those thoughts are separated by large distances or thousands of years in time. The written word has a special impact, a certain umph, maybe because there’s an implicit contract when someone takes the time to put their thoughts into writing. When writing is done right, the reader wants to understand.

You Can’t Speed Up the Clock Speed of your Brain’s Processor…

Hobbyists sometimes use liquid nitrogen to cool an overclocked computer processor. This is probably not a viable option for overclocking your brain.

Reading depends on biomechanical processes, and the speed of these biomechanical processes in our brains are fixed. Some brains may be more adept and some less adept at this (e.g., due to individual differences, development, dysfunction, or trauma) at certain parts of the reading experience. But in the end we are all speed-limited by the boundaries of chemistry and physics, attention and perception, and memory and cognition.

One of the limits on these cognitive abilities is anatomical. Our eyes have a specialized region in the retina called the fovea that we use for detailed visual analysis, such as during reading or driving. The fovea contains about 1% of the surface area of the retina, but the brain devotes about 50% of visual cortex to analysis of the signals coming from this tiny, crucial set of neurons. This special part of the eye has evolved to break down the world in exquisite detail, but the detail that it can see is only a small window of our visual world, typically a little more than a dozen letters, or about 2-3 words wide.

Fovea window and chunking text for speed reading
The brain devotes about 50% of the visual cortex to analysis of the signals coming from the fovea, which is only 1% of the surface area of the retina.

To read or perform any other complex visual activity, we have to move this window across the environment by moving our eyes and turning our head or body. The outside world gets represented in our heads as a series of tiny hi-res snapshots. As we scan our eyes across a visual scene, our brains integrate these snapshots into a seamless experience. Unfortunately for reading comprehension, the peripheral information outside the foveal window is blurry. We use this blurry area (the 99% of retinal real estate outside of our foveal window) for detecting motion. Thus, a real bottleneck in cognition for activities like reading is how information is packed into this narrow foveal window that we see through.

Once information is through the window, other slower forms of processing integrate the words into structure and meaning. Forcing information into the foveal window faster with technologies like Spritz or Squirt doesn’t help with understanding it better. Before your brain can integrate the information from this foveal window, it needs to be in an efficient package. But what’s an efficient package?

An Ideal Package of Information

An efficient package is a small chunk of words – a phrase – one of many structural patterns that occur over and over again in the language. Your brain is fast at recognizing these patterns, and ultimately uses them as building blocks to resolve ambiguities as you build up the structure and meaning of a sentence during processing. Ideally, the information from a single phrase or chunk would fall inside the foveal window – one chunk per foveal snapshot.

Otherwise, your brain has to determine which words belong in which chunks across two different snapshots (separated by an eye movement). This is harder, less efficient, and more prone to errors. Given the small size of the foveal window, optimization is not about faster information flow into the window, but about good packaging of the information. In an ideal reading environment, the eyes would process one package of meaning (one phrase) per window. Good packaging speeds output to the next stages of language processing, where higher order relationships among phrases can be resolved.

There are many ‘packaging’ techniques used to help readers identify meaningful chunks. The white paper Augmenting reading performance: The history and science of chunking via text formatting examines the variety of techniques used to help readers chunk effectively. Reading the white paper will introduce you to:

  • How language is organized into phrases, clauses, and sentences (chunks) during comprehension;
  • a rich history where writing systems fully marked these chunks, but modern writing systems only mark the ends of major clauses and sentences;
  • a synopsis of decades of research that show that chunking improves reading performance;
  • how good readers have eye-movement patterns that correspond to meaningful chunks; and
  • the ways to cue poor or distracted readers to have the same eye-movement patterns of good readers.

The takeaway is that distracted or otherwise challenged readers can have eye-movement patterns identical to those of excellent, undistracted readers if appropriate techniques are used to chunk text. Given the distraction-rich environments in which we read, we see chunking text as a valuable tool to enhance reading ability. We want to bring the benefits of this technique to as many readers as possible.

How to Fight a Perfect Storm?

