This does indeed seem to work for me, but the repeated emphasis on the patents and trademarks makes me suspicious. It looks like a company who cares much more about patent royalties than about their invention - whom I would never be willing to give my money to. Having seen the principle, I could write a browser extension in an hour. I suppose I'll have to wait 25 years (or whatever the French patent expiry length is), though.
If you want to read the patent (in French), it's here: https://data.inpi.fr/brevets/FR3052587. I couldn't find the German patent (10 2017 112 916.2), but I'm guessing it's the same substance.
Yes, I agree. I guess this technology isn't for everyone. This is a real bummer. They did not have to patent and trademark this method, which in principle is such a basic mechanism. Just had to productize their service really well so they'd take the lead in monetizing.
I am alone and I put all my money, time and work into BR. Visible is only a small part. But I think everyone decides for themselves what is best.
But thank you for the critical feedback.
Plus without a patent it would be very easy to capitalize on this without compensating the guy for all the research and tuning his put into it. I think patents are ideal for new ideas that are super easy to steal/copy, right? Sure Academics working at the bleeding edge should patent for example but they're likely to have a little advantage by being the expert and actually knowing how to do the thing. But in this case the result may look simple but took ingenuity and time to produce just the same.
The only way to use a patent “for good” is to prevent someone else from patenting it and exploiting it.
If you’re serious, then release a copy-left license for the patent (but not necessarily your implementation/product/sdk)
Otherwise, just add your commercial licensing terms to your website already so people aren’t misled into thinking they’ll be able to do anything with this idea unless they pay you.
The rules are more clearly depicted on page 4, with different variations depicted on the pages following that.
Looks fairly straightforward to implement. Until this is more widely available "officially", I'll probably just hack up some Chrome extension for personal use and try it out.
> I suppose I'll have to wait 25 years (or whatever the French patent expiry length is), though.
Is that how patents work in general?
I always thought (and could be very wrong here) that if a patent exists for something but you've developed an alternative something in total isolation without seeing the patent then you would have free reign on how to design and develop your thing without violating the patent.
The opposite is true, in fact - a patent is a monopoly on the patented technology, no matter where alternative implementations come from. You might be thinking of “clean room” techniques which protect against claims of copyright infringement.
> You might be thinking of “clean room” techniques which protect against claims of copyright infringement.
Ah yes, that's what I was thinking of -- thanks. Specifically a scene from an episode of Halt and Catch Fire (I won't describe it due to spoilers).
That's a real bummer for patents though and makes me wonder how something like Amazon's 1-click checkout was able to be patented. I wonder how "save billing details for future use" can be considered a novel idea.
For a long time, software patenting law was famously lax. You could basically patent anything software-related if you knew the right legalese. The situation has apparently gotten better after a 2014 SCOTUS ruling, but in the '90s and '00s it was pretty dire.
I believe those scenes were an amalgamation of Tim Paterson’s clean room implementation of DOS from the CP/M manuals [0] and Compaq’s IBM compatible BIOS implementation[1].
They are a some of my favorites, and are fun to watch.
> Standards organizations, therefore, often require members disclose and grant licenses to their patents and pending patent applications that cover a standard that the organization is developing.
> https://en.m.wikipedia.org/wiki/Essential_patent
Please not that this is less of a "You have to offer licenses or else" and more of a "Hey please do that."
I have some feedback regarding the testimonial from “Sangeeth”, whose Memoji shows a Sikh person. Did the user choose their Memoji? The two don’t go together well for reasons of odd cultural differences in the Roman/Latin transliteration of sounds in the south and the north of India. The -th- cluster represents the dental t, IPA /t/, but only in South India. In the North the same sound is represented by -t-. Therefore you might find a Sikh named Sangeet, or a non-Sikh named Sangeeth, but a Sikh named Sangeeth? I’ve never seen a Sikh name using the -th- cluster in my life. There are Sikhs who grow up in South India of course but their names use the North Indian transliteration rules.
What I’m trying to say is: if you chose the Memoji with a turban to show an Indian person, it doesn’t match the -th in the name. As an Indian my first reaction was that the testimonial was fake, for this reason.
Hi kranner
I didn't mean to offend you. Or misrepresent the culture with a memoji. Sorry. If you feel it's out of order, then I can change it too.
But what is more crucial for me is what Sangeeth said. And that is more important to me.
By the way, all testimonials are real. 100%
> Hi kranner I didn't mean to offend you. Or misrepresent the culture with a memoji. Sorry. If you feel it's out of order, then I can change it too. But what is more crucial for me is what Sangeeth said. And that is more important to me. By the way, all testimonials are real. 100%
I think that the general idea is that either you're guessing at the appropriate representation of the testifying users, or that they've picked those emoji. If the latter, then all is good! If the former, then it's generally a bad idea to assume that you know something about a person based on their name, or even, really, anything other than their implicit election. (I would certainly be upset if a business to whom I had offered a testimonial used it and misidentified me, even in ways that may seem trivial, especially if they could have just asked me.)
> In addition to pronouns, let's include emojis that we feel identify ourselves
I assume that was sarcastic, but why not? Do you have any right to choose the emoji to identify me? I mean, you surely have every right to editorialise with an emoji that you think reflects my behaviour, but, if you're going to quote me below a symbolic representation, then I think it's different. If one has to use an emoji to identify someone—and it's not at all clear to me that it's necessary—then, after all, we're talking about who has reached out; why not just ask them?
(For example, surely—right?—we can agree that it would be bad to say "a user from India gave this testimonial:", or whatever—unless the place of origin is relevant, and you actually know it. I don't see much difference between these two.)
