As someone with horrible reading speed, the demo on their home page worked pretty well. I'd like to see a demo with some more advanced text. Marketing language is pretty easy to digest quickly. I'd like to see a selection from Gödel, Escher, & Bach or something to see how well I could comprehend more dense texts.
It's been awhile since I've really looked into anything speed-reading related, but as far as I recall, there aren't that many good techniques for speed-reading something like Godel Escher Bach, because the bottleneck for comprehension for really dense or deep texts is usually time taken to think about it. When I'm reading some like GEB, I usually end up reading it slowly, then stopping and thinking about it even after I've pretty carefully read it. Then I reread it again :)
Conversely, a lot of fiction, especially fiction on the lighter side, tends to repeat itself, or provide more narrative clues as to what's going on, so you can usually pick up what you missed in the context.
Of course, that's a really high level recollection of something I'm not an expert in :P
I also might just be trying to convince myself that I'm not a slow reader, heh
If anyone has any research or info to the contrary, I'd love to hear about it, because I'd definitely like to be able to absorb meaningful/dense information more quickly, even if it were just a small percent increase.
I'd also really like to see some third party research/tests with stuff like this and spreeder
the big thing when speed reading is to read it, just read it. Your brain will churn on it. Then come back and read it again. Your brain will churn more. Make notations in the margin the whole time. Now finally read it again. You'll get it all done faster than one read through and your brain will churn a ton on it. It's like when you come back to code tutorials a few days later after some practice and know what's going on a bit more and get further.
Then go back and read your notations (just put ticks or dots by the stuff that's important in the margins). And now you understand it and you can summarize it in the notes. And this cements the learning.
Of course this is a lot of work. I'd rather slam it once, brain index it and maybe come back if I care.
I've never been able to get the hang of speed reading (I think I lack the knack for skimming), but I listen to a lot of audiobooks, and have a similar procedure for dense sections. Even when I'm distracted and prone to zoning out, I always get essentially 100% comprehension because I just rewind back to the last part that I remember well. In fact, it seems I still pick up something when I'm zoned out, because I'll remember little aspects of the section on the re-read. So just blasting through does seem to lower the cognitive load of a re-read, even if your comprehension the first time around is very low or seemingly non-existent.
So another thing to note is sometimes what you are zoning out on isn't important so if you still get what's going on don't skip back. You can do the same thing going forward in a binary search type pattern. Works really well for stuff like Tolkien or Mellville whe're they're off on a tangent describing the sea or the somthings of somewhere and how onrey they are.
I can understand this; I think it depends on what your trying to get out of the book. There are some books I read (for example, I'm reading Stranger in a Strange Land) where I like the book well enough, but I'm not particularly into the prose and/or writing style, so I'm okay missing some things here and there.
I think that happens to me when I do "decompressing," where I'm reading a book to unwind my brain after having read nothing but papers or technical books for a few weeks.
Other times I'm just in the mood to sit down and really experience what the book has to offer. When I'm in that mood, I don't want to sacrifice really taking the time to let a book evoke imagery and mood, in the name of speed. I guess some folks can get that reading a lot faster, but I can't right now.
Hah! I just finished Stranger in a Strange Land this weekend :)
It was my introduction to Heinlein. I'd avoided older scifi because I'd wondered whether I would relate to it, and felt confirmed in that suspicion with Stranger. It's not that I'm not glad that I read it, and if I consider the context that it was written in circa the 1950s, then I can definitely appreciate how radical it must have been. But to my sensibilities, it feels tawdry and cheap, and thus insincere. It's in no way a fair criticism since it seems to have been part of the zeitgeist that led to the free love of the 60s and subsequent over-indulgence and exploitation of the 70s (and thus the dominance of the idea that sex sells and so on). And from what I gather it was rather sincere in so far as Heinlein was supremely interested in challenging prevailing mores of the time. But outside of that context, to me, it feels, again, cheap and tawdry, and just kind of stale.
No, I completely relate. I feel the same way reading it, and like you said, it's not fair, but that doesn't really take away my general feeling about it.
It's the kind of book where I think I can intellectually appreciate it's contribution and what it was in the context it was born out of, but I'm not really super interested in reading more like it.
As someone who tried using spreeder for a while, I can say that your retention is quite a bit lower than normal reading, especially when reading complicated texts, since you loose contextual clues (your brain is really good at doing look-aheads for clues).
Spreeder shows one word at a time, right? As someone who speed-reads regularly, I find that reading three words at a time works much better than one word, especially w.r.t. retaining context.
It also seems to matter whether chunks are split around punctuation or not, i.e. If the sentence is about to end after two words, the current chunk should not show the first word of the next sentence, just show the last two words of the current sentence. I couldn't find a speed-reading app that did this the way I wanted, so I wrote one for iOS (http://velocireaderapp.com/)
I also agree with the comment below that dense texts are best read without speed-reading. There's a sweet spot between easy and hard, boring and fun for texts that are ideal for speed-reading IMO.
I just bought a copy of Velocireader. I like it so far :)
Is there any way to raise the speed above 900wpm without going all the way to 1200wpm? The book I'm reading now is fine at 900wpm (and <=3 words per screen), but I think I could go a little faster. At 1200wpm I'm losing comprehension, so ideally I'd like to set it to 1000wpm. I can't see a way to do that.
+/- buttons at the top-right of the screen to make +5%/-5% adjustments to the speed without stopping reading would be great.
Currently there isn't a way to do that without adjusting the number of words. I've had other users complain about this though, so in the next update I'm going to make the time slider free to move, and introduce a stepper to adjust to the nearest 50wpm or something. This update has taken a while because I've been working on Pocket/Instapaper support.
Originally the slider was free to move; it was just too finicky to get down to a round-figure WPM on the smaller devices so I added the 'ratchet'.
Wow! Pocket/Instapaper support. I'm really looking forward to that. About 1/4 of the articles I save are too wordy, and this will help me get through them rather than leaving them for a day which never comes.
Yes, the biggest difference I can see is the red character, and there are plenty of other equivalent programs too. The key phrase is "Rapid Serial Visual Presentation".
Retention with spreeder is a matter of skill and practice. Don't give up easily. I've read difficult non-fiction in it with very good retention at 500 wpm. That requires a slower speed than I could read a simple novel (800+ wpm), but still much faster than I could read it in paper (under 300 wpm).
