Tiro Typeworks

Official account of Tiro Typeworks Ltd., a digital type foundry founded in 1994 by John Hudson and Ross Mills, specialising in custom fonts for multilingual publishing and computing.

#Typography #TypeDesign #Fonts #Unicode #WritingSystems #Linguistics #Palaeography #Calligraphy

#fedi22

Public Key

npub1rmg092al9kzwfgzrv57jx3frhm2efcw4eu42n8hfswjewv3autuq0wuntj

NIP-05 Address

TiroTypeworks@typo-social.mostr.pub

Profile Code

nprofile1qqspa58j4wljmp8y5ppk20frg53ma4v5u82u724fnm5c8fvhxg7797qpz4mhxue69uhhyetvv9ujuerfw36x7tnsw43qz9thwden5te0wfjkccte9ekk7um5wgh8qatzfj3dsp

Publishing to

relay.ditto.pub

relay.mostr.pub

Author Public Key

npub1rmg092al9kzwfgzrv57jx3frhm2efcw4eu42n8hfswjewv3autuq0wuntj

Show more details

Last Notes

2026-10-24 16:50:08 UTC

Does this make sense? Do you understand what is diagrammed here?
[See alt text for explanation.]
https://typo.social/system/media_attachments/files/117/496/750/568/707/785/original/663e501905f55a09.png

2026-10-23 17:58:54 UTC

This is a *lot* more complicated than you might think.
https://typo.social/system/media_attachments/files/117/491/514/222/639/888/original/d6e324f93720e243.png

2026-10-22 20:06:51 UTC

Tiro Kannada v2.00 in progress: now with improvements to arbitrary stack handling and new script extension inclusions.
https://typo.social/system/media_attachments/files/117/486/351/183/164/054/original/a0ea66f00e0cdba4.png

2026-10-20 00:34:04 UTC

Editing a machine-transcription of an interview recording with Robert Bringhurst, and quite delighted by some of the errors, especially ‘The Elements of Typographic Sorrow’.

2026-10-18 14:47:20 UTC

New study of palm leaf manuscript preparation. Interesting that prepared palm genera are more easily distinguished than unprocessed palm leafs.
@nprofile…mrzw
https://archaeologymag.com/2026/10/palm-leaf-manuscripts-x-rays-study/

2026-10-17 16:29:15 UTC

Yes, I am kerning obscure Sanskrit syllables with rare anusvara signs.I mean, you would, wouldn’t you.
https://typo.social/system/media_attachments/files/117/457/174/794/721/266/original/1ea15a27b79500fc.png

2026-10-13 22:32:45 UTC

I wonder if anyone has researched when Kannada subscript consonants began to be positioned to the lower right of base syllables?
In palm leaf manuscripts, lithography, and early metal types, the subscripts were centered below the base consonant shape (as remains the case in Telugu typography). But somewhere along the way, typographic practice for Kannada changed.
This looks very much like the sort of thing that might have originated in limitations of a mechanical typesetting system.
https://typo.social/system/media_attachments/files/117/435/917/702/028/859/original/2057b08818c377d6.png
https://typo.social/system/media_attachments/files/117/435/932/364/445/800/original/e55abd20e1971705.png
https://typo.social/system/media_attachments/files/117/435/969/057/024/045/original/fa5ed3841b3a3218.png

2026-10-07 21:22:43 UTC

Kannada subscript kerning checks generate great pareidolia.
https://typo.social/system/media_attachments/files/117/401/702/199/437/912/original/35d432497f002352.png
https://typo.social/system/media_attachments/files/117/401/719/564/153/743/original/65ce794e0548e51b.png

2026-10-03 22:27:23 UTC

I’ll be joining Ziad Matar, Nadine Chahine, and Peter Nowell in the final panel session of the upcoming ILT Font Tech & AI online conference. We’ll be talking about the impact and disruption of ‘AI’ technologies, looking at the reasons people both use and refuse them, and asking whether we can agree on guardrails and where to put them.
https://trust.ilovetypography.com/font-tech-ai/
https://typo.social/system/media_attachments/files/117/379/305/115/694/386/original/9087d25e2fc4824c.png

2026-10-02 15:58:38 UTC

Does anyone have a copy of Footnotes A (the biennial journal from La Police) with which they would be willing to part?

