[1] - http://www.zapreader.com/reader/index.php

Otherwise, it's an interesting thing. I would play with that kind of stuff, but for experimentation.

Also, the brain, or a little more accurately, your visual memory buffer, does indeed "cache" images, which does get deciphered in time, but as the text is flashed to your eyes at a constant rate, it doesn't make a difference. If I flash a stimulus at your eyes for 65ms, you may not consciously detect something, but it will register in your brain and you will process it, after the stimulus is gone. This is the visual buffer in action, but the processing is latent and does not occur at the same speed it takes to send the signal from the optical nerves to the visual cortex. If I flash many short pulses back-to-back, I won't bet on retaining most of the stuff.

The one-line RSVP has a fundamental speed limit before your forebrain cannot consciously register (hence willfully reflect upon) the material. While you can pause the reader, it requires intervention by your hands, which could be an unnecessary delay, because your eyes are far quicker and more fine-tuned for this task.

Anyhow, I'm sure zapreader has its uses. It does force you to read in a less lazy way, and is also useful (very definitely so) for training, but for those serious about reading very "fast," RSVP can only do so much.

I read that for a high school english class, and it changed the way I read. It teaches you how to absorb more information in less time.

The thing with speed reading is that you can increase the speed at which you read fairly easily, but your comprehension and recall will suffer greatly. I can probably read 500 words per minute, but comprehension is pretty low (~60%). At that point, it generally isn't worth it.

Speed reading is kind of a stressful activity, and there's only a very narrow set of situations in which it makes sense to speed read something. 95% of the stuff that I read, I read for pleasure, and in order to enjoy reading something you have to read it the "old fashioned" way.

If you want to teach yourself to speed read, books and the like won't really help much. It is kind of like learning to drive- you really have to just try it yourself. To start, sweep your hand over the words in an "infinity sign" pattern, with the height of each sweep being about 3 or 4 lines high. Don't read the words to yourself, just focus on the words above your finger. After awhile, you'll be able to detach your eyes from the lines, and you can stop sweeping your hand over the page. The key is to take in as much as possible with your eyes, and let your brain assemble everything in the background. You'll eventually get pretty good at associating key nouns with the verbs and adjectives around them. That's pretty much all you need to get 50-80% comprehension with most things.

(P.S. I'm not aware of any speed reading methods that don't come with a reduced level of recall/comprehension. I think it is just the nature of the beast.)

If you're enjoying your book, you don't wish it would end sooner. This would be like wishing your delicious meal would end sooner. And if you're not enjoying it, why are you reading it? The answer for many people is that they need to meet some quota to uphold their imaginary status as intellectuals. That's the real problem to be solved, not reading speed.

Writing is done to communicate ideas, ideas that the author took considerable time to formulate and put into a coherent work. Reading is one of the tasks in life that should be done very carefully. The only way to read quicker and retain the same level of comprehension is to practice reading.

My advice, start reading slower (and closer) and save time by stop reading fluff. Anything worth reading is worth rereading.

I am not trying to do it less carefully, I would like to do it more efficiently. To put in perspective: I know people (mostly English PhD students) who read 2x or maybe even 4x as fast as I do AND retain and understand what they read better than me. Of course, this doesn't apply to them trying to read SICP, but it does when reading Virginia Woolf. Specialization and practice have something to do with it; but, the point remains that there is something to be gained by making the process of information intake as efficient as it can be.

A good novel is a terrible waste of time if read quickly. But a blog post announcing that company X has done something vaguely interesting shouldn't get too much time.

It lets you set how many words are visible at a time, so you can use it to do the exercises jakewolf mentioned.

I could read Crime and Punishment at about 600-700 wpm (around 200-300 is normal) and I think I still had good comprehension. You can also adjust the speed as you go if passages get more difficult.

Also, when I read normally, I don't read linearly. I scan a page and narrow in on interesting words. Usually, if the material isn't too tricky, I can understand a paragraph very quickly this way. It is how I read online.