This took practice. I was reading economics books by Mises at 300 wpm in spreeder for a while and gradually increased the speed. I kept the speed low enough to understand it. However, it was never worse than regular reading.
Also, I started with 3 word chunks. I found that a lot easier than 1 word chunks. I normally use two word chunks now but that took work to get used to. I wanted to lower chunk size to reduce sub-vocalizing and eye movement. I've tried 1 word chunks but I lose over 100 wpm with them.
Anyway, low retention is caused by many things but is not inherent in RSVP.
Interesting! I've always been able to read pretty darn quickly, but I've never played with RSVP before. I gave spreeder a try and found that with a chunk size of three words, 2000wpm is surprisingly comfortable and 2500wpm is pushing my limits but not unrealistic.
It doesn't seem to work very well for me. I tried it in two different browsers and I get this http://i.imgur.com/cEMcR0M.png when I try to enable chunks.
If you use Spreeder, your eye still needs to move sometimes. That slows you down. Spritz has been carefully designed to ensure that your eye never needs to move, and never wants to move.
In my informal comparison tests, these improvements do make a difference.
When I was cramming for the series 7 (in a different life), I would use spreeder.com. I would cycle through chapters and pretty much memorize them before moving on.
When I tried the same site for reading blog content and short stories, however this method falls flat. I found it great for memorizing but not for absorbing or following a narrative.
At first I thought "Oh dear, another gimmick." but the demo is very compelling. I wish it didn't make the syllables in my head sound like such a metronome, but it definitely works. This could really change how we "ingest" knowledge.
Also, I wonder how tired I'd get after reading like that for thirty minutes or a few hours.
Have you tried http://www.spreeder.com/ ?
No association, but when working for it I find I can hit 1000 without losing information.
Good when I need to cram.
I had a summer job once in high school working in the office of a small business that published test preparation books and related products. They had devoted one of their rooms to a speed reading course; one day the teacher told me that as an employee I could take the course at no charge. I said that I could already read as fast as I could think, and he rolled his eyes and said, "You people who think." I believe that this skill might be useful to some people, but I prefer to avoid reading things that could profitably be speed read, because life is too short for that. For example, I haven't read the OA.
Yes, where "comprehend" means to meaningfully understand and appreciate. I was trying to express the idea that if something can be comprehended, in this sense, as fast as it can be speed read, then perhaps it's not worth reading.
One of the things I picked up from "How to Read a Book" (where they put forward that speed reading is usually just bringing slow readers up to speed), is that not everything is worth reading. There is so much to read out there, that you can't possibly read it all (relevant: http://what-if.xkcd.com/76/), and you have to pick and choose. One way to weed out what is relevant is to apportion the correct amount of time and attention to each thing you read. Some things deserve none of your time and attention; others may merely require a once through skimming to get everything of value out of it. Other works you may never exhaust (HtRaB's authors posit that only great books fall into this category, by definition). Perhaps much more important than increasing the speed of reading is developing the skill to be discerning in what and how deeply different texts should be read.
Most informational books are packed with examples and context. Sometimes you need that stuff, but other times you can get 90% of the information you need just by reading the introduction, table of contents, and chapter conclusions.
I think it's important to understand your motivations for reading, and then choose a method that best fits your intended goals.
Reading is like eating — sometimes you want to savor a book and other times you want to grab and go.
The best way to read books is with enlarged font on a tablet or e-reader that's resting on an inclined treadmill. You're already ahead of the game time wise because you're exercising and reading at the same time.
That's interesting. A technology like this was predicted, on theoretical grounds from reading research, in the book Reading in the Brain[1] by Stanislas Dehaene (which I highly recommend as a very good read).
For years 500 words per minute (approximately one printed page per minute) has been my baseline speed for most of the material I read in English. I read slower in my second languages, of course. Back in my college days, when I wanted to make sure I wasn't being slowed down in my studies by a too-slow reading speed, I read a lot of books from the university library about reading skill improvement, and several jointly suggested that improving vocabulary improves reading speed. I took a course about English vocabulary based on Latin and Greek word roots, and that did seem to help for years afterward in both reading speed and reading comprehension.
In this context, "second languages" is a quite normal designation for language(s) acquired after one's native language(s) were acquired. My user profile lists most of mine. I read best in Chinese and in German, besides English.
Ethnologue states that there are 7105 languages in the world[0]. Therefore, it seems reasonable to assume that you can have x second languages, where x is any number between 0 and 7104.
Not true in idiomatic English. Any language other than your 'first' can be called a 'second language'. As a common example of this usage, courses to learn English for non-native speakers are called "English as a Second Language (ESL)" - no matter how many other languages the students already know.
Then I guess only native English speakers can have many "second" language. In many other countries there are 1st foreign language and 2nd foreign language classes in the school. And those books are called "English as a foreign language" (Englisch als Fremdsprache).
I tried an open source version of something similar when I was still in high-school by reading Dracula. The key benefit was't the speed that I could read using the tool, it was that I found I could read much much faster when I later picked up a book.
I recently showed spreeder to my girlfriend who found the same thing.
@ivan_ah Yep I totally use it for proofreading. Often I find mistakes in news articles and such and end up being that person notifying them of a mistake :)
This is called RSVP. It is easier (for the same WPM) with 2-4 words at a time than 1 word. I really hope this catches on more and allows different chunk sizes.
500 wpm with no training is an exaggeration. wpm also depends a lot on which book you're reading. however, RSVP is the fastest way to read and way over 500 wpm is possible with practice.
Their FAQ seems to suggest it's more than RSVP, the addition of the red 'centre' and the differing alignment (not centred on the middle of the word but slightly to the left). I've used both and I found Spritz more comfortable on the eyes, less eye movement.
> Next, let’s talk about subvocalization. This is the process whereby, whenever you read, you talk to yourself, repeating every word in your head. The issue here is that most people can only speak at about 180wpm, maximum. Increasing speed beyond that means not saying every word to yourself.
This makes no sense to me. I've read at 600 wpm while sub-vocalizing every word. That's much much faster sub-vocalizing than they claim is possible. I've experienced it.