2026-09-27 16:31:04 UTC

Do any 2027 type, paleography, etc. conferences have dates — or preliminary dates — scheduled yet? I need to select potential travel blocks for next year before 27 October.

2026-09-24 21:19:43 UTC

Ten years on, still waiting for the variable font UI equivalent of this.
https://typo.social/system/media_attachments/files/117/328/079/017/141/913/original/97652fdd9527570f.png

2026-09-24 14:37:29 UTC

The preliminary programme for the upcoming ILT Font Tech & AI online conference has been published.
https://trust.ilovetypography.com/font-tech-ai/
It’s a bit of a punishing schedule for those of us on the Pacific coast, but I’ll get up that early for Toshi’s test-drive of the platforms claiming to AI-generate fonts.

2026-09-22 02:47:19 UTC

Difficult difficult lemon difficult.
https://typo.social/system/media_attachments/files/117/312/395/492/323/821/original/bc9df6a94fd92547.png

2026-09-08 03:26:10 UTC

Q. What was the first typeface to be given a name, in the modern sense of a unique name for marketing and identification of the typeface, rather than a label for a style or class or size of typeface?

2026-09-07 22:42:28 UTC

Should I take the lack of response as ‘No, there is no reason to ever use mark attachment class as a data type.’?

2026-09-07 21:21:07 UTC

Regardless of your response — possibly more or less nuanced — to the tooted opinion, I recommend reading Eli’s blog entry about nastaliq and his in-progress approach to it.

2026-09-05 22:14:17 UTC

Can anyone think of a reason to use GDEF mark attachment class in any situation instead of mark glyph set?
I am pondering whether a new tool should just always assume that a mark filtering group should be compiled as a mark glyph set in the GDEF table, and just ignore having an option to instead compile to a mark attachment class.

2026-09-03 19:43:33 UTC

In the perfect storm of stupidity and capitalism that is AI, Monotype have made a font identification app accessible in ChatGPT that initial tests show a) incorrectly identifies fonts and b) then directs users to MyFonts where they can purchase the font they’re not looking for. More testing is needed to confirm the pattern, so maybe test with images of your own typefaces and report here:
https://typedrawers.com/discussion/5684/myfonts-in-chatgpt

2026-08-24 20:48:39 UTC

On the one hand, the detail change I thought of requires editing 185 glyphs.
On the other hand, its an improvement, so...

2026-08-21 14:47:12 UTC

- reply

@nprofile…26hz The question not being asked here is ‘Why are the ascent and descent metrics in the fonts so generous?’ It’s because of font makers trying to work around the inconsistent interpretation of font metrics in different browsers and other software, so putting default linespacing into the ascent and descent metrics and setting the actual linegap metric to zero. If metrics were set according to the font format specification, instead of font makers trying to accommodate the ways in which the metrics have been misapplied by various kinds of software for nearly forty years, the questions you are asking would not really arise.

2026-08-21 05:52:38 UTC

- reply

@nprofile…ale9 I am re-reading it, for the first time since the 1990s, in preparation for interviewing Bringhurst about it at the end of the month.

2026-08-19 18:38:07 UTC

- reply

@nprofile…n00n Interesting. Found a not great digitisation online, and it is indeed using a Latin transcription.
https://typo.social/system/media_attachments/files/117/123/619/292/320/026/original/22773641ba7c66dc.png

2026-08-19 18:25:17 UTC

- reply

@nprofile…n00n Can you share some images of the internal pages? I would like to see the Devanagari and or Nastaliq types used.

2026-08-18 14:54:36 UTC

- reply

@nprofile…9ng9 I’m familiar with the script ordering, which, frustratingly, ignores the most relevant and reliable data point which is the dlng tag in the meta table.
The first versions of the Tiro Indic fonts had a relatively small Latin subset, and they were sorted with by the Indic script. The new versions have a larger Latin subset, and ID is sorting those newer versions with the Latin fonts, even though the meta table specifies the Indic script as dlng.
_
Something else seems to be going on with this one font, though, because the error is affecting only one member of the font family. All the fonts have identical glyph sets and cmap tables. And the problem is limited to ID 2026: in the previous version it is fine.

