This is an experiment from 2011 in which the author produced a font by averaging all the fonts on their system.
I'm reposting it here because I noticed that this looks a lot like the uncanny valley produced when an image AI tries to make text, which makes perfect sense: it's a statistical average of fonts.
I've been called out previously here for having unknowingly introduced some undefined terms to readers, and which they found to be perplexing.
And I took that to heart, because I don't want my words to be perplexing. I instead want them to be clear and easily understood.
In my corner of the world, I haven't held a mimeographed document in my hands for ~35 years. I found it reasonable to assume that a non-zero amount of people here might find the term to be unfamiliar.
So I provided definition of the term on the basis that it may be unknown to some readers, and that more information is better than inadequate information.
In this instance I would have preferred to use hyperlinked text for visual brevity, but that's not a thing on HN. The normal and accepted style on HN consists instead of using footnotes.
And at this point, generating footnotes is nearly entirely muscle memory for me. So a footnote (with a URL) was included.
Thank you for your attention on this matter, fsckboy. I'm pleased to discover that you've found my footnote to be so unusually compelling.
Same. It looks like the print you see in old books. Very pleasing to the eye. The lower case 'm' sticks out to me though, the second hump is raised a little too high.
Interesting how modern designers think readable fonts (with serifs, so people can reliably distinguish between Al and AI, for example) are "uncanny" because they don't follow the latest trends in ultra-minimalist "design" and other fashions.
Most serif fonts look too noisy to me, I made a website the other day and set a serif font and it immediately stood out to me as cluttered. I have zero design sense, though, so this is just an opinion.
I've used Averia (Serif Libre, specifically) for at least a decade as my primary font for email, web pages in 'reader' mode, writing long-form text, etc. I find it extremely legible, and even calming.
Ironically, I've been a typographer for decades, both for print and online. Averia might seem an odd choice for someone intimately familiar with typographic theory/history and the vast catalog of possible fonts. But there's a certain pleasure and comfort in a font that is not trying to stand out or do anything particularly special.
The trick is that there are two "averages" in play. Let's call them the "attractiveness average" and the "physical average". Your intuition concerns the attractiveness average: you know that there are beautiful people and ugly people, and "average people" must be somewhere in the middle, yes?
But when scientists average faces to create a perceived attractive face, they're averaging together the physical characteristics of each face: distance between the eyes, position of nose and ears on the head, size of mouth, symmetry, etc. The claim is that we have an intuitive, perhaps instinctual, notion of what humans "should" look like and our perception of attractiveness is roughly a measure of conformation to that standard. So an intuitively "average-looking" person is more correctly stated as having a medium amount of deviation from the human mean.
The main reason it has this "calming" feature is because it's imperfect. By averaging different, sometimes incompatible font faces the result looks like a letter pressed on a soft paper, with all it's natural imperfections. It looks real.
Somehow I was not aware of Averia and used Old Timey for exact same reasons in the past.
On the other hand, someone here mentioned "Lato", which to me looks exactly how two robots would write holiday postcards to each other.
Does this font simply ... Not look good to anyone else? It is visually kind of uncomfortable to behold. Maybe it's because it's a bit blurry feeling.
It sort of suggests to me that there's a lot going on with typeface design that we take for granted.
Edit: on closer inspection, the letter forms are kind of all over the place. The humps on the 'm' are lopsided, letter heights are sort of random. I think it's an interesting idea but to make it a more useful font would take a lot of manual fine tuning.
It's because the website is using cufon, a very early attempt at supporting custom fonts on the web using HTML canvas - basically every word you see is rendered as an image rather than text. The end result does not look good on hi-dpi screens like modern Macbook displays, probably they did not exist back then. The site mentions Google Font has a hosted version of it now and you can look at how it is meant to be rendered https://fonts.google.com/specimen/Averia+Libre
This is really cool. There's something very pleasing about precisely how unobtrusive it feels. You can also view the specifically serif-only and sans-serif-only versions here:
I think it would be really cool if a designer used these as a starting point for overall metrics, but then regularized and cleaned them up to exhibit consistent proportions and elements from character to character, without the wobbly parts. It really feels like it would become an ideal font family for reader mode, for journaling, just any time you want to focus on content and have a font that just "gets out of the way".
I’m surprised by how good it looks. This is really cool! I do feel like the Q and 4 characters need a little manual tweaking since the blur+threshold technique leaves some artifacts in the corners but those are such minor issues given how readable this font is overall. Love it.
> I call it Avería – which is a Spanish word related to the root of the word ‘average’. It actually means mechanical breakdown or damage. This seemed curiously fitting, and I was assured by a Spanish friend-of-a-friend that “Avería is an incredibly beautiful word regardless of its meaning”. So that's nice.