This makes no sense to me. Spoken English does not reach 10 words per second even in New York; to hear it, even inwardly, would be to hear a jam of syllables with no respite and hardly any time to parse out the phonemes.
you're basically denying the definition. echoing in the brain isn't the same as actually getting the muscles in the throat or the mouth to move. It does slow things down but the muscle movement is way slower and some people don't get rid of it.
As an Australian crossing the US, I looked forward to NYC because finally I'd be in a place where they talked at a proper speed. But when I got there, while it was nice to not have to slow my speech to be understood, I still found the locals were speaking a touch too slow for my liking.
I wonder whether this uses the front-facing camera for blink detection, particularly at higher WPM settings. A blink typically lasts 100-400ms (per Wikipedia), and 500 WPM is .8333... words per 100ms. A blink thus seems very likely to miss a word or two once you're using this the way its creators intend. I don't see any mention of blinking or mitigating its impact on the site, though.
Blinking might explain why I strongly prefer 2-3 word chunks, instead of 1 word chunks, when I read with RSVP. Interesting. I can read some texts at 800 wpm with 2 word chunks but find it a lot more problematic with 1 word chunks so it's flashing twice as fast.
Thank you for helping me figure out why my eyes started burning when I was spritzing. I think you subconsciously try to avoid blinking becuase you know you'll miss a word.
Perhaps it's called Spritz because the user is meant to keep a spray bottle full of water on hand, spritzing occasionally, so as to obviate the need for blinking :)
I guess a software could insert a "blink break" every couple of minutes. They could even display an a video commercial during that time, to monetize your blinks.
Since everyone seems to try to patent everything: in this comment I claim the invention of a killer business model for speed reading apps. 25 Feb 2014, 4:30pm Sydney time :)
Congratulations on being one of the millions of internet users who have no idea how the patent system works. First to file is purely for determining priority of an application; the date at which the patent office (or another patent office under the Paris convention) comes to know about the invention. It's used to resolve a) which prior art is relevant (anything published publicly before this date) and b) if two people try to patent the same thing, which one gets it (gross simplification here). America was the only country which had the "First to invent" priority, which the rest of the world has deemed for decades as completely daft and a huge administrative nightmare.
The "invention" has been made public (in paulftw's dated comment) before anyone has filed for a patent. It's no longer patentable material in a first-to-file system as there's now evidence that it's obvious to practitioners of the field, and this evidence predates any filing gcb0 can make. In first-to-invent, he could've claimed an invention date prior to the comment. Sound right?
All I wanted is take this patent out of the equation forever.
My thinking is that I've created prior art that would make the idea much much harder to patent. Not a patent lawyer, so may be wrong.
As for how harmful it was for my anonymity to mention Sydney - well, NSA knows way more than that.
Besides, I could be sending you all down the wrong trail. or better - confusing a ML algorithm.
Very interesting stuff. This demo would be more interesting if I could paste in my own text to confirm my comprehension. I could process Spritz's marketing statement at 500wpm, but it's fairly simple text and may have been optimized for the delivery mechanism. I'd like to try it on some text that I actually have an interest in reading.
I played around with something like this a while ago (my files have 2004 timestamps, but those may have updated when I copied stuff to a new computer). I had a command-line based program that took text and displayed it a word at a time. If anyone would like to play around with it, here's my source.
Compile: c++ speedread.cpp
Usage: ./a.out [delay_in_ms] < text
Should work on Mac and Linux, and maybe Windows if you uncomment the first line. (Looks like a #include <stdlib.h> needs to be included to make it compile on Linux. Not sure what I don't need that on Mac).
It's kind of fun. I played around with showing words at a constant rate, and with throwing in an extra delay whenever there was punctuation. The version below does the latter. I had thought of this as a possibly good way for displaying text on the limited phones of the early 2000s. Also, I thought it might be an interesting way to display status information on many processes at once. A display line wide enough to display, say, 10 words could be used to display 10 status messages at once instead of one.
Oh...2 clause BSD license, if anyone cares.
//#include <windows.h>
#include <stdio.h>
#include <iostream>
#include <string>
#include <unistd.h>
#include <signal.h>
#include <sys/time.h>
typedef unsigned long DWORD;
bool cont = true;
void
stop( int sig )
{
cont = false;
}
void
Sleep( int msec )
{
usleep( msec * 1000 );
}
DWORD
GetTickCount()
{
static unsigned long base = 0;
struct timeval tv;
gettimeofday( &tv, 0 );
if ( base == 0 )
base = tv.tv_sec;
return (tv.tv_sec - base) * 1000 + tv.tv_usec / 1000;
}
int main( int argc, char *argv[] )
{
int delay = 250;
std::string word;
if ( argc == 2 )
delay = atoi(argv[1]);
signal( SIGINT, stop );
std::cout << "\n\n\n";
int words = 0;
DWORD start = GetTickCount();
std::string display_word = "";
while ( std::cin >> word && cont )
{
Sleep(delay);
++words;
if ( display_word.length() )
display_word += ' ';
display_word += word;
//if ( display_word.length() < 5 )
//continue;
word = display_word;
display_word = "";
int punc = word.find(".");
if ( punc == std::string::npos )
punc = word.find(",");
int pad = 15 - word.length()/2;
for ( int i = 0; i < pad; ++i )
std::cout << " ";
std::cout << " " << word << " \r";
std::cout.flush();
if ( punc != std::string::npos )
Sleep(2*delay);
}
DWORD stop = GetTickCount();
float seconds = (stop - start) / 1000.0;
printf( "\n%d words in %f seconds\n", words, seconds );
if ( seconds != 0 )
printf( "%f words/second, %f words/minute\n", words/seconds,
(60*words)/seconds );
return 0;
}
I also did an RSVP app many years ago. I did mine for j2me - http://kybernetikos.com/bookreader-and-bookreaderdeployer/ and read books from gutenberg on my mobile phone at the time. It had pause and jump forward/backward as many people in this thread are saying is important to them.
The FAQ tries to make the claim that Spritz is not just another RSVP speed reader (although I'm not sure how much I believe it).
> Reading with RSVP is somewhat stressful and the reading speed is limited. Reasons therefore are the word alignment and display times. Briefly, RSVP still forces the eye to perform saccades (eye movements) while removing targeting these saccades by peripheral visioning. Spritz instead takes saccades out of the game by applying a new method of word alignment. Our technology is based on the science of how people read, how they learned to read when they were young, and what your eyes expect when you are reading.