2026-08-17 22:23:59 UTC

Anyone else come across this sort of thing in the Adobe InDesign 2026 font menu? The Regular weight of my latest Tiro Bangla build is showing up as a separate family labeled
(Unknown Script 13)
while the other three weights are sorted together under the preferred family name as expected.
In other apps, including InDesign 2025, all four weights are correctly showing in the font menu.
https://typo.social/system/media_attachments/files/117/113/153/539/822/523/original/78ac083bd73796fe.png

2026-08-15 22:35:06 UTC

Done! I didn’t consciously leave this glyph to the end, but its where I ended up through the parallel filters of enthusiasm and reluctance.
https://typo.social/system/media_attachments/files/117/101/894/681/030/561/original/26d9554f850fed97.png

2026-08-15 21:32:08 UTC

- reply

@nprofile…wxfp The term defines itself pretty well. From a practical perspective, the most important factor for me is which glyphs feed into most others in terms of shapes, details, and proportions. Those are sometimes frequent letters, but not always.
That said, when I am working on Indic scripts with Fiona Ross she likes us to start with the letter ka because it is considered auspicious.

2026-08-14 21:31:28 UTC

- reply

Yoga.
https://typo.social/system/media_attachments/files/117/095/982/190/833/385/original/c3e45a4eb536c67a.png

2026-08-14 16:42:51 UTC

Do you ever feel that you’re being trolled by a writing system?
https://typo.social/system/media_attachments/files/117/094/850/651/575/894/original/8088d194ff23054e.png

2026-08-14 14:28:03 UTC

@nprofile…lwfy The deterministic precision claims for Primework reminded me of your $1 bill rendering challenge. I wonder if we have a contender?
https://typedrawers.com/discussion/comment/72088/#Comment_72088

2026-08-13 16:20:56 UTC

- reply

@nprofile…mrzw @nprofile…w44f U+271D ✝ has never been included in many fonts, so U+2020 † is used in a lot of places, giving the impression that the majority of people who died in Germany were stabbed to death.😁

2026-08-13 15:02:47 UTC

- reply

@nprofile…65my I don’t understand what aspect of this irritates you.

2026-08-12 15:25:39 UTC

Bangla, you are so strange.
https://typo.social/system/media_attachments/files/117/083/221/758/715/853/original/37a491c76cc9d472.png

2026-08-11 15:43:36 UTC

- reply

@nprofile…8xdc @nprofile…yeul That’s the approach I took in Aldhabi, but apparently only when preceding lam (which in Aldhabi only happens if the default ligation is disabled). I need to update to apply this attachment also for other initial letters (along with a *lot* of other improvements and fixes: it was a project in search of adequate tools).
https://typo.social/system/media_attachments/files/117/077/634/696/685/949/original/27308a2df5e5bb04.png

2026-08-11 15:39:52 UTC

- reply

2026-08-11 15:10:42 UTC

- reply

@nprofile…dfhn @nprofile…g56g I wondered if this query might be related to a documented pantograph master size.
‘Midpoint’ implies something halfway, and I wasn’t sure if that was what you meant. Given that optical size modifications accelerate towards the small end, I wonder if one could indeed identify a midpoint in terms of the average of the stem thickness at the smallest and largest sizes?
Applying that calculation to the Sitka family, I end up very close to the nominal Subhead size (which in Sitka targets 16.7pt).

2026-08-11 04:04:08 UTC

- reply

@nprofile…dfhn @nprofile…g56g What does ‘midpoint’ mean in this context?