Given society’s large shift towards mobile devices like phones and tablets, our reading experiences can now happen anywhere and anytime. Cutting cords has given us new freedoms, but also new burdens. Not only has more data and content been made available for us to read, we are now more and more distracted by mobile surroundings. The rising currents of information combined with a climate of distracting devices creates a veritable hurricane that wreaks havoc with understanding what we read.


This is why at Asymmetrica we have a sense of urgency. Our transition from the information age to the networked age has deluged all of us with an ever increasing wall of information. This deluge is only going to get more intense. To develop a solution, we carefully examined what we can and can’t do. We know you can’t speed up the clock speed of your brain’s processor, but you can remove unnecessary operations by chunking information into a more efficient representation. This ‘wasting less time’ approach is the heart of our solution.

Waste Less Time

Readers of content that use chunking aren’t as much reading faster as wasting less time while reading. With effective chunking, each eye movement is more efficient – our phrase-marking gives your brain natural cues for where to pause during reading. The content remains the same, but because eye movements are more efficient, the brain can efficiently integrate every chunk of meaning into the stream of understanding. With Asym, people aren’t cramming in more information faster, they’re comprehending more effectively.

Better Comprehension Can Lead to a Better World

Speed reading (1000% faster) won’t change the world beyond teaching people how to skim text. Improving comprehension is a different story. Significant improvement in comprehension is not only possible, it is practical. It can help companies keep users on a webpage. It can improve conversion rates on calls to action. In today’s post-mobile, highly networked, distraction-rich world of information overwhelm, better understanding might just lead to a better world. The 10-35% improvements we’ve seen from chunking are the difference of a letter grade (or more) for a student in school. It may be the difference between understanding and not understanding what we read. And it could be the difference between someone learning (or re-learning) to read, and someone giving up in frustration.

In the words of Herbert Simon more than 40 years ago:

In an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes


He concludes by saying:

. . . a wealth of information creates a poverty of attention.

We designed Asym to address the poverty of attention so elegantly described by Herbert. If our solution to modern distractibility helps even a few at-risk readers, it has more than met our mission.

Asym technology is uniquely positioned at a crossroads where design and science meet typography. We hope it’s worth putting on your map.


Introducing Asym Spacing

The spaces between the words we read every day have remained essentially unchanged since the Early Middle Ages. Uniform, unvarying word spacing dominates how type is arranged despite radical technological developments in the past millennium in the way text and documents are created and distributed. But why are even spaces the norm, when decades of science have proven that uneven spaces add significant value to comprehension?

This is why we founded Asymmetrica Labs – to make this nascent discovery an accessible technology – and this is our story.

Almost every modern writing system uses punctuation. This function seems crucial and obvious to us now, but punctuation, including word spacing, took centuries to develop and catch on as practice. Brace yourself for a brief history of the status quo.

Greek Papyrus
Example of continuous writing in Ancient Greek.

The earliest punctuation was developed around the 5th century BCE by the Greeks who used the paragraphos to separate text passages into paragraphs, or text sections. Before this point:


Examples of paragraphos, one of the earliest forms of punctuation, used to mark document sections.

The paragraphos was followed by the dicolon and tricolon, used by ancient Greek playwrights to indicate when an actor should pause for breath when reading the work aloud. Aristophanes of Byzantium in c. 200 BCE used punctus, which were dots at various heights on a line: kōlon, komma, and periodos to indicate pause length. As you may guess, these marks evolved into their modern namesakes.

Adding spaces between words on the other hand is, perhaps surprisingly, a relatively recent development. In the 7th-8th centuries Irish scribes began adding visual cues in order to help make Latin more readable to non-native speakers. These visual cues were the spaces between words that we know today.

But the use of word spacing with modern hierarchical punctuation (reminiscent of Aristophanes’) took until the 19th century to become a systematic practice. This system of separating words with uni-­size blanks is used in many modern writing systems for Latin, Cyrillic, and Arabic scripts.

In 1951, North & Jenkins published a study [1] which showed that participants read faster and showed better comprehension (10.9% better) when the phrases were made more obvious. Multiple studies on phrase-based (also called “syntactically cued”) formatting have been published throughout the latter half of the 20th century which corroborate this important finding.

But why have these studies not been applied to mainstream reading materials?