Interesting. One thing that struck me is I would like a small pause on full stop or comma.. to digest that thunk before continuing on and allow for blinks.
This is very interesting (and as has been pointed out in other comments, not exactly a new idea). I think its proper application, though, is in a tool for practicing reading, not for actually reading tests.
One major thing this approach looses is the inherent non-linearity of text. If I miss a fact and want to back up a bit, or want to pause for a moment and think about something, I can (when reading) without even thinking about it. Even if this approach is faster overall, it makes reading more like listening to audio or watching a video; it's a big pain to rewind or pause.
Secondly, this is actually slower than true speed reading or skimming because it forces you to read every word. Truly accomplished readers will often read material at a very superficial level, only dipping in and reading consistently when they encounter a novel concept. Essentially, they can use semantic compression to increase their reading speed by only bothering to read what they find to be relevant. This operates on every level, from the page to the chapter to the paragraph to the sentence. It isn't perfect, of course, but it's always possible to back up if one finds crucial information has been missed.
Finally, even though this tool is truly excellent for breaking the subvocalization habit that hampers most slow readers, once you learn how to read without subvocalizing it becomes a bit redundant. For example, looking at Spritz, I cranked the speed to 500. It felt pretty good, like I was reading fast. Then I went and took a traditional reading speed test and clocked in at 700wpm, with 95% comprehension. So I'm not sure my overall speed is better with Spritz.
That said, I'll probably keep coming back to this or technologies like this, now that I'm aware of them. They seem a really good way to force oneself into the speed-reading mindset - I have a feeling that doing this for 60 seconds before a normal reading session would improve reading speed substantially.
Whenever you hear something, you hear it word by word. And similarly you can't immediately go back (you have to stop it, "rewind it" and play it again).
Huh, strange thing is it seems like I just can read that quickly anyway. Started on the banner at 250wpm, pretty comfortable, went straight to 500wpm on the banner and still fairly comfortable to read. So I thought I'd have a go at just reading some plain text as quickly as I comfortably could and it turns out that I can read more quickly than I do. I don't know why I don't read more quickly than I do, perhaps it down to comprehension, but I think in the future for articles I'm not trying to fully absorb that I may try to just read them more quickly!
I've thought about this before too, and I've realized that unless I give it some thought, my eyes read at about a talking speed, especially when reading essays and comments such as yours. My guess is since we hear the words in our heads, we prefer to hear them as if they were naturally spoken.
My comfortable reading speed is a bit above 700wpm. On a book with proper line length, or on my Kindle. Strangely, or not at all, I only read at that speed when parsing non-fiction texts. Reading a novel, the reading times on my Kindle boil down to 300wpm.
Some stuff needs to be savored, I guess. Even at that speed, I'm known for devouring books, so I am probably reading too fast.
On a side note, if Spritz is that good, I'd like to try it out above my comfort zone. Say, 1000wpm.
I thought the same thing. Nothing a little JavaScript can't fix:
// Insert drop down item.
var li = document.createElement('li');
li.innerHTML = '<a href="#">1000wpm</a>';
li.setAttribute('data-value', '1000');
document.querySelector('.speed ul').appendChild(li);
// Create 1000wpm script
var s = document.createElement('script');
s.id = 'learntospritz_en_1000';
s.type='spritztext';
var data = JSON.parse(document.querySelector('#learntospritz_en_500').innerText);
data[17] = '1000';
s.innerText = JSON.stringify(data);
document.body.appendChild(s);
I just did a random reading speed test and got 1640wpm with 90% comprehension, and I definitely do not read linearly -- it's hard to explain, but I often find that I scan far ahead and see the ends of sentences before their starting words, then wander around somewhat randomly in the middle of them. I'll notice typos and grammar errors several sentences or even a paragraph after where it occurs, probably meaning that I'm taking huge "gulps" of words and internally buffering it, then processing it very nonlinearly. I've never taken any courses on speed reading, but I guess I developed this habit from reading code (where a lot of the time is spent seeking). The Spritz experience was a little disorienting since I kept wanting to seek around in the text, and being linearly spoonfed it gave me the feeling of losing track of what I'd just "read".
Gave it a go and while I did ok, I agree with the sentiment about reading being non-linear.
That said, I don't see it as being a blocker exactly. A UI that let you pause the stream, and then if asked revealed some kind of tome navigation can't be too difficult to do. Also, images (photos, diagrams) could be revealed in a similar way.
That way the benefit of speed reading is retained, but with the added benefit of being able to skip backwards and forwards, and pausing to look at graphics.
What traditional reading speed/comprehension test did you take? I just took the one at readingsoft.com, and I felt like I could have guessed most of the answers in the test without even reading the text. I'm wondering if mosts comprehension tests actually do a poor job to testing comprehension.
This stuff always interests me because I'm a terribly slow reader (I think I did ~180 wpm on the above test). I'd really like to be able to improve without hurting my comprehension. However, when I discuss an article with someone who reads quickly I usually find that they've missed some important pieces of information. The sort of gaps exposed by real conversation may not show up on a simple multiple choice quiz. I'm just guessing here though. Based on my own experience, hearing that someone read 700wpm with 95% comprehension is like hearing that someone has broken the laws of physics.
(Not the OP, but I just take the test.) I got 6 out of 11 without reading the text. I didn't even skim the text. And in most of the questions I got wrong, the correct answer was my second choice.
> this tool is truly excellent for breaking the subvocalization habit
Which, according to all the research I've seen, is exactly what you don't want to do if you want to retain any information. (I couldn't find the study I remembered reading recently, but this article references some such studies http://skeptoid.com/episodes/4229)
I was speed reading since Third Grade and remember making a deliberate effort to slow down in Junior High because I was not retaining enough information to write essays. During High School, I broke the curve on A.P. US History on an open notes final without having a single note to bring to class.
Auditory learning styles may require the subvocalization to succeed.
You don't want to break it in the sense that you stop doing it altogether. Just as lukev said, you would want to learn controlling it to your advantage - slowing down and speeding up depending on what you a reading.