2026-08-10 17:43:07 UTC

- reply

@nprofile…av8d I have occasionally been tempted to purchase a copy of Fry’s Pantographia, but then I read bits of it and am not sure about spending that amount if money on something whose primary value is comedic.😜

2026-08-10 15:25:04 UTC

- reply

@nprofile…8xdc @nprofile…yeul Getting the unconnected alif raised relative to a height determined by the following letter group is tricky in OTL. The cursive connection anchor type was defined with a couple of assumptions in mind: a) that the letterforms visually connect, and b) a single pair of entry and exit anchors was sufficient for all connections. What this means in cascading Arabic is that one can’t define a entry anchor on an initial letter without all other initial and medial letters with exit anchors connecting to it.
The isolated alif needs to be handled as a special case. In theory, one could define a contextual cursive connection lookup, in which the entry anchor on the initial letter would only be active when preceded by an alif. I have not tried this, though.

2026-08-10 00:15:13 UTC

- reply

@nprofile…8xdc I’m really surprised that Gulzar doesn’t include code to raise non-final alif to align with the following letter group.

2026-08-08 23:48:34 UTC

- reply

@nprofile…8xdc I also encourage you to submit a proposal for a demo/talk to the online ILT font tech conference:
https://trust.ilovetypography.com/font-tech-ai/
(Deadline is in a couple of days, but may be extended?)

2026-08-08 23:45:32 UTC

- reply

@nprofile…ampu The term ‘smart font’ has been used colloquially since the mid-1990s as a generic label for any font format that contains layout information beyond simple spacing and kerning data, so e.g. AAT, OTL, Graphite, ACE, etc.
Without knowing exactly how your approach works ‘under the hood’, I am guessing this fits into the broad category of parametric fonts, which may or may not help in coming up with a unique product name.

2026-08-08 23:38:37 UTC

- reply

@nprofile…8xdc Thank you for writing this detailed description of the process. i have a much better understanding of how you have approached this now.

2026-08-05 16:51:51 UTC

Shape of the day.
260805 Tiro Bangla bja conjunct.
https://typo.social/system/media_attachments/files/117/043/928/680/332/954/original/1aa4b6c8e55f9da9.png

2026-07-31 17:48:07 UTC

- reply

@nprofile…58hj @nprofile…mrzw Think about any piece of text, presented to a reader in some medium. So a physical book constitutes a text system involving particular typefaces, page layouts, and quite probably subsystems for different kinds of text within the whole.
An individual piece of calligraphy also constitutes a text system.
And this toot is presented to you in a text system that doesn’t know in advance what the content of the text will be.
Because of the varied nature of these media and the way the systems are devised, I wanted a term that is general and not tied to a specific technology such a typography.

2026-07-31 17:24:16 UTC

- reply

@nprofile…58hj @nprofile…mrzw Then there are what I have recently started calling text systems, i.e. the particular application of a writing system to instantiate an individual text. I needed a technology-agnostic term that covers not only typography but all the other media through which writing systems are applied to texts.

2026-07-31 17:20:54 UTC

- reply

@nprofile…58hj @nprofile…mrzw It’s a continuum of difficulty, I think, with important things to recognise about the parallels of learning how to design for different writing systems within a script, and learning how to design for the writing systems of another script.
The most important thing to recognise is that we’re mostly learning about writing systems, and then abstracting to the script level, rather than learning about a script per se. A script is a system of graphical elements and general rules for how they are combined, but it is only at the level of the writing system that the system is applied in a specific set of elements and where exceptions to the general rules might occur.

2026-07-31 17:11:40 UTC

- reply

@nprofile…58hj @nprofile…mrzw Your observations about relative perceptual fluency between your Hebrew, Russian and English writing systems are interesting. Something to bear in mind is that individual scripts may be more-or-less readable than others, and within a script some writing systems may be more-or-less readable that others, and within a writing system different styles of text may be more-or-less readable. Funnily, a lot of people who naturally accept the last of these resist the notion that the same applies between writing systems or between scripts.

2026-07-31 16:57:59 UTC

- reply

@nprofile…58hj @nprofile…mrzw As a literate user of a native language, you have a native writing system, and that certainly gives you advantages in learning other writing systems within that script, especially for related languages with similar phonetic mappings. But there are characters used in some writing systems that you have likely never seen before, which you can’t read and about which don’t know things that a that a type designer needs to know, e.g. how do they vary idiomatically, in different styles of text? what do the cursive forms look like? with what other characters do they occur?
Consider the first time a non-German designer approaches an ß. Being a native reader of English or French etc. doesn’t help much, even though this is a Latin script character.