Finally, it really helps to think about what the author is saying, especially if I try to anticipate what's next. This gives me something to compare and contrast with, which makes me retain and understand what I read much better.

i think a large majority of arguments would go more smoothly if people would actually read all the words instead of guessing what the other person is saying.

of course that doesn't work for math textbooks and some sciences.

Seriously though, read all the words vs guess is a false dilemma.

I think that is the stuff of misunderstandings in a serious discussion. If someone is trying to say something new to you, then anticipating and trying to skip words based on knowing what he's saying won't work so well.

If reading for just fast consumption of information, I often will use a multiple sentence technique where you read 2-3 lines as you go across and reconstruct the ideas as you go. It is unusual, but it works eventually. It is not meant for deep understanding though. Kind of like reading something and only getting the highlights (albeit very detailed highlights).

BTW: good luck on your YC application ;)

I can read at somewhere between 575 & 600 wpm with http://www.zapreader.com/ (I hadn't tried that before - interesting idea). That's not comfortable, and I need to concentrate to do it, but I suspect I'd still be getting close to 100% recall. I'm not a reading technique expert, but it seems to me that the zapreader software is specifically designed to stop people using skimming techniques, so if you can read at a good speed on that you are doing pretty well.

When I'm reading a novel I generally read at 1 minute per page. On average that's probably around 250 wpm, and I find that relaxing and easy.

eye movement is the bottleneck in the i/o process. your mind will process the words fast, you're losing all of your speed on moving your eyes across each word and focusing on them. learn to use your peripheral vision more.

on each new line, instead of focusing on the first word, focus on the second, but use your peripheral vision to read the first. finish up the line in the same way but backwards. you'll find you'll be reading markedly faster already, without really losing any of the meaning, context, or beauty of language.

you can speed it up even faster by taking the same technique to an extreme, but i don't like to do that, myself.

Peripheral vision serves as priming mechanism. You can better prepare your mind to piece together relevant words and form meaning, but don't expect it to support reading like the center of gaze.

I admit that it is difficult to get used to, but for reading non-technical documents or for something other than pleasure, it is quite effective.

Also, do not vocalize the words externally, it should just be read naturally.

Stay away from speed-reading products. http://en.wikipedia.org/wiki/Speed_reading

Generally, proponents of speed reading don't know their stuff. Speed reading is based on the notion that the brain is underused in comprehending text; training would supposedly untap this potential. Don't count on it. Some say that if you read faster, you force your brain to focus, and so comprehension goes up. This works to an extent, noticeable, sure, but far less than hyped. And it is not a technique problem: it is a discipline problem. If you're generally focused at what you do, you're probably among the faster readers.

So most people read at about 250-400wpm, and this is related to the speed of the "inner voice," aka "subvocalization" (or, better, the phonological pathway/loop). So the idea is that if you reduce, or eliminate the inner voice, you can read faster. Programs that focus on these will claim 1000+ wpm. There is some truth to this, but these programs aren't all it takes.

Now when your eye sees text, it the brain takes time to parse it and register meaning. Conscious meaning, IIRC, happens after a 350ms mark, which means it is biologically impossible to read word by word, line by line, beyond the speed of the inner voice. This means that if you want to read faster, you're reading in blocks of text .

What does this mean? It means that when a normal reader sees a word, you're seeing a sentence or paragraph. Peripheral vision? Not necessarily. Peripheral vision doesn't have enough acuity to support parsing at a level required for detailed reading. IOW, most of the peripheral vision talk by these programs is BS.

The cases where PV is useful is when you have good familiarity with the text contours. How do you train this? The hard way: build your vocabulary, and read a lot. Reading is a "self-bootstrapped process," if you will. This is, AFAICT, the ONLY fool-proof, non bullshit method.

FYI, I recall two speed readers, who both read very fast but in different ways. First one reads blocks of text in a glance. He would see a paragraph and understand it in the time you read a sentence, in a kind of real-time, bulk-parsing process. The second one would read text in a "serial" fashion without regard to the meaning, into some kind of buffer, then think about it after he's done. The catch: both are geniuses.