It certainly seems clear that renato_cassutt, who it is clear can taken constructive criticism and who has given very useful replies throughout the comments, meant no-one any harm; and neither I nor, I am sure, kranner meant to impute any ill intent. I can understand the appeal of pictorial representations of one's users; I was just, and I suspect that kranner was also, trying to mention the possible inadvertent offence that might result in this way, not to ruin anyone's fun but to save someone from an inadvertent misstep.
Thanks JadeNB. I'm pretty sure that renato_cassutt didn't mean any harm and that the testimonial is from a real user. The context that many are missing here is that a turban is religious headwear like a yarmulke. As a Sikh you have to be authorised to wear one in a ceremony called a "dastaarbandi", and there are rules of conduct while wearing one in public. It's inadvisable to add one to the depiction of a person unless they choose it themselves. This kind of thing (e.g. inappropriate Sikh representation in movies) regularly becomes a political issue in India. In renato_cassutt's case, it's an unnecessary distraction from the actual product.
I do very much agree that it's inadvisable to add a turban to depiction of person unless they choose themselves.
That being said, as an pedantic, unshorn hair, turban wearing Sikh I'm not sure I'd describe dastaarbandi as any sort of ceremony granting authorization.
Hi renato_casutt, I'm not offended at all and congratulations on launching. I only commented to point out that the turban+name combination makes the comment look inauthentic which may be an issue in the intended use of the comment as social proof.
It’s probably a review written by somebody with a brown skin tone. Page authors don’t know the social intricacies of what constitutes Sikhs and chose this one as “close enough”.
Maybe, but Sangeeth is not here to tell us if they are Sikh or not, so it's hard to say.
I think it's unfortunate that (at the time of writing) the comments about the product and its effectiveness are ranked lower than this discussion about the _possibility_ that a non-Sikh person has been represented as Sikh.
If Sangeeth themselves had popped up in this thread to comment on the choice of emoji, then sure, but it's just somebody making some assumptions based on a 't' vs 'th' cluster in the name.
Sangeeth is not a Sikh name. In India people don't convert to Sikhism, or at least I'm a middle-aged Indian and I've lived here all my life, I've never seen it happen. There's a certain foreign subculture of Americans converting to Sikhism, but then they take on Sikh names. It's extremely improbable that Sangeeth is a Sikh person.
Oh I agree, the lesson learned should be: let's not spend our attention on race and identity because it leads to these bizarre situations among many other things. I barely understand how we ended up like this.
Thank you. I was trying to keep it understated, but it was the equivalent of seeing a Jewish-sounding name and depicting the person as a yarmulke-wearing Orthodox Jew without their consent. Anyone with a broadly shared cultural background as that user would get hung up on the incongruence (and many would mind being stereotyped) instead of focusing on your product. I don't blame you for not knowing that turbans are religious headwear, not fun hats people wear and take off on a whim as depicted in American media (which has greatly influenced global cultural stereotypes).
You are absolutely right. I really wasn't aware of it. And therefore I thank you very much for the hint.
I am an open person and am therefore always happy when I can learn something new. Everyone should live as he wants.
We are individuals and I think that's great.
Best regards, Renato
Not sure if this intentional or just because this site is unfinished, but the repetition of the same paragraph throughout the page probably has the effect of making you believe you've read it faster, but it's more likely just recent familiarity of those words in that order that is allowing your brain to skip ahead rather than a specific visual cue.
I actually found the 'bumpiness' in reading a distraction and I had to read the paragraph a few times to truly get the meaning. So any speed gains from the reading methodology were lost during the understanding phase
> Bionic reading aims to encourage a more in-depth reading and understanding of written content
by reading faster? I mean the examples are pretty compelling and it did feel like I was reading faster, but at the same time their goals are at odds - faster reading doesn't necessarily mean more attentive reading, or being able to absorb more information.
Still, interesting technique. I just wish they made their examples / comparisons actual text instead of images; surely they can just write a snippet of JS or some HTML to render the effect in the browser itself. Or would that be giving up their secret sauce?
Also if any of the authors are here, please get rid of the "small text in between big text", I'm not sure if I need to read the one and then the other, or refocus my eyes on the big then the small text.
I think the less friction there is between the words on a screen and you having them in your head, the more readily you're going to absorb it. And attention is a demanding resource to expend, even if non-ADHD people don't notice it due to having ample supply.
But yes, this would be awesome to see as a browser extension. Maybe I'll get around to it at some distant point in the future.
It was explained in the link title before the change - their hypothesis is that our eyes are the bottleneck - the brain can understand more quickly than we can read. So by speeding up our eyes, we can understand more information.
I saw it as "reading with less effort" and "faster" is kind of a shorthand for "less likely to get bored waiting for my eyes to catch up".
I'd be more likely to read a longer piece of text (average content, not something I'm deeply interested in) if it was "easier" to read -- I get through it faster, or have to make less effort to get to the end.
As another comment said, this would be good for those with short attention spans or those with reading disabilities like dislexia to focus better. I would want to see an epub renderer so I can read ebooks with this method.
I'm on mobile so maybe that is why, but I noticed that I skipped over the sentence a bit. As in, I didn't notice the periods and comma at first. Would bolding the punctuation marks cause any issues?
Actually this made it harder for me to read. The text felt "bumpy", unlike normal in text where it felt smoother and flowing. Maybe it's just a matter of habit.
Anyways, I don't believe speed-reading to be effective, especially for comprehension and long-term retention. The best and fastest way to get an overview of a written piece is not to read it faster, but to skip most of the content. Most of it is just supporting the core arguments, and can thus be skipped for an overview.
I'd say it almost halves my reading speed. In regards to speed reading I've always been under the impression it means you take in entire lines/sentences in one go rather than reading faster, skipping the sub-vocalisation step. For instance the homepage of HN is a quick scan vertically down the page until a headline catches my eye as interesting, no real side to side eye movement needed unless the title is a bit long
Maybe this brain reading faster than eyes thing is best aimed at sub-vocalisers? I'd be interested in knowing if the people this helps read 'out loud' in their heads
Yes, corobo
that can be that it prevents you from reading. However, I would argue that you are a speed reader.