One answer is that both the publishing industry and typography in general are conservative and slow, if not resistant, to change.

Another key reason is the intersection of technology, distribution, and ease of use, which have not historically been available to embrace and leverage the science for everyday use.

One more answer is that some approaches to phrase-formatting try to change people’s reading habits and how documents are arranged instead of augmenting the natural reading habits of good readers.

Take attempts to reform the English spelling system as a case study in technology reform. There have been many (well intentioned) attempts to simplify the unwieldy English spelling system. How successful have they been?

Not very. Yes, English does have one of the most complex system and rules for spelling. But once you’ve learned it, who wants to learn a new system – even if it’s simple to learn? And what about backwards compatibility – why obsolete all previously published books?

We wanted to steer clear of technology solutions that require users to actively learn a new system. Historically, no good solution has been available for applying this science to everyday reading materials.

Until now.

Asymmetrica leverages new technology to bring advances in reading science to massive audiences. Several recent advances makes this technological feasible, including improvements in: (1) web typography, (2) web and ebook distribution, (3) and mobile display hardware. Incombination, these advances now make it possible to represent word spacing in a way that augments comprehension of a language’s intrinsic pseudosyntactic structure.

But what is pseudosyntax? Pseudosyntax is an early stage of comprehension where your brain makes quick, rough estimates of the syntax based on statistical properties of the language. Later in comprehension, your brain periodically revises this guesstimate and tries to integrate it into a true syntax, a slower, more robust representation that handles exceptions to the guesstimates.

Asymmetrica’s novel approach is to directly compute the pseudosyntactic structure of a language and apply the measured structure subtly, but visibly, to the spaces. We do this seamlessly and in real time to improve everyday reading. Just as punctuation and word spacing have improved reading and made writing accessible to the masses, Asymmetrica’s spacing (Asym Spacing) takes the next significant step in improving everyone’s reading experience.

What may be a surprising statistic is that 43% of Americans have low literacy and have difficulty with moderately challenging literacy activities. Reading ability can’t be taken for granted, and our success as readers can’t be judged by a simple pass or fail. Our lives in the information age are complicated, distracting, and tiring with constant demands to read and process more and more information. Those fortunate to be above average in reading ability may not be performing at their best in today’s highly-distractible, always-connected environment.

Everyone can benefit from the reading improvements Asym Spacing delivers. Even if you think you’re already a good reader, consider the following scenarios in which phrase-­based formatting has demonstrated improvements to reading performance:

  • Students young and old who want to improve their learning performance, either for their primary language or one or more secondary languages;
  • Below average readers (Lake Wobegon children excepted);
  • Busy professionals who read every day in a distracting environment;
  • Readers who find that they often read while they’re tired, and want to maintain and improve their reading performance even under poor conditions;
  • Anyone who feels overwhelmed by today’s information overload and wants to improve their experience.

Whether you read news articles, journals, books, or even your Facebook feed, adding Asym Spacing can improve your experience.

Today, Asymmetrica Labs is launching the Asym Browser Extension for Chrome, Firefox, and Safari for pages in English. In the background, the Asym extension will dynamically apply our language models to the contents of your web pages. After your pages load, our spacing will quickly “snap” into place. We never change the text content, we just augment it with better word spacing.

We’re pleased to provide this benefit to everyone and, in our initial release phase, the extension will be free. Support for Spanish and other languages are in the pipeline.

Content creators and distributors can benefit as well. Asymmetrica offers testing plans for creators to format their content using our secure APIs. Run A/B tests between the original content and Asym Spacing, and find out how much Asym Spacing can improve your content for your readers – we only benefit if you do. Delivering your content with Asym Spacing to hundreds, thousands, or even millions of people will improve your users’ experience and your bottom line.

Keep an eye out for our future developments, and enjoy your reading!

This entire post was formatted using this technology. Did Asym Spacing improve your reading experience?

Let us know what you think on Facebook, Twitter, Google+, or LinkedIn (@AsymmetricaLabs) or in the comments below.

[1]    North, A. J., & Jenkins, L. B. (1951). Reading speed and comprehension as a function of typography. Journal of Applied Psychology.