Zethos – Speed reading tool in 1 kb

http://www.zolmeister.com/2014/03/zethos-35-million-in-1kb.html

by kumarski • 12 years ago

105 28 12 years ago

zolmeister.com

robertkrahn01• 12 years ago

Another Spritz-like implementation: http://lively-web.org/users/robertkrahn/speedread.html

garretraziel• 12 years ago

As a Smalltalk enthusiast, I am really glad to see lively-web in actual use.

_asciiker_• 12 years ago

Hello,

It looks looks, however Spritz went a little further, the sweet spot of the human eye is a little to the left of the center, that is where the "red" character should be. On your example you see the larger words do jump around a bit and it is not as efficient.

gkoberger2• 12 years ago

I love you made something and open sourced it, however your multiple mentions of $3.5 million ignores a lot of what goes into software development:

- Most of Spritz's work went into the R&D, not the actual code. Anyone can copy the end result.

- Zethos ignores a lot of the science that makes Spritz work so well.

- Spritz does localization, not just best-guess-English.

- Usability (as in, how do I even run Zethos on, say, a book?) is important.

- Spritz has an SDK for Android, iOS, JavaScript and more. They support Google Glass, watches, etc. Ubiquity is important.

- Marketing, bizdev, support, design, QA, legal, HR.

- Money raised is a horrible metric for code. Even revenue wouldn't work here.

Think about it this way: a ton of people cloned Flappy Birds in HTML when it came out, but none of them are worth anything. I could do a decent rendition of Facebook in a few hours, but that doesn't make me Mark Zuckerberg.

rjzzleep• 12 years ago

- Most of Spritz's work went into the R&D, not the actual code. Anyone can copy the end result.

conveniently ignoring the fact that most of it was not actually spritz' research. the only "extension" if you so will is that colored cursor. acereader[1] for example has been around for 10 years(or almost 10 years). go to google scholar, and search for saccadic eye movement. for example here is a paper from 1992! [2]. here's another one from 2006 [3]

Rapid serial visual presentation (RSVP) of text has been reported to foster higher reading rates than presentation in a continuous text (CT) format, possibly because scanning eye movements are minimized. We investigated how this might be relevant for persons with congenital nystagmus (CN).

the fact that spritz may even get credited with this is the very problem of our buzz society. i refuse to believe that creating a crossplatform sdk or "researching" the fact that using a fixed cursor to reduce said saccadic eye movement even further requires 3 million usd. don't get me wrong i'm a little happy that they created enough buzz for this to become popular. but if we really need that much money to do basic tasks, god help us.

- Zethos ignores a lot of the science that makes Spritz work so well.

:D like what?

- Usability (as in, how do I even run Zethos on, say, a book?) is important.

how do you run spritz? spritz is an SDK(!). it's not an ebook reader, it's not a website reader it's an sdk. it's up to you to integrate it in your app.

EDIT: this was written before parent appended SDK information to his post

- Money raised is a horrible metric for code. Even revenue wouldn't work here.
Think about it this way: a ton of people cloned Flappy Birds in HTML when it came out, but none of them are worth anything. I could do a decent rendition of Facebook in a few hours, but that doesn't make me Mark Zuckerberg.

your analogy is completely faulty. first of all mark zuckerberg didn't "make" facebook contrary to popular belief. second, flappy bird itself was already a clone that got popular, similar to spritz. third, no, you can't just roll your own facebook in a few hours. you can create something that looks like it, being far from functionally the same. spritz is different. create a spritz clone though? see below:

https://github.com/OnlyInAmerica/OpenSpritz-Android

the difference? it's not a library. meaning noone can really reuse the code provided in here. i doubt spritz is going to be that easy to integrate with academic pdf's. though i have a feeling that that's what they might be using their couple million funding on. i have a todo list that's at least 5 years old with a speed reader for pdf's in the list. but is it worth investing time in? sure, if you give me 3 million(i'm not actually being fully serious here, so spare me logical inconsistency analysis of my post)

[1] http://www.acereader.com/

[2] http://www.citeulike.org/group/6810/article/3309870

[3] http://www.ncbi.nlm.nih.gov/pubmed/16909075

EDIT: there have already been a few rsvp browser extensions for both firefox, and chrome over the last couple of years, but there never was any buzz around any of them.

http://www.ilovefreesoftware.com/11/windows/internet/reasy-r...

http://youtu.be/MuszcoVL3lA

edraferi• 12 years ago

The major missing feature appears to be intelligent hyphenation. Long words should be broken up by syllable, not raw character count. it's missing the little hash mark at the focus letter as well, but that's an easy fix. intelligent hyphenation likely requires some real NLP work, or at least importing a real dictionary of some kind

3rd3• 12 years ago

You don’t acutally know what Spritz’ research exactly looked like. Maybe they tested thouthands of configurations (custom fonts, hypthenation, color coding of clauses etc.) to find that the optimal one is still rather simple (single words, monospaced font with one red highlighted, centered letter).

scrrr• 12 years ago

I want to see a decent rendition of Facebook that was made in just a few hours. ;)

72deluxe• 12 years ago

Google+? Sure seems like they only spent a few hours on it anyway haha

Sorry, couldn't resist.

kranner• 12 years ago

I don't have a study to back it up, but in my experience speed-reading multiple words at a time is much better for comprehension compared to the Spritz way of a single word. You can afford to slow down (which helps around blinks) and still have a high effective WPM rate.