> this tool is truly excellent for breaking the subvocalization habit
I found the exact opposite when playing with the demo on the site. Perhaps I fall into the camp of the 1k+ wpm readers, and by extension am a non-linear, non-vocalizing reader, but going this slow, and seeing only one word at a time made me conscious of my mind "saying" every word.
> it forces you to read every word
Also a pain in the ass - I found if you blink, you'll end up missing words. At a consistent 500 wpm clip, a blink (avg 100-400 ms) could miss anywhere from 1 to 3 words. As you get faster, blinks are even more costly. Somehow (I assume through years of practice), when speed reading, your mind is able to properly account for blinks. This also brings up an interesting point - the creators claim that this method reduces eyestrain, but it makes me feel like I'm forced to hold my eyes open for fear of missing something. I can't imagine that's any less straining on the eyes.
Maybe in the future they'll add eye tracking and pause the text when it detects a blink. And we thought Kindle tracking your reading habits was scary.
Over the past year, I have dedicated hundreds of hours towards improving my reading speed. I usually use Acceleread on my iPad; most recently I used it to read D.H Lawrence's Sons & Lovers at around 750-800 wpm. I also use Beeline -a different type of reading tool with some additional functionality for dyslexics- when reading web pages.
I've learned that certain types of prose make for poor speed reading material. Without a "rewind" option, there is no way that I can use these apps to read anything instructional or anything involving a high level of detail (such as a mystery novel). I understand that allowing the user to rewind would sacrifice speed, but speed can't always take precedence over clarity.
Also - in terms of the text- there needs to be a better method for handling quoted text. With speed reading apps that give the user the option of seeing chunks of words, it is far too easy to lose track of who is speaking when one chunk of words has an opening quotation mark and the next chunk of words [may or may not] have a closing quotation mark. I'm guessing that this problem is exacerbated by the fact that - in Spritz's case- words are only presented one at a time.
Here are some things I would I would like to see adopted by speed-readers:
Rewind functionality
Punctuation-based text color:
If the text is in between quotation marks, parenthesis or brackets, this text should be shaded a slightly different color than the rest.
Contextual deceleration:
I would love it if a speed reading app would automatically slow down at the end of a chapter. The last few sentences of a chapter are usually crafted with especially deliberate care and are meant to be savored.
More customization options for dyslexics:
For example, Beeline allows users to chose a specific type of font that is specifically geared towards dyslexics. In the future, I'd like to see speed readers enable these types of fonts (as well as customization options for line height and letter spacing).
For the punctuation-based text color, it would be great if the software was able to keep track of, or at least attempt to keep track of, what character was speaking and assign a specific color to that characters' text.
For speed reading something like a novel, this would help the brain quickly associate a color with a character and avoid the need to catch the text that makes it clear who is speaking.
Ok... but almost all that you said is a UX Issue...
Imagine an UI with a circle and the spritz box above. You put your thumb in the middle, push right it starts to show the words forward, push farther from midle => faster. Push it left => slower. Left after middle, rewinding text.
You could even use up, down to go back entiry frases or paragraphs.
Regarding linearity, you could use color/color fading to wrap phrases, or show two to four words per time moving just the red letter according to speed.
Why wouldn't they just implement a stop/start button and a zoom out/in button? If you want to return to a missed word or reread a section, zoom out to see a page and then touch a word.
Fun to make. I'm also curious about the overall benefits. I find myself distracted by other thoughts if I don't have subvocalization happening.
If you want the ability to move backwards, I feel like that's possible with a good NLP library. If you skip back two sentences, or enough sentences to match some word or syllable threshold, it may be useful. In that case, you can play a few words, pause for a second or so to let the user decide if they've found the right spot (and context switch into reading) or if they should continue going back. I tried hacking in a quick back button that went back X words, but it was very disorienting. But in any case, it isn't as interaction free as doing so when reading a block of text is. But to counter your argument: you still have to keep your position even with text or seek for your position again, so there's a downside to those benefits.
I'm not sure how Spritz handles things like this, but I felt like making my own based on what I saw on the demo really quick to get a better understanding. It's a curious thing, maybe I'll check out the application sometime.
Sweet fiddle, but putting the emphasis on the center letter puts it way too far to the right for any long word. Compared to the example on the Spritz website it's quite awkward. My eye constantly wants to shift left, while I can comfortably zone out on the highlighted letter in the Spritz example.
I'm curious as to how they calculate the "Optimal Recognition Position", but there is probably a better way to approximate it. I'm going to play around with some different positions.
I find this system really drives up my concentration by sustaining the pace and letting me read without eye scanning.
In the past few years I begun using text to speech coupled with visual reading with the same effect - it makes my concentration more powerful. I don't even like to just read any more - or God forbid - post a comment like this without listening it out aloud. TTS support during reading also has the advantage of allowing me to look away for a time, maybe do something in the room while continuing reading (or reading while I drive).
My favourite voice is Alex from Mac OS. I read dozens of books, thousands of articles and forum threads like this. With forums I still need to scrape out some formatting text with a quick Grease Monkey script though, but it's so fun to hear you out aloud!
This is amazing. Not only can I now magically read at 350WPM, but I noticed what I believe is a spelling error at 350WPM ("Integrate" was spelled as either "Intergate" or "Intergrate")
As a german i find the name a bit unfortunate.
To spritz = spritzen means to squirt/spray coat and in german is often associated with ejaculating. I don't know if this name would fly around here, but maybe thats why it would. Could make for some funny conversations.
i don't speak a word in german other than beer varieties and i also read that as something like spraying words. only on the second take i got the sprint similarity. ...Completely missed the climax though.
Hehehehehehe! Do you know what Spritz means in German? ROFL! LMAO!
One of our founders is from Munich, so yes, we know. We bet you
won’t forget it though, will you?
It may be sour grapes, but why should any one read at such high speeds. What is the virtue of reading at 500 wpm. I don't think that the time saved is such a big deal unless one is reading 40 - 50 books a year or reading 300+ technical papers a year. And comprehension suffers at high speed. I do just fine reading at 250 wpm. Some times I feel just being good enough has lost its value.
I have never had much luck speed-reading technical papers. It's too difficult for me to interpret and think critically together -- words flashing by will go unevaluated I think.