2026-07-31 16:50:55 UTC

- reply

@nprofile…58hj @nprofile…mrzw My point is that English and Russian are languages, and the application of the Latin and Cyrillic script to those languages constitute specific implementations of the script: a writing system. An individual writing system typically involves only a subset of the graphical signs that constitute a script—especially true in the case of massively extended scripts like Latin and Cyrillic—, in combinations that are limited by the phonetic inventory of the language.

2026-07-31 15:59:32 UTC

- reply

@nprofile…mrzw I also challenge the notion that anyone has a ‘native script’. What we have is one or more native writing systems, i.e. implementations of a script for particular languages, and as type designers we have to study and learn about other writing systems within that script. This has implications not only for parallels in how we learn about other scripts, but also for the limitations of information on those scripts from native users who might only know how they apply to one language.

2026-07-30 17:25:00 UTC

- reply

@nprofile…8xdc When I see grid based letterforms with no curves, I never know whether I am looking at a viable proof of concept or at a concept that ultimately can’t manage anything else.
Let’s see the nastaliq.😁

2026-07-29 21:22:43 UTC

Stumbled on this today, while digging around in the Murty Classical Library of India project archive. One of the first texts prepared for publication had been submitted in an 8-bit hack font using visual ordering. Karsten Luecke wrote a script to convert the text from that 8-bit encoding to Devanagari Unicode. I prepared this table of sample complex mappings and reorderings to illustrate for Karsten the kind of things that would need to be handled in the script.
https://typo.social/system/media_attachments/files/117/005/343/612/134/377/original/aede824c5b605dc4.png

2026-07-29 18:07:55 UTC

This online Indian prosody tool is very cool. It’s intended to help people writing metrical verse, but I am enjoying dropping in 16th Century samples to figure out in which classical metre they were composed.
https://chandas.org/
https://typo.social/system/media_attachments/files/117/004/593/956/182/948/original/ef35c8003b858d3d.png

2026-07-29 17:31:08 UTC

There are people who will tell you that the Latin letter S is a difficult shape to get right.
https://typo.social/system/media_attachments/files/117/004/448/945/833/838/original/cc8153666af5e36b.png

2026-07-28 17:54:46 UTC

- reply

@nprofile…gj52 Thanks to this, I just spent the past hour reliving the summer of 2009 via the listserve discussions that led, eventually, to the charter of the W3C webfont working group. I’d forgotten how cheeky I was.

2026-07-27 18:31:04 UTC

Shape of the day.
260727 Tiro Bangla Bold jja conjunct.
https://typo.social/system/media_attachments/files/116/993/357/424/084/748/original/56000f509ca72dd6.png

2026-07-20 18:20:36 UTC

- reply

@nprofile…ahkr @nprofile…ale9 It depended on the environment. At the target ppem size it rasterized to b/w shapes that matched the existing bitmap font, regardless of whether hinting was active. That was the requirement for one, backwards compatible environment. Within other environments it was either b/w or greyscale rendered and hinted for a range of sizes.

2026-07-20 17:35:12 UTC

- reply

@nprofile…ale9 Not bitmap per se, but your query reminded me of a *very* early project Ross and I worked on in the mid-1990s: a TTF for a terminal emulator maker that needed to match existing bitmap shapes at a specific size (without using hinting) in one target environment, but which would scale in others. The results are, as one might guess, weird.
https://typo.social/system/media_attachments/files/116/953/498/919/492/012/original/5e458399d15319cb.png

2026-07-20 17:14:05 UTC

- reply

@nprofile…ale9 A couple of years ago, Mark Simonson succeeded in getting the ancient Mac bitmap Font Editor from the 1980s running on an emulator.
https://typedrawers.com/discussion/5070/apples-ancient-macintosh-font-editor-app

2026-07-20 17:10:43 UTC

- reply

@nprofile…ale9 Ben Zotto’s article about the original Mac bitmap fonts:
https://bzotto.medium.com/hidden-sheep-and-mac-typography-archaeology-efce770da76c