One test person in the study was also a speed reader. She had read two different texts. Once with and once without bionic reading. But her speed stayed at 800 wpm each time.
But believe me, there are a lot of people who have difficulty with reading. Surely also in your circle of friends. And for some of them Bionic Reading can help.
I felt the same way. It slows me down. But I imagine it would be very useful for dyslexic people. I have a family member who struggles to read; it's as if they have to consciously process what each character means. From that perspective, it's not so much speed reading, but bringing up people with slow reading to normal levels.
Could you make one that reduces the opacity of the second half of the word, while keeping the same boldness for the complete text, to compare the effect?
I am very glad that Bionic Reading is being discussed in HN.
I am aware that it cannot help everyone.
But I also get a lot of feedback from people who are helped a lot by this reading mode.
Imagine you have problems with reading. You don't realize what a huge hurdle that is until you talk to people who are affected.
Being ashamed in front of society.
The problem of learning new things.
Being called stupid.
And these are just a few examples of people affected.
Therefore I thank you very much that you discuss the problem critically.
Best regards from the Alps,
Renato
I saw the opacity features but I was curious to see if with that feature, the bold text could stay as-is, and the regular text could take on lower opacity so that the full word is still there but only the bold parts are in full color. The opacity demo showed the whole paragraph as taking on an opacity reduction. I’m wondering if making the regular letters less opaque would disrupt the flow or further help focus on the important bits in bold?
I spotted a couple of language issues on the webpage. "most concise parts of words" doesn't make sense, as "concise" means brief or short. Maybe you meant "important"? "very great" isn't proper English.
This would be an awesome enhancement to Pocket- which already extracts web page texts into a lighter format - I'd probably read everything through that
Are there any actual studies on the gained speed being done by Independent scientists? I could not finde anything on their (quite beautifully done) marketing/landing page.
I see patent registrations, but no mention of any studies underpinning the claims being made.
To me this smells like all the other pseudo-scientific BS marketing and product people are trying to spoon-feed me every day.
Also as said in a sibling comment, for me it killed my reading speed. From their examples I would estimate by about half.
So thanks, but no thanks - this is not an offering for me. But it will probably find its audience. I could imagine it will be a big hit in the self optimization scene.
We have just completed the preliminary study but I cannot publish any results yet.
And you are right, of course. It does not help everyone. That's the way it is. But there is a lot of feedback thanking us and hoping that it will be available to as many services as possible.
I don't doubt that this will help people. I know how it helped me to train myself in the way I read today during my university years.
So as said - there will be people who benefit from this. Still - being an empiricist I just like to see independent and scientifically valid representative studies being done before I believe marketing claims. ;-)
I don't have links to studies but last time I checked and from anecdotal experience reading speed with high comprehension is limited by information throughput in the brain. It's also why languages spoken faster carry less information per word, because the limit is information processing. To experience that if you are a good reader, you can try to speak out loud a sentence while memorizing a second sentence and see that you'll not have process and understand the words until you have spoken them, even if you have read and stored them (short sentences like in TFA, obviously doesn't work on HN).
What! This blew me. This magic happening before my eyes. Forget the negative comments here on HN, the fact that you had such an insight, it's awesome.
Going constructive on the HN comments, if someone found something, you got a right to protect and earn from it anyway, charging for however a simple idea helps it spread and let's you do more innovation on it. But that said, maybe a library with paid access is a better idea than api which needs to travel across 14 seas and 7 oceans!
It’s also obnoxious and utterly frustrating as an end-user. (I know that’s not the aspect of it you’re remarking on but rather the monetisation, but I find their demo just so painful and the entire idea quite ill-conceived. Scrolljacking is always bad. Stop trying to take over the page flow and just make it a normal page that I can scroll as I desire.)
I think the idea is great but the greediness on display here kind of disgusts me.
Here is something that could legitimately be useful for many people, help the consumption of human knowledge when open-sourced, but instead it is patented and monetized to the maximum extent.
I wonder if "word" is the correct unit, here; perhaps "morpheme" would work better, especially for multi-morphemic words or in compound nouns (e.g., in German). I think "syllable" may be a bit too much, unless the purpose was to teach reading. For example:
ANTIDISESTAblishmentarianism (initial 40% highlighted)
vs.
ANtiDisESTAblishMEntARianIsm (highlighting rules over morphemes)
vs.
AnTiDisEstAbLIshMEntArIAnIsm (highlighting rules over syllables)
An API that provided this as a service would be less trivial.
The lay-typographer in me really hates the look of it too. Aesthetics don't trump readability, but perhaps there's a middle ground where the font weight drops off (e.g., with exponential decay), rather than having the binary "bold vs. not bold".
For everyone who think this is new stuff: It is not new. There was another company in the German speaking space, Vienna more exactly, see craftworks.at, that offered almost the exact same product already back in 2019, perhaps even earlier, also called "Bionic Reading".
In the current version of the website this does not appear, but it is saved in archive.org: https://web.archive.org/web/20190119032121/http://br.craftwo...
Perhaps the author of the current offering worked at craftworks at that time, then left and took his IP with him, I don't know.
The thing is, this (and craftworks.at or whatever) is still not a fully polished technology.
It probably can be optimized in a bunch of ways:
* Less drastic transitions between accentuated text and the rest
* An ML algorithm to select quantifiably better patterns (with distillation if necessary)
* Enrolling visually- and cognitively-impaired persons (e.g.dyslexics) into the study to avoid undercutting accessibility
We've been doing printing and written/illuminated books for many centuries, and statistical readability studies for the last 80 years or so. If a variation of this technique is better than the existing 'flat' type (even with ligatures), NASA/DARPA/NIH would be investigating it shortly.
