Hacker News Reader

Comment by steve_adams_86 | Hacker News Reader

I find it incredibly fascinating how readable the end results are. They’re so wrong, so fun and attention-grabbing, and somehow it makes sense.

I’d really like to understand better how it is that we can decipher language so well. At least, it seems like we’re good at it. Why can I read that font or that calligraphy? It’s almost an abomination, haha. Yet my brain troops on, parsing and processing, turning those squiggles back into what it knows.

Truly fascinating stuff. I also enjoy the back story here. Exploring written language as a kid was a lot of fun. I settled on a strange way of writing, not much unlike the runes on the inside of the ring in the LOTR movies, and it was entirely arbitrary and almost an artistic decision. Totally impractical. I still write like that to this day.

View on HN

Reminds me of this classic:

> Aoccdrnig to a rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be a toatl mses and you can sitll raed it wouthit porbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.

I think that paragraph oversimplifies. (For example, splitting up the gh in "rghit" or the tt in "ltteers" would make them much harder to scan.) But clearly what it's getting at is true, that there are certain details of words and letters that we tend to focus on, and we can ignore quite a lot of noise as long as those details are in place.

Snopes doesn't definitely prove or disprove it but I don't think there ever was a paper that was published. https://www.snopes.com/fact-check/can-you-raed-tihs/

It's also nearly impossible to read longer words so I doubt it. e.g. Cigdrbame Uietisrnvy is nearly impossible to read.

I think for native speakers deciphering that that says Cambridge University isn't all that difficult and doesn't require much mental energy.

If you were primed for it by the earlier post, sure, but if you saw it out of the blue I doubt you'd know what it sefgiiind.

Most likely but doesn't that prove the point that reading is ultimately heavily context based and provided enough context you can distort the letters themselves to some absurd lengths and still preserve meaning?

Sure, writing has very low entropy, but the claim that you can easily tell what a word is as long as you preserve the first and last letters seems false.

It's not hard to understand what it signifies though.

Longer words seem to work OK if you add another constraint: keep all letters with ascenders and descenders in roughly the same relative order, eg, Cmbadigre Uisrevitny. Also, it works much better in a whole paragraph.

I think an important part is that ascenders and descenders are in approximately the same position as before.

> Cigdrbame Uietisrnvy

Sounds like a Welsh-Russian couple ...

Something I just noticed about that text is that "important" is spelled wrong - there is an E and no A. Searching for the first line shows me some results that have it wrong and some that have the A in there, I wonder how that came to be.

Aindroccg to a rccrsheeah at Ciardgbme Uistvinery, it d'nseot metatr in waht oedrr the leretts in a wrod are, the olny ioatnprmt tnihg is taht the frsit and lsat ltteer be at the rghit pclae. The rset can be a ttaol mses and you can slitl raed it wouitht pberlom. Tihs is bscueae the hamun mnid deos not raed eervy lteetr by ieltsf, but the wrod as a wolhe.

That was a strange sensation looking at "ioetnprmt" and feeling there was a spelling mistake there! (There is)

Ah! Good catch. I've edited the comment.

There is also a book about speed reading, where you can skip some words and still understand the sentence.

In my experience, most books on speed reading are such that you can just skip the entire book without losing anything important.

The ultimate form of speed reading

I spent about 8 seconds reading "The rset" as "The reset" and noting it was wrong.

It breaks down fast if you sort alphabetically the letters between the first and last letter.

Can we calculate how many bits of entropy we are saving in this compressed form? There seems to be (n-2)! ways of writing a word in a different way that is still readable.

I guess that could be a form of lossy text compression - where the end result is not completely right (the letters not being completely in the right order) but it's good enough to be able to read the text.

Information density of English text is 1ish bit per letter. There was some research about it, and also it's the compression rate you get from state-of-the-art algorithms.

Some research by none other than Claude Shannon :)

I have to subvocalize to read the quote, which I don't normally do. But once I do it's easy to read. I wonder what that says about language processing.

Wasn't this an early spam technique?

Yes they have been "shuffled" except for the 1st and last and they have kept a general ordering through the word

I too was surprised by how readable it was. I'd skipped to the examples at the end without reading the original alphabetical list and was still able to read them.

I found myself identifying the letters by focusing on the four corners / extremities of the letters.

Part of the reason is, we don't actually parse letters individually, but the patterns of words/letter-groups, sentences and general grammar.

Another responder mentioned the infamous Cambridge study. My Layman's take would be that the patterns are recognizable enough that the pattern-matching portion of verbal cognition can still associate them properly without conscious thought. And any "misses" are filled in or trained by contextual clues.

Watch cracking the mayan code (pretty sure that was what it was called). Spoiler alert. Turns out mayan letters are kinda like this. Lots of freedom as long as the basic shape is there. Lots of artistic freedom.

Douglas Hofstadter spent some time interested in letterforms and asking questions like "what variations are still recognizable as an 'A'?" He wrote two essays that ended up in his Metamagical Themas book, Variations on a Theme as the Crux of Creativity and Metafont, Metamathematics, and Metaphysics. Very interesting reading if you are interested in type and lettering. The second one involves the Metafont program that Donald Knuth wrote (in part) to create his Computer Modern font (used in TeX).

I wonder how OCR would perform on this script. Could it decipher as easily as we can?

Reply on news.ycombinator.com