I wrote http://velocireaderapp.com specifically to speed-read multiple words better than existing apps (splitting sentences around punctuation the way I prefer). From the emails I get, it seems most users prefer to read multiple words, with 2-3 being the most common length. I should track this.

camillomiller• 12 years ago

Nice proof of concept (the concept being that sometimes the whole funding thing is a bit difficult to properly understand). Now, a call to programmers: why the heck there's still no readability-like chrome extension? After I saw the spritz demo I thought that it would have been immediately consequential...

gkoberger2• 12 years ago

There's this (a bookmarklet):

https://gun.io/blog/openspritz-a-free-speed-reading-bookmark...

However, it no longer works due to him going over the Readability API limit.

davidjhamp3• 12 years ago

I made one too that has key bindings as well: http://www.davidhampgonsalves.com/spritz-like-rsvp-reader-bo...

mitchtbaum• 12 years ago

https://chrome.google.com/webstore/detail/sprint-reader-spee...

zk000063• 12 years ago

Is it just me, or other people are also not able to get the content of the text while testing these method? Reading is highly non-linear and it is necessary to spend more time on some parts of the sentence than the others. Streaming the words linearly just does not work for me at all.

rplnt• 12 years ago

This was the major issue I've had some other tool I tried some time back. It spent the same time on, for example, "inconvenience" and "cat". It also didn't really worked with prepositions. Pauses on commas and periods were weird as well. That's just something you could work on. Then there's contextual complexity (that you probably talk about), where sentence could be complicated even if it contains only simple words.

ozh• 12 years ago

These tools are not to read. They're to quickly scan texts and get the general meaning. You would never, never want to read a novel using this.

zk000063• 12 years ago

From my experience, I get the general meaning also by focusing on first and last sentence of a paragraph and skipping everything else. Also, thinking for a while about individual words gives a lot of information about the whole content without the need to go through every word one at the time.

seren• 12 years ago

If I want to quickly scan an article, I start by a few opening sentences, then go directly to the conclusion, and if it is not enough I will quickly scan the opening of some paragraphs to look for the most compelling arguments/experiments. I don't see how using something like Spritz I could be more efficient. The point is that you don't need to read every single word, even very quickly, to grasp the author's ideas. Maybe I am missing something ?

superasn• 12 years ago

For those who haven't had great success with Zethos/Spritz, can give magicscroll.net a try. Another reading tool but very different technology. I've found it quite easy to read books with it for some reason (unfortunately it does not work with PDF or general text content.)

Like try this (click the play button at bottom right to start):

http://www.magicscroll.net/#reader;md5=597e52af574ea413a5886...

robobro• 12 years ago

If someone were to make a Spritz-esque bookmarklet that added more to the basic idea we've seen online, what would you consider to be useful additions? I think that letting users select text and do a run of the highlighted area would be great, or letting them click "pause" to jump the page down to the currently displayed word would be cool too.

davidjhamp3• 12 years ago

I made one with those features(except that pause doesn't jump to that word)

http://www.davidhampgonsalves.com/spritz-like-rsvp-reader-bo...

pkhamre• 12 years ago

How do I install and use it?

zk000063• 12 years ago

I believe these speed-reading tools are missing the concept of meaning. While the idea to fix the eyes allows to deliver more data to the retina, the linear speed does not work well. It would be great to combine these methods with some text processing algorithms that are able to extract the key words from a sentence. While I am not expert in text processing I am convinced that methods able to compress text into meaningful keywords must have been proposed in the literature already. If not, it would be very interesting research. Those keywords should be then presented longer to the reader, while the less important words should be skipped.

luisivan• 12 years ago

I made a library that does the same a week ago: https://github.com/luisivan/spritzer

gren• 12 years ago

Nice concept, It would be wicked if this could work with something http://store.neurosky.com/products/mindwave-1 (an attention detector basically) and speed up when your attention raise! Also the thing can detect your eye blink, and you basically want to stop a bit the reader when this occurs ;-) I assume google glass already have thought about that.

Future here we go!

davidjhamp3• 12 years ago

I wrote one of these things(bookmarklet) but I never really use it. Reading in this fashion while faster just isn't that enjoyable in my experience.

kriro• 12 years ago

I'd like to use this for PDF-papers. Rough benchmark that would be cool: Read an entire conference paper (usually in the 6-8 pages range, bibliography can probably be excluded for starters) in the time it currently takes me to read the abstract :D

[if you build this you can probably sell it to one of the reference management software vendors]

Zethos – Speed reading tool in 1 kb

Related Stories