BeeLine founder here; interested to see this. My first reaction was that the "before" text in the before/after demonstration is both thin and grey.
It is well known that text that is too thin and too light-colored is not great to read. The text modification shown here makes certain letters easier to read, but it appears this is partially because the rest of the text is specifically difficult to read. The background of the page very light grey, it looks like, which would magnify this effect even more.
I wonder what this would look like with a not-too-thin-font that is black or near-black. I also wonder what it looks like on a full-width paragraph, as the sibling commenter mentioned.
Regardless, it's always nice to see innovation in this space!
Since you work in this field, I'd be interested if you would really qualify this as innovation. I have posted already in several places in this thread about the fact that Bionic Reader is at least 3 years old already. Someone else also dug up the patent application: The rules to make the text bold in Bionic Reader in its simplest form (i.e. ignoring things like setting Saccade manually) apparently is just a very very simple if-else that takes 3 lines of code.
I wonder how messy it would look to combine the two? BeeLine really works, especially in cases where the designer has disregarded the need for reasonable column widths. I feel this works too, but as with BeeLine the visual appearance is a bit jarring at first.
Hi mdp2021
I understand your skepticism. The point is that the reader needs a fixed point to absorb the text. That's why this definition is placed.
As shared in other comments, I unfortunately had to make very bad experiences. And the responsibility I perceive among others Marco (feedback website), I do not want to jeopardize.
But I understand your objections.
How do you decide which letters to highlight? Is it the first half in every case? Did you conduct any trials to see whether this yields the fastest reading speeds?
As a non native English speaker, this does not work.
Maybe because the part of the brain used to read/write/listen/speak a non native language is not the same as a native language. I really feel that I need to focus more while reading the example paragraph with the bionic stuff.
I tried the other examples, but I memorized enough of the text starting the 3rd example that I could not really infer anything. Putting the same text for all the examples is not really helpful in this case.
As a native English speaker, I also felt like I had to work harder to read the bionic text.
I am first grade teacher and have taught many children to read. Bionic's principles and algorithm do not match what I understand about the science of reading. This just holds the first part of a word, regardless of what the letters are. I would want an algorithm that is smarter than that (and I certainly wouldn't use bionic in the classroom, regardless!).
Hi cardamomo
I understand your skepticism and thank you for your criticism. What you see so far from Bionic Reading is the base. Of course, children should still learn to read. Because only when children have acquired a vocabulary, their brain has the representation products.
Reading is already very well researched. I absolutely agree with you on that. Unfortunately, people are reading worse and worse. There may be many different reasons for this. But I'm sure you agree with me that reading is a cultural asset that everyone must use.
Who reads worse, has thereby no advantage. Doesn't it?
Reading transports knowledge. That's why I think it's important to take a closer look at the way we read.
That's why I get a lot of feedback from people who have dyslexia and can read better again with Bionic Reading.
Thank you for teaching the children and best regards from the Alps,
Renato
Yes, that's just me, Renato Casutt. From a small town in Switzerland.
I discovered BR in my studies as a typographic designer.
And I think it's time to show that typography is not just well-designed text pages. There are also ways to make reading easier for other people with reading difficulties (e.g. dyslexia sufferers).
Yes, you are right.
The problem with Spritz is that you lose the overview of the text and therefore you can't optimally absorb the context of the content.
Some users have already confirmed this to me.
But the technique is cool...
There are many good RSVP apps and browser plugins. I find it better for news or other things I want to skim read, than trying to power through a novel.
Any preferences for plugins and apps these days? I used to use RSVP tech a decade ago in the Windows phone days, but have not seen great modern clients.
There's also spreeder on Android in similar spirit which I found useful to read web content. The downside is I have to copy and paste it from the browser
Site images blocked for me, and no examples in HTML.
Looks like “shortpixel.ai” blocked by AdGuard Base.
“Data on Swiss servers”, but shortpixel.com and shortpixel.ai owned by ID Scout of Romania. Their Wordpress plugin and CDN both superficially ‘act like’ plenty of free beacon tracker type image hosts, though this service offers paid plans and copy paste data legalese doesn’t reference advertising/measurement/aggregation third parties.
I haven’t the time to look into why AdGuard is blocking, but perhaps it’s that ad banner or ad injector companies like it:
Surprisingly pleasant. I haven't speed read in a while, but this felt better than highlighting text, or centering word on screen (although those work better with audio support, but I guess that could be mixed together).
Few questions:
From what I understand BR figures out 'what part of the word is unique enough to be a good enough approximation of the word'. Is this somewhat context aware or generic (with parameters available for developer)?
Are there any specific, measurable claims (and evidence) being made here with regards to what this supposed to achieve? "[Encouraging] a more in depth reading and understanding of written content" is pretty vague. Does it help with recall? Memorization etc. Personally it reads like those "dyslexic" fonts which actually make reading harder for a lot of people.
My brain just automatically ignores the paragraph with the bold letters. I'm having to force myself to focus on reading it. Does anyone else experience the same?
I like how this leverages the fact we don't have to read the entire word, or that letters in the middle of the word can be jumbled and we still read it correctly.
Very cool insight in emphasizing the "important" parts of the words. For me at least, it seems to have worked - it felt like less effort to glide over the paragraph of text and take in the content.
As far as it becoming a product.. I'm sorry, but people will immediately rip this off. Someone will make a free browser extension, or add this to an existing one like Outline / Mercury Reader / etc. To enforce patents you have to have money to pay the lawyers, so for small fry they're kind of useless.
Very cool idea all the same, and I would love to see this embedded in anything that displays text.