What I really love about both of them is that they instantly give you the impression of a real print made with real ink. Especially Averia - which makes sense, since it was averaged from all sorts of different fonts - has a lot of, for the lack of better word, excess fat on it. Something that may happen accidently while pressing "precise" font letter on soft paper.
Very cool project, thank you for sharing! To me, it raises some interesting questions around attribution of sources in derived works, in the same way that AI training does.
This is an experiment from 2011 in which the author produced a font by averaging all the fonts on their system.
I'm reposting it here because I noticed that this looks a lot like the uncanny valley produced when an image AI tries to make text, which makes perfect sense: it's a statistical average of fonts.
It also reminds me a bit of what text looks like after multiple rounds of photocopying. Like the handouts we'd get in grade school.
And the smell of weird, purple mimeograph[1] ink.
[1]: https://en.wikipedia.org/wiki/Mimeograph
For some reason, that smell is one of my favorite memories from my early school years. I really liked it.
That combined with the magic of "Wow! Copies?!"
[2]
[2]: I just wanted to add the most unnecessary footnote formatting possible as a kindness, so yours would no longer hold the crown as worst
That's clever, but:
I've been called out previously here for having unknowingly introduced some undefined terms to readers, and which they found to be perplexing.
And I took that to heart, because I don't want my words to be perplexing. I instead want them to be clear and easily understood.
In my corner of the world, I haven't held a mimeographed document in my hands for ~35 years. I found it reasonable to assume that a non-zero amount of people here might find the term to be unfamiliar.
So I provided definition of the term on the basis that it may be unknown to some readers, and that more information is better than inadequate information.
In this instance I would have preferred to use hyperlinked text for visual brevity, but that's not a thing on HN. The normal and accepted style on HN consists instead of using footnotes.
And at this point, generating footnotes is nearly entirely muscle memory for me. So a footnote (with a URL) was included.
Thank you for your attention on this matter, fsckboy. I'm pleased to discover that you've found my footnote to be so unusually compelling.
>I've been called out previously here for having unknowingly introduced some undefined terms to readers, and which they found to be perplexing.
i wasn't criticizing putting in a link; just put it in, it doesn't need layers of boiler plate [in this instance]
like, if i use the word sesquipedalian https://en.wiktionary.org/wiki/sesquipedalian
[in this instance] the square brackets add nothing but noise
I am appreciative of your criticism and have taken it under advisement.
However, I have noted that you do not pay me. It therefore does not behoove me to try to emulate your particular writing style.
Kindest regards.
Yes, I saw the exact same thing when you posted it - "oh, AI text looks like an averaging of fonts".
I don’t get uncanny valley feel from this one. It feels kind of great for me as a font.
Same. It looks like the print you see in old books. Very pleasing to the eye. The lower case 'm' sticks out to me though, the second hump is raised a little too high.
Interestingly it evokes Open Dyslexic.
Does AI look like this from an average or from training on the reams of copyright free books from a century ago? It seems more like the latter.
I wonder if you can ask AI to use a particular font for text in generated images.
Interesting how modern designers think readable fonts (with serifs, so people can reliably distinguish between Al and AI, for example) are "uncanny" because they don't follow the latest trends in ultra-minimalist "design" and other fashions.
Most serif fonts look too noisy to me, I made a website the other day and set a serif font and it immediately stood out to me as cluttered. I have zero design sense, though, so this is just an opinion.
I like readable serif fonts but this one really looks like an uncanny AI image.
I've used Averia (Serif Libre, specifically) for at least a decade as my primary font for email, web pages in 'reader' mode, writing long-form text, etc. I find it extremely legible, and even calming.
Ironically, I've been a typographer for decades, both for print and online. Averia might seem an odd choice for someone intimately familiar with typographic theory/history and the vast catalog of possible fonts. But there's a certain pleasure and comfort in a font that is not trying to stand out or do anything particularly special.
It's kind of like how if you take the average of enough male or female human faces, the result is a very pleasing, attractive face.
Same with music, a large group of people singing slightly off-key (each in their own way) tends to sound pretty good in aggregate
That's interesting, because my intuition of an "average" face is, well, average and uninteresting. Can you share your source?
Here's one of the papers concerning "averageness=attractiveness": https://www.sciencedirect.com/science/article/abs/pii/S10905...
The trick is that there are two "averages" in play. Let's call them the "attractiveness average" and the "physical average". Your intuition concerns the attractiveness average: you know that there are beautiful people and ugly people, and "average people" must be somewhere in the middle, yes?