The bottleneck in deciphering meaning from text is not at the text->parse stage (this is what SR programs focus on). It is from the parse->meaning stage. IOW, brain power. Fortunately you can train this, but probably not with some book or some program.

This is pretty thoughtstream. If you have other questions I'm happy to discuss.

Who are the two geniuses?

I've heard the block of text at a time thing before. In fact, I even think most people who I know that read significantly faster than me do so this way. The strange part is that when asked, people have told me that they have a hard time deciding whether they read a word a time or in larger blocks of text. My only question would be: can this skill be learned/trained? If so, how does one go about doing that?

I'm a bit ambivalent about the bottleneck, at least personally. I guess it depends on what kind of reading you do most of the time, but I'm inclined that I read more materials that are written in simple text that was meant to be consumed easily and that in those cases the bottleneck is in the text->parse stage. When reading technical material (which I also spend a fair amount of time doing) the bottleneck is definitely on the parse->meaning stage and that process probably takes several orders of magnitude more time, but speeding up the text->parse stage would be worth its while still. I say this because the process is a little bit different when reading technical material. You get through the text and you think about it for a long time, do problems, discuss, perform thought experiments, etc. But everything in the list seems to be a different stage than the parse->meaning stage. It is more of a meaning->true understanding/intuition stage. Something like first you read and you need to comprehend what you have been told at a "language" level. Then you go and understand the concepts more abstractly and at different levels. However, it is difficult to draw the line between those two stages.

You're right about the bottleneck, which makes my previous post's discussion not so good. There are multiple bottlenecks, and improving one area will not guarantee a result like those programs promise. If it's technical material, obviously, if you were Feynman you'd be able to quickly read hard paper without feeling uncomfortable (hence my reasoning about brain power). If it's Bearstein Bears, then, yes, you can fly over pages with total comprehension. As you said, here the speed of parsing text would make a big difference.

This is related to the "block of text" thing, which can be trained. It's not "reading in blocks" per se, but the training of fast visual pattern recognition. In this sense, it is similar to the pattern recognition the Rain Man used to count 270 or so matches in one glance. The most direct strategy then, is simply twofold. One, increase your vocabulary. Not just so you can answer test questions: you really need to know these words cold. Unfamiliar words take longer to retrieve from memory. Obviously, Charles Dickens would take longer to decode than Douglas Adams. (Hitchhiker is interesting though, since there are many invented terms; maintaining speed on this requires a relevant, but different, skill, and shouldn't be confused with that involved in vocabulary building). Two, read more, and ideally, when you read non-essential material (daily news perhaps), consciously push yourself (this takes discipline). The point of this is not for speed of parsing words. It's for recognizing the shapes of lumps of words. To illustrate: if you see "the Jedi reader reads fantastically quickly" a thousand times, you'll begin to parse that as a combined shape, in a single fixation. You will slow down, though, if you see "fantastically quickly, the Jedi reader reads." On the most part these two sentences are semantically equivalent, but the latter will take longer to parse, simply because it takes a different shape. But if you read that many times, you'll start chunking it like the other sentence. So reading a lot is simply to build up your arsenal of "familiar phrase shapes."

Again, this is the most foolproof way of training yourself. There are other things that may bring small improvements. But this is basically the heart of the matter. There are people who can parse unfamiliar word forms very quickly. The "bottleneck" here, as I understand it, is at the pipeline between CPU and RAM, and cache size. Genetics plays a big role here, so I usually don't bother.

Only if they're bad novels. When I read (good) novels, I'm constantly trying to slow myself down - to the point of literally mouthing the words or forcing myself to linger five minutes on a page. It takes so much longer to _feel_ a novel than it does to intellectually take in what's going on - so if I read too fast I end up knowing what happened, but not having experienced the novel.

As somebody pointed out on the 'heroes' thread, fiction often has a very high information content - it's just buried in symbolism and multiple meanings and the like.