If you had said you do just fine at 150 wpm may be i can accept it. But you are saying 1 wpm which means you are not trying to give constructive criticism. Most of the gen-pop is around 200-250 wpm which is why I said it is good enough. Don't be a troll.
Someone has implemented a similar process in an iOS app called Velocity (http://velocityapp.com). I downloaded it a few weeks ago and have tried it off and on, I have trouble for the use case of reading feeds (it integrates mostly with rss feeds I think) in that I tend to do that reading in distraction-prone environments but it might work better for other mediums
I know my eyesight is bad (test scheduled for today and new glasses overdue), but I really struggled with this.
I tried multiple times to read the example, but would stumble to comprehend the shape of a word or two, lose my way, and have to rewind back to the beginning.
Overall it felt considerably slower and more broken, less rhythmic. My comprehension felt lower, I felt I spent far more time comprehending the shapes of the words rather than the totality of the meaning expressed.
I know my eyesight is bad, that I am long-sighted and colour blind, but this was a painful and slow reading experience.
Some words took so much concentration to figure out that I'd forgotten the words that preceded it, and lost context.
Without the ability to rewind this or to see a larger context, I would avoid this like the plague.
Very interesting, and I did manage to read 500WPM without too much of a struggle, but I did not enjoy reading with this technique. I feel uncomfortable not knowing when a paragraph is going to end, and the relentless onslaught of words took the fun out of it.
215 comments
[ 3.6 ms ] story [ 246 ms ] threadConversely, a lot of fiction, especially fiction on the lighter side, tends to repeat itself, or provide more narrative clues as to what's going on, so you can usually pick up what you missed in the context.
Of course, that's a really high level recollection of something I'm not an expert in :P
I also might just be trying to convince myself that I'm not a slow reader, heh
If anyone has any research or info to the contrary, I'd love to hear about it, because I'd definitely like to be able to absorb meaningful/dense information more quickly, even if it were just a small percent increase.
I'd also really like to see some third party research/tests with stuff like this and spreeder
Then go back and read your notations (just put ticks or dots by the stuff that's important in the margins). And now you understand it and you can summarize it in the notes. And this cements the learning.
Of course this is a lot of work. I'd rather slam it once, brain index it and maybe come back if I care.
I think that happens to me when I do "decompressing," where I'm reading a book to unwind my brain after having read nothing but papers or technical books for a few weeks.
Other times I'm just in the mood to sit down and really experience what the book has to offer. When I'm in that mood, I don't want to sacrifice really taking the time to let a book evoke imagery and mood, in the name of speed. I guess some folks can get that reading a lot faster, but I can't right now.
It was my introduction to Heinlein. I'd avoided older scifi because I'd wondered whether I would relate to it, and felt confirmed in that suspicion with Stranger. It's not that I'm not glad that I read it, and if I consider the context that it was written in circa the 1950s, then I can definitely appreciate how radical it must have been. But to my sensibilities, it feels tawdry and cheap, and thus insincere. It's in no way a fair criticism since it seems to have been part of the zeitgeist that led to the free love of the 60s and subsequent over-indulgence and exploitation of the 70s (and thus the dominance of the idea that sex sells and so on). And from what I gather it was rather sincere in so far as Heinlein was supremely interested in challenging prevailing mores of the time. But outside of that context, to me, it feels, again, cheap and tawdry, and just kind of stale.
It's the kind of book where I think I can intellectually appreciate it's contribution and what it was in the context it was born out of, but I'm not really super interested in reading more like it.
As someone who tried using spreeder for a while, I can say that your retention is quite a bit lower than normal reading, especially when reading complicated texts, since you loose contextual clues (your brain is really good at doing look-aheads for clues).
It also seems to matter whether chunks are split around punctuation or not, i.e. If the sentence is about to end after two words, the current chunk should not show the first word of the next sentence, just show the last two words of the current sentence. I couldn't find a speed-reading app that did this the way I wanted, so I wrote one for iOS (http://velocireaderapp.com/)
I also agree with the comment below that dense texts are best read without speed-reading. There's a sweet spot between easy and hard, boring and fun for texts that are ideal for speed-reading IMO.
Is there any way to raise the speed above 900wpm without going all the way to 1200wpm? The book I'm reading now is fine at 900wpm (and <=3 words per screen), but I think I could go a little faster. At 1200wpm I'm losing comprehension, so ideally I'd like to set it to 1000wpm. I can't see a way to do that.
+/- buttons at the top-right of the screen to make +5%/-5% adjustments to the speed without stopping reading would be great.
I'll get back to reading...
Currently there isn't a way to do that without adjusting the number of words. I've had other users complain about this though, so in the next update I'm going to make the time slider free to move, and introduce a stepper to adjust to the nearest 50wpm or something. This update has taken a while because I've been working on Pocket/Instapaper support.
Originally the slider was free to move; it was just too finicky to get down to a round-figure WPM on the smaller devices so I added the 'ratchet'.
This took practice. I was reading economics books by Mises at 300 wpm in spreeder for a while and gradually increased the speed. I kept the speed low enough to understand it. However, it was never worse than regular reading.
Also, I started with 3 word chunks. I found that a lot easier than 1 word chunks. I normally use two word chunks now but that took work to get used to. I wanted to lower chunk size to reduce sub-vocalizing and eye movement. I've tried 1 word chunks but I lose over 100 wpm with them.
Anyway, low retention is caused by many things but is not inherent in RSVP.
If you use Spreeder, your eye still needs to move sometimes. That slows you down. Spritz has been carefully designed to ensure that your eye never needs to move, and never wants to move.
In my informal comparison tests, these improvements do make a difference.
When I tried the same site for reading blog content and short stories, however this method falls flat. I found it great for memorizing but not for absorbing or following a narrative.
Also, I wonder how tired I'd get after reading like that for thirty minutes or a few hours.
Most informational books are packed with examples and context. Sometimes you need that stuff, but other times you can get 90% of the information you need just by reading the introduction, table of contents, and chapter conclusions.
I think it's important to understand your motivations for reading, and then choose a method that best fits your intended goals.
Reading is like eating — sometimes you want to savor a book and other times you want to grab and go.