But when scientists average faces to create a perceived attractive face, they're averaging together the physical characteristics of each face: distance between the eyes, position of nose and ears on the head, size of mouth, symmetry, etc. The claim is that we have an intuitive, perhaps instinctual, notion of what humans "should" look like and our perception of attractiveness is roughly a measure of conformation to that standard. So an intuitively "average-looking" person is more correctly stated as having a medium amount of deviation from the human mean.
The average face should be perfectly symmetrical assuming iid. Perfectly symmetrical faces are generally seen as attractive across all cultures.
The main reason it has this "calming" feature is because it's imperfect. By averaging different, sometimes incompatible font faces the result looks like a letter pressed on a soft paper, with all it's natural imperfections. It looks real.
Somehow I was not aware of Averia and used Old Timey for exact same reasons in the past.
On the other hand, someone here mentioned "Lato", which to me looks exactly how two robots would write holiday postcards to each other.
Does this font simply ... Not look good to anyone else? It is visually kind of uncomfortable to behold. Maybe it's because it's a bit blurry feeling.
It sort of suggests to me that there's a lot going on with typeface design that we take for granted.
Edit: on closer inspection, the letter forms are kind of all over the place. The humps on the 'm' are lopsided, letter heights are sort of random. I think it's an interesting idea but to make it a more useful font would take a lot of manual fine tuning.
It's because the website is using cufon, a very early attempt at supporting custom fonts on the web using HTML canvas - basically every word you see is rendered as an image rather than text. The end result does not look good on hi-dpi screens like modern Macbook displays, probably they did not exist back then. The site mentions Google Font has a hosted version of it now and you can look at how it is meant to be rendered https://fonts.google.com/specimen/Averia+Libre
Wow, that looks completely different than how Safari rendered the site. Thanks for the link. I like the look of it hosted at Google.
Aaahhh, thank you.
It still looks a little funky but way more readable.
Try the GWF Serif variant. It has more satisfying symmetry. Other than 4.
This is really cool. There's something very pleasing about precisely how unobtrusive it feels. You can also view the specifically serif-only and sans-serif-only versions here:
http://iotic.com/averia/preview.php
I think it would be really cool if a designer used these as a starting point for overall metrics, but then regularized and cleaned them up to exhibit consistent proportions and elements from character to character, without the wobbly parts. It really feels like it would become an ideal font family for reader mode, for journaling, just any time you want to focus on content and have a font that just "gets out of the way".
I’m surprised by how good it looks. This is really cool! I do feel like the Q and 4 characters need a little manual tweaking since the blur+threshold technique leaves some artifacts in the corners but those are such minor issues given how readable this font is overall. Love it.
Btw, "Avería" means "failure" in spanish
This is mentioned:
> I call it Avería – which is a Spanish word related to the root of the word ‘average’. It actually means mechanical breakdown or damage. This seemed curiously fitting, and I was assured by a Spanish friend-of-a-friend that “Avería is an incredibly beautiful word regardless of its meaning”. So that's nice.
Arabic ʕawāriyya for (goods) damaged in transit > Catalan avaria for a breakdown, damage > Spanish avería for a breakdown, something that has failed
https://en.wiktionary.org/wiki/avería
"Average" comes from Arabic for "damaged goods."
Reminds me of Old Timey [1] a lot.
What I really love about both of them is that they instantly give you the impression of a real print made with real ink. Especially Averia - which makes sense, since it was averaged from all sorts of different fonts - has a lot of, for the lack of better word, excess fat on it. Something that may happen accidently while pressing "precise" font letter on soft paper.
[1] https://webonastick.com/fonts/old-timey-mono/
I kind of dig this. It seems like it might look good on an ereader. Might have to upload it to my kobo!
It's been used in some visual novels by Nova-box:
https://store.steampowered.com/app/957820/Across_the_Grooves...
https://store.steampowered.com/app/738650/Seers_Isle/
I'd love to see the results for the same process used on monospace fonts.
yesss, waiting for it. i started using this font in my text editor and i find it super comfortable so i would love the same experience in my terminal.
Very cool project, thank you for sharing! To me, it raises some interesting questions around attribution of sources in derived works, in the same way that AI training does.
Reminds me of Supernormal font [1] averaged from widely popular fonts.
[1]: https://research.public.services/typography/
It looks like they need to turn down the ink flow on the press or the plate is a bit past its prime. I like it.
I think this font would look great for printed clues for a mystery game. Or on treasure map where the fonts tend to be over the top and illegible.
We already have the average font and it’s the execrable Lato.
https://fonts.google.com/specimen/Lato
https://fonts.google.com/?query=Averia
Looks blurry on my phone.
The site uses bitmap images, not web fonts.
Thank you, I love it!