2026-07-20 15:43:14 UTC

It’s been almost seven years since anyone posted to it, but the Punchcutting sub-forum on TypeDrawers remains a treasure trove of practical information as well as a delight in itself (much of it thanks to Ramiro Espinoza). When tired of staring at beziér curves, I like going and looking at the photographs of the tactile processes of our metal heritage.
https://typedrawers.com/categories/punchcutting
https://typo.social/system/media_attachments/files/116/953/050/668/369/658/original/d4eb02b1a8bd5141.png
https://typo.social/system/media_attachments/files/116/953/060/688/979/120/original/2fc702abb884cffb.png

2026-07-17 15:26:38 UTC

- reply

@nprofile…ale9 One of my favourite examples of this kind of thing is the work Christian Thalmann did for the Shadow & Bone television series, especially the Hangul-inspired Shu script.
https://dedalvs.tumblr.com/post/712063868348841984/i-mentioned-that-christian-thalmann-got-the-chance

2026-07-13 23:28:02 UTC

- reply

@nprofile…8xdc It is possible for your prediction about the longterm potential for AI in font making to be legit and for there *also* to be a current AI bubble. The 1990s dotcom bubble didn’t mean the World Wide Web failed to become a major, transformative technology, but the collapse of the bubble affected the entire tech sector for years, including companies not directly involved in the bubble financing. The amounts of money involved in that bubble were miniscule compared to those sunk in AI.

2026-07-13 19:41:26 UTC

- reply

@nprofile…uju4 Context.
https://typo.social/system/media_attachments/files/116/914/364/594/469/818/original/302c112352843330.png

2026-07-13 19:39:24 UTC

@nprofile…uju4 To be very clear: ParaType is a company that is materially complicit in this. ATypI should not be providing a platform to that company’s representatives.
I say this as a former vice-president of the association, and as someone pained by ParaType’s institutional support of of the murder regime in Russia and its aggressive war of territorial expansion in Ukraine. I was friends with the founders of the company, and am dismayed to see their international legacy wrecked in this way.

2026-07-13 01:12:42 UTC

- reply

@nprofile…ale9 Starting to take the weekend after years of working every day was one of the best decisions I ever made. I began by not working on Sundays, but felt so much better for it that I started taking Saturday too.

2026-07-10 22:43:48 UTC

Shape of the day. This is one for which I have ambivalent feelings: I enjoy the shape, but I resent spending time designing a character that does not occur in any language.
260710 Tiro Bangla Bold ḹ vowel letter.
https://typo.social/system/media_attachments/files/116/898/095/941/352/427/original/0e4e09f2e77d1b98.png

2026-07-10 15:46:57 UTC

Final tally.

2026-07-09 15:10:57 UTC

Boosting this to see if we can get more responses before the poll closes.

2026-07-08 23:21:48 UTC

Shape of the day.
260708 Tiro Bangla kṭra conjunct.
https://typo.social/system/media_attachments/files/116/886/911/652/385/657/original/93134dbbfc88c424.png

2026-07-08 22:37:17 UTC

This lecture by Tal on his work for National Geographic is great. The story of the invention of the Bumstead Photo-Lettering Machine is little treasure in itself.

2026-07-08 19:49:26 UTC

- reply

@nprofile…8xdc Thanks for that link. Zak is great.
His 1979 book is the most frequently consulted text on my Arabic shelves: it is slimmer than most, but none of the others do such a good and systematic job of identifying the different styles, even the obscure ones.

2026-07-08 19:00:57 UTC

- reply

@nprofile…8xdc You are very quick to extrapolate from your own preferences and processes to thinking that the ways other people like to work are becoming obsolete and the tools they use are not going to survive. I like that you are finding new ways to work, and maybe those ways will appeal to some other people, but that doesn’t imply anything general about processes and tools.
Tool choice seems to me quite personal. I know many people who love Glyphs, while it would be hard to overstate just how much I don’t. Luc de Groot still uses FontLab Studio 5. I still do my OTL production in a tool written thirty years ago in Visual Basic that requires me to maintain a Windows machine.🤷

2026-07-08 00:17:10 UTC

- reply

@nprofile…w44f Literally, zonder schreef, but the compound schreefloos (serifless) is far lovelier.