Great work, congrats! Another app Beeline Reader[0] seems a lot similar to this. It was launched many years ago I suppose. All the best with the product!
BeeLine founder here — I'd say we are similar in that we both work to improve visual tracking and speed up the reading process. BeeLine's focus is perhaps more on return sweeps, but in the grand scheme of things, we're both pushing in the same direction!
BeeLine improves visual tracking by displaying text with a color gradient, which wraps from the end of one line to the beginning of the next. Like Bionic Reader, it improves visual tracking across the line.
It also improves visual tracking during the 'return sweep' — when you move your eyes from right to left, to find the new line.
The colors are user-configurable, and we make first-party tools like browser plugins, [1] PDF converters, [2] and iOS apps. [3] We also partner with reading and education platforms like Blackboard [4], who have embedded our tech in their products.
We do have a few other tricks up our sleeve, but for now this is the main event. It's not too shabby, and has won social impact awards from MIT Solve [5] and the United Nations Solutions Summit. [6]
For me at least, it made little difference. My brain refused to ignore the parts that aren't boldened.
I think what I don't understand is the concept: "and so just a few letters are enough to recognize whole word".
This isn't expressed in the example. The rest of the word isn't hidden, it's still there in my peripheral vision (?). It is basically the same as I normally read.
If the unboldened parts of the words were actually discarded, I am not sure if I would be able to recognize all the original words, at least not efficiently and faster than the original.
You never look at every single character while reading, your eyes only dart between select tokens. Your peripheral vision and your brain fills in the blanks. I think the idea is to aid your brain quickly parsing these tokens.
A nice side effect is that it also helps your vertical tracking of lines by adding more visual anchor points.
A small observation: all the "normal" sample text has low contrast with the page background and the kerning seems awfully narrow making it harder, for me at least, to read than normal text. I tried to inspect the text to play with the css but alas, the text is an image and doesn't allow experimentation. I did run it though Chrome's Lighthouse accessibility check and it agreed about the lack of contrast.
I’ve used both Bionic Reading while on Reeder, and the Open Dyslexic font while on Kindle, for several years to similar effect.
My personal finding is that it makes the work my eyes need to do to keep on the line, and move from line to line easier.
But I will note that the largest increase in reading speed for me, up to 800 words per minute, came from single word focusing. There are tools, such as speed reader mode in Emacs, or an old browser extension called Squid, that take the readable text and put it all on one line, then move the text through a window that stays fixed on the screen.
By training yourself to relax and not read the words aloud, I was able to increase my reading speed from around 200 wpm to over 800 wpm. The trick is to slowly ramp up the speed, and not just jump up to the fastest speed possible. It’s also important to remember to blink.
Alternatively, text to speech works for to in a similar way, and I’ve found that I can handle 2.5x - 3.5x speeds after a few hours of training. As a bonus, nobody else can decider what you’re listening to, and you get some funny looks.
Let me know if others have found similar techniques to work for them.
Out of curiosity, how do dyslexic friendly fonts work? Are they easier for everyone, or is it a case of compensating dyslexia related deficits specifically?
There are a couple popular dyslexic-friendly fonts, OpenDyslexic [1] and Dyslexie. [2] My understanding is that the first is free and the second is not.
What they have in common is that both are quite heavy fonts (thick lines), and with a bit of extra kerning (space between letters). They are also designed to have 'b' and 'd' (and other mirror-image pairs) that do not look like mirror images.
I have worked in this field for years, as the founder of BeeLine Reader. [3] We received a lot of requests to offer one of these fonts, and we've used OpenDyslexic for quite some time. I know some people really like it. At the same time, dyslexia researchers (especially those in the US) insist that dyslexia is not a visual condition, and that fonts cannot make a difference.
I tend to believe that even if it is not an issue that is primarily visual, that doesn't meant that adjustments to how text is presented can't make things better. This seems to be the sense among (1) researchers outside the US, (2) individuals with dyslexia worldwide, and (3) many assistive technology or SPED workers in the US — despite what US-based researchers say.
You could check out Helperbird, which offers various readability tweaks. [1]
Glad you like BeeLine — we've integrated with several HN/YC'ers, ranging from blogs [2] to mobile apps. [3] Please feel free to reach out (developer@beelinereader.com) if you're interested in working together.
This presumes you're scanning lines but if you really want to speed read you should scan whole blocks of text as you draw the eye down, there should be little to no horizontal movement when you're really reading fast.
- - - -
An interesting and effective method of speed reading is Rapid Serial Visual Presentation (RSVP). These days it's mostly used in the context of psychological/perception studies, but it came about as a I/O method for military machines IIRC. Anyway, the trick is you display text one word at a time, in the same location, so the eye doesn't have to move at all. With a little practice and some tweaking (i.e. slight extra delay after periods and commas, that sort of thing) you can read very very fast. (For me I could read faster than my internal mental voice can speak.)
- - - -
There's an interesting effect where you can take some text and, for each word longer than ~5 letters, you leave the first and last letter in place but scramble all the "interior" letters. The resulting text is still legible!
(I'd give an example here but I'm too lazy. Exercise for the reader, eh?)
> There's an interesting effect where you can take some text and, for each word longer than ~5 letters [...]
I think you're referring to this (dubbed "Typoglycemia"):
> Aoccdrnig to rscheearch at Cmabrigde uinervtisy, it deosn't mttaer waht oredr the ltteers in a wrod are [...]
It's sort of true (although there was never any research at Cambridge that anyone has been able to unearth), but it's way more limited that "any word longer than 5 characters". This is a good explanation of it (arbitrary link found after a search):
I found this to be an unbelievable improvement in reading for me. It was automatic and effortless. I cannot believe how much faster that helped me to read without sacrificing on comprehension.