For years 500 words per minute (approximately one printed page per minute) has been my baseline speed for most of the material I read in English. I read slower in my second languages, of course. Back in my college days, when I wanted to make sure I wasn't being slowed down in my studies by a too-slow reading speed, I read a lot of books from the university library about reading skill improvement, and several jointly suggested that improving vocabulary improves reading speed. I took a course about English vocabulary based on Latin and Greek word roots, and that did seem to help for years afterward in both reading speed and reading comprehension.
[1] http://readinginthebrain.pagesperso-orange.fr/intro.htm
AFTER EDIT, TO REPLY TO QUESTION:
In this context, "second languages" is a quite normal designation for language(s) acquired after one's native language(s) were acquired. My user profile lists most of mine. I read best in Chinese and in German, besides English.
[0] https://www.ethnologue.com/statistics
Otherwise you introduce ambiguity, if I learned 50 words in Spanish and then mastered French, does that mean French is my third language?
I recently showed spreeder to my girlfriend who found the same thing.
This was great.
@achalkley what speed do you use? This is where I have mine: http://i.imgur.com/nIO0gRw.png It's a little quick, but it keeps thing interesting ;)
Mine is a little slower than yours - http://i.imgur.com/WGW1QWk.png - I use cmd + alt + shift + s as my short cut.
500 wpm with no training is an exaggeration. wpm also depends a lot on which book you're reading. however, RSVP is the fastest way to read and way over 500 wpm is possible with practice.
> Next, let’s talk about subvocalization. This is the process whereby, whenever you read, you talk to yourself, repeating every word in your head. The issue here is that most people can only speak at about 180wpm, maximum. Increasing speed beyond that means not saying every word to yourself.
This makes no sense to me. I've read at 600 wpm while sub-vocalizing every word. That's much much faster sub-vocalizing than they claim is possible. I've experienced it.
no respite is a matter of practice. time to parse the content is not that hard if you've already learned to RSVP faster.
you're basically denying what you don't know how to do, for no apparent reason.
According to (a) if someone files a patent first then if there is any prior art that and any subsequent patent applications will fail.
According to (b) if someone files a patent first then out of that and any later patent applications the first one will win.
Seems to fit with "first to file" to me.
My thinking is that I've created prior art that would make the idea much much harder to patent. Not a patent lawyer, so may be wrong.
As for how harmful it was for my anonymity to mention Sydney - well, NSA knows way more than that. Besides, I could be sending you all down the wrong trail. or better - confusing a ML algorithm.
Compile: c++ speedread.cpp
Usage: ./a.out [delay_in_ms] < text
Should work on Mac and Linux, and maybe Windows if you uncomment the first line. (Looks like a #include <stdlib.h> needs to be included to make it compile on Linux. Not sure what I don't need that on Mac).
It's kind of fun. I played around with showing words at a constant rate, and with throwing in an extra delay whenever there was punctuation. The version below does the latter. I had thought of this as a possibly good way for displaying text on the limited phones of the early 2000s. Also, I thought it might be an interesting way to display status information on many processes at once. A display line wide enough to display, say, 10 words could be used to display 10 status messages at once instead of one.
Oh...2 clause BSD license, if anyone cares.
https://gist.github.com/gosub/9206636
http://pastebin.com/MyB0UUFD
The FAQ tries to make the claim that Spritz is not just another RSVP speed reader (although I'm not sure how much I believe it).
> Reading with RSVP is somewhat stressful and the reading speed is limited. Reasons therefore are the word alignment and display times. Briefly, RSVP still forces the eye to perform saccades (eye movements) while removing targeting these saccades by peripheral visioning. Spritz instead takes saccades out of the game by applying a new method of word alignment. Our technology is based on the science of how people read, how they learned to read when they were young, and what your eyes expect when you are reading.
http://www.spritzinc.com/faq/
http://jsfiddle.net/P7c3s/
One major thing this approach looses is the inherent non-linearity of text. If I miss a fact and want to back up a bit, or want to pause for a moment and think about something, I can (when reading) without even thinking about it. Even if this approach is faster overall, it makes reading more like listening to audio or watching a video; it's a big pain to rewind or pause.
Secondly, this is actually slower than true speed reading or skimming because it forces you to read every word. Truly accomplished readers will often read material at a very superficial level, only dipping in and reading consistently when they encounter a novel concept. Essentially, they can use semantic compression to increase their reading speed by only bothering to read what they find to be relevant. This operates on every level, from the page to the chapter to the paragraph to the sentence. It isn't perfect, of course, but it's always possible to back up if one finds crucial information has been missed.
Finally, even though this tool is truly excellent for breaking the subvocalization habit that hampers most slow readers, once you learn how to read without subvocalizing it becomes a bit redundant. For example, looking at Spritz, I cranked the speed to 500. It felt pretty good, like I was reading fast. Then I went and took a traditional reading speed test and clocked in at 700wpm, with 95% comprehension. So I'm not sure my overall speed is better with Spritz.
That said, I'll probably keep coming back to this or technologies like this, now that I'm aware of them. They seem a really good way to force oneself into the speed-reading mindset - I have a feeling that doing this for 60 seconds before a normal reading session would improve reading speed substantially.
Whenever you hear something, you hear it word by word. And similarly you can't immediately go back (you have to stop it, "rewind it" and play it again).
Very interesting concept, though.
Some stuff needs to be savored, I guess. Even at that speed, I'm known for devouring books, so I am probably reading too fast.
On a side note, if Spritz is that good, I'd like to try it out above my comfort zone. Say, 1000wpm.
That said, I don't see it as being a blocker exactly. A UI that let you pause the stream, and then if asked revealed some kind of tome navigation can't be too difficult to do. Also, images (photos, diagrams) could be revealed in a similar way.
That way the benefit of speed reading is retained, but with the added benefit of being able to skip backwards and forwards, and pausing to look at graphics.
This stuff always interests me because I'm a terribly slow reader (I think I did ~180 wpm on the above test). I'd really like to be able to improve without hurting my comprehension. However, when I discuss an article with someone who reads quickly I usually find that they've missed some important pieces of information. The sort of gaps exposed by real conversation may not show up on a simple multiple choice quiz. I'm just guessing here though. Based on my own experience, hearing that someone read 700wpm with 95% comprehension is like hearing that someone has broken the laws of physics.