2026-07-07 23:25:56 UTC

- reply

@nprofile…6t4x The way I explain it to Americans is that Sharjah is to Dubai as Kansas City Kansas is to Kansas City Missouri: part of a single conurbation but in a separate state.

2026-07-06 15:50:27 UTC

- reply

@nprofile…ns6x @nprofile…wguy I wouldn’t at all describe ILT as pro-AI. They’re providing a forum for critical discussion of the technology and to help people understand how it is already being used and what the implications of that are. I think those are important conversations to have.

2026-07-04 04:11:46 UTC

I read about this 1965 BBC interview with David Jones in Thomas Dilworth’s bioography, and was thrilled to find it on YT.
https://www.youtube.com/watch?v=psQkOT7eNwE

2026-07-03 15:41:16 UTC

Anyone else encountered this?
In InDesign, using the italic keyboard shortcut results in a weight change in some font families, e.g. Regular upright text becomes Black Italic. I have examined the fonts closely, and don’t see anything that would explain this behaviour. It occurs only in InDesign, not in Illustrator or Photoshop.

2026-07-03 15:25:09 UTC

- reply

@nprofile…tp2m @nprofile…nqgz I think we are going to see a huge amount of new things being created, across all areas, that are a mix of original human authoring and AI generated parts. The latter will include
a) parts derived from other work by the same human, i.e. locally trained machine learning,
b) parts derived under license from work by other humans, and
c) parts derived from obscure soup of general AI training, in which it is difficult to identify the original works from which the AI agent is deriving the parts.
I doubt there is going to be a one-size-fits-all intellectual property solution even in a single jurisdiction.

2026-07-03 15:15:29 UTC

- reply

@nprofile…tp2m @nprofile…nqgz Copyrightability of the derivative work is a red-herring. Let’s say that I, a human being, create a work that is derivative of another. Technically the work I made could be subject to copyright because I am a human, but any claim I make to it is null because the derivative work is subject to the original copyright of the prior work from which it is derived. If I am creating a derivative work under license from the original copyright holder, I am doing so subject to the terms of that license, which may include terms specifying the downstream license under which I must publish the derivative work.
Substituting an AI agent in place of me in any of that sequence doesn’t change the nature of the relationship of the derived work to the original work.

2026-07-03 15:05:06 UTC

- reply

@nprofile…tp2m @nprofile…nqgz That isn’t what I argued. Derivative works are a well-established notion in copyright law. The fact that there is now a new technology for producing derivative works does not magically make them not derivative. What matters is the copyright of the original work, not the putative copyrightability of the derivative work.
Scott’s original question was about license, which follows from the original copyright, i.e. it is the holder of the copyright who has the right to grant licenses for the use, reproduction, distribution, or making of derivatives of a work. Some libre license models have a clause that imposes a downstream license on derivative works. Scott’s question as to whether that applies to derivative works made by AI seems valid and open in law.

2026-07-03 02:05:50 UTC

- reply

@nprofile…tp2m @nprofile…nqgz The counter argument is that a derivative work is a derivative work regardless of how it is derived, so regardless of whether it is itself copyrightable or licensable it may still be subject under law to the copyright or license of the work(s) from which it is derived.

2026-07-01 15:48:20 UTC

- reply

@nprofile…e9hc If you’re looking for something in Europe, maybe see if Takahiro kettles from Japan are available? They’re a simpler shape than the Hario Buono, so in theory should heat more quickly and evenly on an induction hob.
From North America, there’s the Barista Warrior pour over kettle, which I like. The built-in thermometer has both C° and F°.
https://baristawarrior.com/products/pour-over-kettle

2026-07-01 15:24:53 UTC

- reply

@nprofile…aane It’s going to be a total bloodbath.
https://typo.social/system/media_attachments/files/116/845/406/155/813/105/original/443c96c642529e3e.png