Unreal. Reading the non-bionic (?) text on the website suddenly felt like a sluggish chore, and for the first time I am realizing that I am actively using not insignificant amounts of energy to mentally parse text.
I can totally see how it could benefit some people but for me I don't find it works, I feel like it forces a slow on few letters words or group of words that I could have read in one glance. Also I tried a bit of speed reading and I feel like I already read faster than I can process information and that my eyes are not the bottleneck when deep reading.
There is a cliff in speed reading (at least in my experience) where you're reading speed for a particular reading exceeds your ability to process and/or enjoy the words you are reading.
Similar to audio books ... some podcasts/audiobooks I can listen to at 2.5x or higher and others I need to reduce to 1.25 or 1.5x. Some readers are already fast enough and I listen a 1x; Mary Robinette Kowal for instance.
I like this. I've always had to force jumping down line by line to pick up words if I was trying to speed read, but with the extra highlighting I'm picking them out and jumping almost effortlessly. It's pretty neat.
I'm honestly wondering if this would make me lazy for reading non explicitly partway bolded text.
I am very happy to hear that.
I don't think you will have problems if you can't apply BR. But you will probably find that the "reading pull" is less strong without Bionic Reading.
Tbh I scrolled down the page reading (and checked the about section), and am still none the wiser as to what exactly is supposed to be happening process-wise, and - assuming I'm supposed to be reading things more deeply, what the evidence for that is.
At the very least I think that should all be clearer given the primary claim.
We were taught "speed reading" in my first year of an electronics diploma. This feels very similar. I found speed reading only good for searching through text. Comprehension and memory goes out the window when I try it. If I want to get information in fast text to speech at 4x speed is more effective.
Works for me and my partner. Very impressive, I wish you the best. However, I think you're trying to capture something that's too big and elusive to be captured. If this improves peoples ability to consume information with a relatively trivial change (not calling your technology trivial, but relative to other infrastructure costs, it is), then there is no way that you won't see hundreds of spinoffs by companies and open source engineers, even if they are inferior products. And if you try suing them all, you will become the enemy in the public eye.
How do you plan to make money at the scale that this system could apply at, without alienating the world by forcing text flow through your system or pay some kind of royalty?
Just reading the example paragraph, I did feel at times speed was increased, but there were times the highlighted letters caused my brain to assume the wrong word, and when that happened it took extra effort (& time) to focus on the actual word.
Not 100% sure it's possible, but maybe you could turn this into a font family and license it. It would be a much better way to make money than an API that can be replaced with a 3 line function.
Interesting! I've been using this casually on iOS (via Reeder [1]) for at least a year. One thing that absolutely flips the effect on it's head (for me, personally) is the switch to "dark" mode, or simply light text on dark backgrounds vs. dark text on light backgrounds.
I suspect this is likely a typographic side-effect, similar to traditional print where dark pages "swallow" light text where as the opposite happens when inverted.
Could anyone explain the rationale behind it? It seems that how we represent the letters and the words has significant impacts on how easily our brains digest them.
Also a bit of tangent, but I wonder, could fellow English native speakers generally skim books like less than three to five seconds on each page, and roughly get the idea of those pages? I'm a native speaker of Chinese and it's pretty easy for me to do it in Chinese, but there's no way for me to do the same in English despite I've been reading English for years.
This is how I speed read for decades (autodidact); I swoop over the page once with my eyes and then just jump over it more or less like they show here. Allows me to read multiple books per day.
I was actually not able to read their example text at all. I had to slow down, to read this slowly word, by word.
I trained myself since my time at university to mark important parts in text (either with text marker in a book or with bold typesetting in a digital text).
I then primarily scan only the bold/colored parts and skip over the non marked parts. So my eye/brain jumped from bold letters to bold letters. As these were not full words, I just had gibberish in my head.
Maybe for the untrained reader this might work. I know that I am just anecdata.
It would be interesting to see what demands this places on the source material.
There's a fair bit of work in say philosophy where for example the concepts being discussed may be things like relationships between things, or things themselves, or being, also unavoidably something that crops up in the actual discussion, you're comparing comparisons or saying that that that thing that is is. You kinda usually have to slow down considerably to break these things down to have any hope of actually understanding them.
I’ve been consistently suspicious of speed-reading solutions. I have a loud inner voice that insists on annunciating everything that is read before comprehension takes place. Speed-reading typically outpaces my inner voice.
Like many others, I scan over sentences multiple times. This has made me more a apt reader technical writing than creative or conversational writing.
Maybe I’m yet another smooth-brain falling for the placebo effect, but I truly perceived an increase in reading speed that did not outpace my inner voice. Nice!
I often hear this sentiment in book reviews. "This book could have been expressed in the length of an article, but then they couldn't have published a book."
I suppose the more common version is "this article could have been expressed as a tweet!"
I'd like to try this as an option in firefox's reader mode as that's the format in which I read most long texts.
This did on the surface feel like it is a help for me, in particular not so much the reading speed but the 'pull' of the words meant I was less likely to get distracted halfway through a paragraph and tab away. (Yes, I suspect I may have undiagnosed adhd).
Seems neat, but is there any research to suggest that this works? I'm skeptical because I'm given a claim, "This will help you read faster", and then asked to test the claim, which primes me to think I am reading faster, regardless of whether I actually am.
While in general I find that speedreading is mostly for poor content, since our speed of information comprehension is slower than any reading, this seems to be the most effective implementation I've come across.
It might help the website more if the testimonials themselves were in the same way.
This seems obvious from the fact I love listening to audio books at 1.75-3x speed but hate actually reading because the latter feels too slow. It even feels hard to me to understand the information (let alone concentrate the attention) at the speed so low as the eyes can supply.