Which, according to all the research I've seen, is exactly what you don't want to do if you want to retain any information. (I couldn't find the study I remembered reading recently, but this article references some such studies http://skeptoid.com/episodes/4229)
I was speed reading since Third Grade and remember making a deliberate effort to slow down in Junior High because I was not retaining enough information to write essays. During High School, I broke the curve on A.P. US History on an open notes final without having a single note to bring to class.
Auditory learning styles may require the subvocalization to succeed.
I found the exact opposite when playing with the demo on the site. Perhaps I fall into the camp of the 1k+ wpm readers, and by extension am a non-linear, non-vocalizing reader, but going this slow, and seeing only one word at a time made me conscious of my mind "saying" every word.
> it forces you to read every word
Also a pain in the ass - I found if you blink, you'll end up missing words. At a consistent 500 wpm clip, a blink (avg 100-400 ms) could miss anywhere from 1 to 3 words. As you get faster, blinks are even more costly. Somehow (I assume through years of practice), when speed reading, your mind is able to properly account for blinks. This also brings up an interesting point - the creators claim that this method reduces eyestrain, but it makes me feel like I'm forced to hold my eyes open for fear of missing something. I can't imagine that's any less straining on the eyes.
Maybe in the future they'll add eye tracking and pause the text when it detects a blink. And we thought Kindle tracking your reading habits was scary.
Over the past year, I have dedicated hundreds of hours towards improving my reading speed. I usually use Acceleread on my iPad; most recently I used it to read D.H Lawrence's Sons & Lovers at around 750-800 wpm. I also use Beeline -a different type of reading tool with some additional functionality for dyslexics- when reading web pages.
I've learned that certain types of prose make for poor speed reading material. Without a "rewind" option, there is no way that I can use these apps to read anything instructional or anything involving a high level of detail (such as a mystery novel). I understand that allowing the user to rewind would sacrifice speed, but speed can't always take precedence over clarity.
Also - in terms of the text- there needs to be a better method for handling quoted text. With speed reading apps that give the user the option of seeing chunks of words, it is far too easy to lose track of who is speaking when one chunk of words has an opening quotation mark and the next chunk of words [may or may not] have a closing quotation mark. I'm guessing that this problem is exacerbated by the fact that - in Spritz's case- words are only presented one at a time.
Here are some things I would I would like to see adopted by speed-readers:
Rewind functionality
Punctuation-based text color:
If the text is in between quotation marks, parenthesis or brackets, this text should be shaded a slightly different color than the rest.
Contextual deceleration:
I would love it if a speed reading app would automatically slow down at the end of a chapter. The last few sentences of a chapter are usually crafted with especially deliberate care and are meant to be savored.
More customization options for dyslexics:
For example, Beeline allows users to chose a specific type of font that is specifically geared towards dyslexics. In the future, I'd like to see speed readers enable these types of fonts (as well as customization options for line height and letter spacing).
For speed reading something like a novel, this would help the brain quickly associate a color with a character and avoid the need to catch the text that makes it clear who is speaking.
Imagine an UI with a circle and the spritz box above. You put your thumb in the middle, push right it starts to show the words forward, push farther from midle => faster. Push it left => slower. Left after middle, rewinding text.
You could even use up, down to go back entiry frases or paragraphs.
Regarding linearity, you could use color/color fading to wrap phrases, or show two to four words per time moving just the red letter according to speed.
Personally... I think it's an amazing Idea! :)
Fun to make. I'm also curious about the overall benefits. I find myself distracted by other thoughts if I don't have subvocalization happening.
If you want the ability to move backwards, I feel like that's possible with a good NLP library. If you skip back two sentences, or enough sentences to match some word or syllable threshold, it may be useful. In that case, you can play a few words, pause for a second or so to let the user decide if they've found the right spot (and context switch into reading) or if they should continue going back. I tried hacking in a quick back button that went back X words, but it was very disorienting. But in any case, it isn't as interaction free as doing so when reading a block of text is. But to counter your argument: you still have to keep your position even with text or seek for your position again, so there's a downside to those benefits.
I'm not sure how Spritz handles things like this, but I felt like making my own based on what I saw on the demo really quick to get a better understanding. It's a curious thing, maybe I'll check out the application sometime.
Thanks for the JS Fiddle. It was the basis for my chrome extension, which your fiddle greatly expedited.
The following Chrome extension allows a single-word-reader (or whatever they are called?) to pop up and run for any text you select in your browser.
https://github.com/keithmgould/speedreader-chrome-extension
I'm curious as to how they calculate the "Optimal Recognition Position", but there is probably a better way to approximate it. I'm going to play around with some different positions.
In the past few years I begun using text to speech coupled with visual reading with the same effect - it makes my concentration more powerful. I don't even like to just read any more - or God forbid - post a comment like this without listening it out aloud. TTS support during reading also has the advantage of allowing me to look away for a time, maybe do something in the room while continuing reading (or reading while I drive).
My favourite voice is Alex from Mac OS. I read dozens of books, thousands of articles and forum threads like this. With forums I still need to scrape out some formatting text with a quick Grease Monkey script though, but it's so fun to hear you out aloud!
And yeah, I find the name very unfortunate as well, especially in combination with call to actions like "Click here to Spritz".
Don’t spritz me you spritzin’ spritzer! (wtf?)
Will spritzing burn my retinas? (only if he spritzes in your eye...)
What is fine? What is good enough? I think it's an absurd statement that in no way is an argument against reading at 500 wpm.
>Most of the gen-pop is around 200-250 wpm which is why I >said it is good enough. Don't be a troll.
At around 200-250 wpm you are in the middle of the normal distribution hence good enough.
I tried multiple times to read the example, but would stumble to comprehend the shape of a word or two, lose my way, and have to rewind back to the beginning.
Overall it felt considerably slower and more broken, less rhythmic. My comprehension felt lower, I felt I spent far more time comprehending the shapes of the words rather than the totality of the meaning expressed.
I know my eyesight is bad, that I am long-sighted and colour blind, but this was a painful and slow reading experience.
Some words took so much concentration to figure out that I'd forgotten the words that preceded it, and lost context.
Without the ability to rewind this or to see a larger context, I would avoid this like the plague.
Clearly an outlier as everyone else is raving.