2026-06-27 17:08:46 UTC

https://mastodon.social/@gutenberg_org/116821334604477915

2026-06-26 19:14:42 UTC

When a human lies to me, I am dealing with sin, which is something I know and understand and, crucially, must be open to forgiving. When AI lies to me, what are my responsibilities, given that it does not seem capable of repentance?
https://typedrawers.com/discussion/comment/71696/#Comment_71696

2026-06-24 23:55:32 UTC

New series, maybe?
Shape of the day: something I am working on, or looking at, or remembered.
260624 Tiro Bangla jja conjunct.
https://typo.social/system/media_attachments/files/116/807/781/508/180/293/original/c14e10d1ab2dac03.png

2026-06-24 23:43:29 UTC

- reply

@nprofile…aswa I’ve wondered if these two shapes — also phonetically unrelated — ever cause confusion.
https://typo.social/system/media_attachments/files/116/807/727/803/448/232/original/f5f4fb359caa4eb2.png

2026-06-23 17:31:02 UTC

- reply

@nprofile…ale9 Just finished listening to Brandon Sanderson’s 2026 Tolkien Lecture on the nature of fantasy, which you might enjoy.
https://youtu.be/fKvmYcGOCkE?si=4V0UNf3gG3veuwGN&t=104

2026-06-22 15:20:50 UTC

Essential reading.

2026-06-20 22:27:48 UTC

Short article, republishing a text I wrote for ISType in 2015, considering how we understand emergent properties, in type as in birds.
https://www.tiro.com/articles/reflections-on-the-common-swift

2026-06-20 15:14:27 UTC

- reply

@nprofile…narm @nprofile…mrzw I was thinking phonetically unrelated, i.e. the same shape is being in conjuncts that represent different underlying letters representing different sounds.
Ben’s ‘etymologically’ unrelated examples suggest probable explanations for some of the shape convergences I see in Bangla and other scripts. But, of course, unless one can trace the evolutionary steps, such explanations are speculative.
Some years ago,, I wrote a short text for one of the ISType programmes about how genetically unrelated organisms evolve similarities based on their environmental circumstances, and how this is also true of typefaces. Hmm. Maybe I will dig that out and put it on the website....

2026-06-19 01:42:37 UTC

One of these things is not like the others.
The way in which Bangla re-uses the same shape different letters in conjuncts is disconcerting.
https://typo.social/system/media_attachments/files/116/774/207/888/941/433/original/d65c8af8aff54d2d.png

2026-06-18 01:52:57 UTC

- reply

@nprofile…yeul Hmm. In my Indic naming convention LLa and Lla would indicate different things. Would this normalisation potentially be an issue?
https://typo.social/system/media_attachments/files/116/768/590/676/772/334/original/3a0bce14b0d6e6a0.png

2026-06-15 20:05:10 UTC

- reply

@nprofile…vwag Yes, this seems to be safe. Almost all system fonts have visible glyphs for these formatting control characters (not always consistent in design, but that”s another story) and text layout engines and libraries reliably supress their display. I am not entirely sure where in the text stack the code for supressing display of formatting control glyphs lives, but I think it is not generally something for which individual apps are responsible.

2026-06-15 17:13:35 UTC

- reply

@nprofile…yk7r ZWJ and ZWNJ are present on most Indic keyboard layouts, but in inconsistent locations across platforms, even on what are nominally the same layout. And they’re not made visible in on-screen keyboad tools, although they could be since the fonts used to populate those tools include visible control character glyphs. Unhelpfully, on-screen keyboard displays tend to treat the key cells as if they were output text, suppresssing display of the control characters.
Banging on about this:
https://www.youtube.com/watch?v=mUbK8BUQZQg

2026-06-15 17:07:12 UTC

- reply

@nprofile…vwag Yes. Word processing and page layout apps usually offer an editing mode that exposes text formatting, e.g. paragraph beginnings, different kinds of line-breaks, and also control characters like ZWJ, ZWNJ, and directional controls. Display of these characters may depend on the font, which is why visible glyphs are provided. The display of these is suppressed in normal text output, and only shown in these editing modes.

2026-06-15 15:02:10 UTC

Wishing there were a resource that documented the location of ZWJ and ZWNJ on every keyboard layout in which they occur, searchable by language and platform.