> this book author did not write his texts in written German, but in a form of the Swiss language, which I did not understand with reading. So I noticed that I have great difficulties to read the text.
This makes me angry. The German Swiss have deprived themselves of their own culture by normalising the practice of writing a German variety that is not theirs, and that is not spoken. The other three Swiss ethnicities wouldn't dream of doing such a nonsense.
It is nigh 100 years past time that someone who cares – following the footsteps of Adelung, Karadžić, Manzoni – pushes through with https://als.wikipedia.org/wiki/Dieth-Schreibung and takes it to the schools, so that in future generations people like Renato won't have trouble reading their own native language, and the pupils of Romandy learn a (secondary) language that is actually useful for orally communicating with their fellow countrymen, unlike what happens today.
This is really interesting. I wonder how difficult it would be to have a set of "training texts" where the reader/user is presented with different parameters to observe optimum points for speed vs retention.
Excited to read the study when it's published.
Honestly fixed word length per column independent of device or screen size is the #1 feature I wish for. I can't stand flow templates. When the nature of the paragraph changes per device it's very hard feel like you've been there before.
I prefer TTS with 1.5x speed coupled with highlighting the current words, and restyling the text to have 160% line height. There's something about having both the sound and visual at the same time that allows deeper concentration.
I read it and it did feel a little easier/quicker, but it's hard to discount it just being 'novelty' factor making it more engaging being the cause of that perception. But the real test would be a longer text.
> I have started to watch more videos at > real-time speeds.
I started doing this too.
But mostly because it helps with those videos where the speaker is speaking about an interesting topic but is speaking very slowly, or with large gaps or with many fillers. It helps making the viewing experience much more bearable.
I just wish most software were smarter about the speed ups. Naively playing everything X times faster can be weird. Either the pauses are too short, the words spoken too fast, or something. But it should dynamically stretch and squish more, I feel, to make it sound like a person actually speaking faster.
It was faster for me, I learned to read in Australia which apparently doens't have a very sophisticated reading education, sub-vocalisation kept us slow compared to other systems.
I normally have to make a conscious effort to speed read so this is intriguing. I’ve had a lot of success with Edge’s reader mode with verbs and nouns highlighted in different colours.
Thank you for your feedback.
Yes, this version of Bionic Reading is the beginning of the reading mode. Of course there are ways to structure the word types as well as the syllables.
I guess my only argument against these approaches is that I’d like to have a method that works for physical texts as well so practising speed reading seems the most impactful thing.
This is a pretty interesting find, it did help me read faster. I think the website would be better served by having the example texts be different though.
A friend of mine from another country with a different teaching style, when I showed him some speed reading apps laughed, saying that he was taught not to sub-vocalise words, and he is a smart guy. So maybe you have something similar?`
I had heard about that before, and I definitely try not to subvocalize now, so maybe that's the difference. I'm mainly surprised that so many people recognize it as definitely faster or slower, and so few that see no difference.
We have just finished a study.
The publication still needs some time.
But I can already say in advance that 25% of the subjects get great benefits in the method.
My finding before the study was that 10% of all readers gain an advantage in Bionic Reading.
And 10% is already very much...I find
What's your point? If you are implying the patent is just about ceil(len*0.3) you completely missed the application. In contrast, if you make/invent something new that uses 2+2=4 as an underlying fundamental mechanism, go ahead.
The rules implemented are very simple (taken from the patent application [1], the exact numbers are configurable).
If there are <= 3 letters, one letter is bold.
If there are == 4 letters, two letters are bold.
If there are > 4 letters, 40% of all letters are bold.
There is this claim as the first text on the website:
We are happy if as many people as possible can use the advantage of Bionic Reading. For this reason, Bionic Reading should be able to be integrated into existing apps and services. The benefit for the reader should be the main focus.
If that is truly correct, why not publish these 3 rules? Why hide it behind multiple patent applications and trademarks? Why spend all this time and money on patents and an API that probably involve sending all text to some servers, just so that these 3 lines of code can be executed?
I don't really see it that way. Yes, the about link leads to a page where the goal of improving the way people can read is described, but it would be odd to see them say, "and we expect to make money from working on this project".
Each of us is motivated by different goals. I am motivated to have a comfortable life, and I've worked for different companies over time to achieve this. I also find personal satisfaction in helping others so I've occasionally make charitable contributions of time and money, but I don't expect that everyone else is going to make the same choices I have. We don't know the motivation that drove the inventors of this new reading technology. Maybe it was to simply make money for riotous living, but it could be that they have a critically sick partner that needs expensive surgery. These are personal and private choices that I don't feel a need to weigh. Improved reading technology sounds like a good thing for not just the inventor, but for all of us.
When I buy some peaches from a farmer at a stand by the highway, I want the peach more than I want the dollar I have in my pocket. The farmer wants the dollar more than the peach. After I buy the peach we are both better off. I don't need to feel that the farmer is a non-profit farmer. I expect the farmer to be happier after the sale, for otherwise next year he may not even be there for my benefit during peach season. What I do expect, is that the farmer doesn't misrepresent his product, that he doesn't steal my credit card number, that he holds up his side of the bargain; if so, next year, I will be there to buy his peaches again.
I've invented a number of important technologies and benefited from doing so. Some, when I worked for others, ended up being patented. Such is the environment we find ourselves in, but for many of my ideas I chose not to protect them by patents and let these be adopted by standards organizations instead. Personally, I don't think patents are a good fit for software, and I would rather see better use of copyright protection for software or a simpler much shorter patent, say three years for software. The open source software movement has been an amazing gift to humanity, but so has the deluge of products (and peaches) produced for the profit motive.
I have mentioned this a few times here already, but I will mention it again, even if I risk looking like a spammer, as it seems this point hans't been made strongly enough in this thread: This discovery is not new.
So I am absolutely against patenting something that is 1) not new 2) that has a very simple implementation and could thus easily be disseminated widely if patents weren't "protecting" it. Reminds me strongly of the various patent wars regarding scrolling on the smartphone and other simple user interface gestures; in the end everyone looses out.
If we expect everyone to innovate and build useful tools solely out of the goodness of their heart, then we can expect far fewer useful tools & innovations in the future.
I think the opposite is true. Patents are hampering further progress that would built upon the innovation or make use of it to increase productivity. but said progress often is not happening, if the costs of obtaining a license is high, or if first tedious communication has to happen with the author to obtain green light to use his invention. Rapid innovation looks different.
If you innovate without filing a patent, you get a head start. The time the you have until other catch on - that is how you can monetize your invention.
I wonder how bolding more on uniqueness would work? Instead of just first characters.
For instance now THeir and THere will be bolded the same, on a quick glance that can be confusing. (Using caps as HN can't do bold)
Would it increase pattern recognition in the brain if it instead was THeIr and THeRe, or some other variant where the bolding highlights the unique parts of a word compared to other common words?
Well I wrote the idea here, so don't try patent it, hehe.
Haha...nice
The basic problem with reading is that your eye needs a fixed point...that's why they should be "blocks". This one doesn't have to be at the very beginning though ;)
But cool comment Hehe
True. You made me remember a different concept, where words would flash at the same point, but centered around some fixed highlighted character per word.
https://m.youtube.com/watch?v=ky8DP55YEO0
Edit: but thank you for sharing your concept here, very interesting, and I like the discussions it created.
I had an iOS speed-reading app at the time and considered integrating their API because users requested it. Not only would they have required individual users to sign in to their API via my app, IIRC their terms allowed them to track what each user was reading. Naturally I declined. I hope OP’s terms of use are more respecting of user privacy.
As GP said, the exact numbers are configurable, therefore your simplification may not work. BTW, none of the examples (the first text and all the subsequent variants) on the web site follows exactly these 3 rules, it must be something more.
Great product that might get lost because of overthinking the business model (say that even as a fellow Swiss). Think the giga-opportunity here is for a chrome extension, releasing it for free or freemium won't interfere with your patent (and if you don't, someone else will, regardless of the legal aspects). Either way, congratulations on this innovation, it's elegant.
Ja ich hatte bereits zwei Browser Extensions (Chrome, Firefox) als Beta-Versionen. Aber die Erkenntnis war, dass sehr viele Websites — da diese ja individuell erstellt werden — keine gute Experience ergab.
Deswegen habe ich mit Silvio Rizzi (Reeder5) getestet, wie es am besten nutzbar gemacht werden kann.
Hab aus Fehlern gelernt, aber wir denken dass es viele Entwickler gibt, die BR in ihren Leseprodukten anbieten möchten...
Aber vielleicht liegen wir da ja auch falsch. Man muss es versuchen und dann weitere Erkenntnis daraus ziehen.
The problem is that since you've patented these 3 lines of code, I'm not sure many other can legally write a browser extension for this functionality. I'd like to use this in Safari on iOS, for example, but... I can't.
Clearly these three rules are a fact about human pyschology rather than an invention and should be ineligible for patents. I understand that New Zealand law for example classifies pure software inovation as categorically not being invention. I'm unsure of what hoops you'd have to jump through to take advantage of this fact to release a browser extension for example, but it seems like it should be possible, though code signing and stuff might cause problems.
Thank you very much noeltock.
Yes, I already had two browser extensions (Chrome, Firefox) as beta versions. But the finding was that a lot of websites - since they are created individually - did not give a good experience.
That's why I tested with Silvio Rizzi (Reeder5) how it can best be used. Learned from mistakes, but we think there are many developers who want to offer BR in their reading products...
But maybe we're wrong about that. You have to try it and then draw further knowledge from it.
A dear greeting from Chur Renato
I came across BR during my studies as a typographic designer.
There was a language problem there.
That's why I realized that reading and listening are completely different.
You can understand a language by listening to it.
But reading the same language does not mean that you can understand it.
No problem, but the second half of my question was genuine.
I can't imagine this working too well with German words ? e.g. how much of Rechtsschutzversicherungsgesellschaften, Kraftfahrzeughaftpflichtversicherung, Lebensabschnittpartner or Donaudampfschiffahrtsgesellschaftskapitän would you need to highlight ?
I don't speak German, but surely with compound words, your concept doesn't work that well ?
Slightly off-topic - while I know that there are these lists of “crazy long German words” to make fun of each one of your examples besides the “Donau..” I actually used normally in conversations in the past.
Even just a couple days ago I came across “Personalisierungsinfrastrukturkomponente“ [0] in a news article, so it is not just a rumor that Germans like long-ass words :)
[0] Apparently these are the devices used by the government to verify passports and / or fingerprints of refugees.
For languages with compound words it looks like you'd need to parse out the sub-words and bold the first few letters of each one. That's doable but now you need language-specific dictionary files etc.
> If there are > 4 letters, 40% of all letters are bold.
Very cool :)
I would have thought there is also value in bolding the last-x-amount of letters.
I forgot the name of the "concept" but it's been shown that the beginning and ending parts of words are the most "recognizable". You usually see a sample of that on those 'silly-facebook-posts', example:
"I cnduo't bvleiee taht I culod aulaclty uesdtannrd waht I was rdnaieg"
Looking at the example in one of the apps (Reeder 5) doesn’t work 100% the same as the patent. For example, neumann (a last name) is bolded as NEUMANn the first time it appears and NEUMann the second time it appears.
If you want to read the patent (in French), it's here: https://data.inpi.fr/brevets/FR3052587. I couldn't find the German patent (10 2017 112 916.2), but I'm guessing it's the same substance.