Variable fonts, axes values

KP Mawhood

@Simon Cozens This is a bad analogy given that LEGO have made custom bricks, and the systems (1955) with those custom bricks played a large role in their success.

In fact the first lego bricks (1949) did not stick well together and so play was limited, impacting conversion/sales. The success of these products relies on the combination of design and technology. It's not a competition between the two. 

With sales picking up around Christmas and plummeting the rest of the year, Godfred felt their LEGO toys were missing something essential.

He realized they were giving kids ready-made toys which were not challenging enough. The toys needed an idea and a system built around it. This way children were able to use their imagination and creativity.

This is the second breakthrough moment in LEGO history. Instead of offering kids ready-made toys, LEGO gives them the opportunity to build their toys — a much more challenging activity which kept kids engaged for hours."

https://brand-minds.medium.com/legos-success-story-3-defining-moments-d17f5ddb5380

Simon Cozens

This is, of course, the problem with analogies: people pick at the analogies, and not the point you're trying to make... :-) The point is where you apply limitations and simplifications. If you put them on the technology, you restrict its expressiveness. If you put them on the user interface, the underlying technology is still expressive but the user is guided as to how best to use it.

It's like Unix: Unix was not designed to stop you from doing stupid things, because that would also stop you from doing clever things.

KP Mawhood

@Simon Cozens Agreed, it can be a problem with analogies – but they're also a useful tool. I enjoy debate, and analogies are real asset in provoking positive feedback (aha moments!). 

Simon Cozens said:

It's like Unix: Unix was not designed to stop you from doing stupid things, because that would also stop you from doing clever things.

I grew up on Risc OS (ARM chip, check the history) and also had my time with Ubuntu. Now running the BSD-derived Apple (with BSD derived from Unix) which can definitely be a barrier to stop you from doing clever things… if only because it's closed source, etc.

Of course it depends what you want to do. Whilst we could all work in the command line, it has a steep learning curve… and most people don't know how to use it.

Simon Cozens said:

The point is where you apply limitations and simplifications. If you put them on the technology, you restrict its expressiveness. If you put them on the user interface, the underlying technology is still expressive but the user is guided as to how best to use it.

I support the idea of catering to different skill levels. But, first let's address this concern around limitations and simplifications. I'm not advocating for limits on the user (at least not the advanced user)… what do you think is being suggested?

John Hudson

If you offer users the two options, showing variable fonts with the steps model and the current model, which one would they use the most?

If you offer users two options, they are almost certain not to think about all the other possible options. 

There are so many interesting and useful ways in which interaction with variable fonts could be managed, all on top of the existing data structures.

John Hudson

Katy, 

On the subject of the target audience/market for variable fonts: this is something I have thought about a lot. In March 2020, I was supposed to be participating in a Granshan Type Tech panel on marketing variable fonts, and before it was cancelled due to Covid—before people got used to organising and attending video conference events— I started looking for examples of how the original Linotype machine was marketed and how phototypesetting systems were marketed. Because that’s what needs to be marketed at this stage—a new technology—, which is something that—with the limited and not far from entirely successful experience of marketing of OpenType Layout features in fonts in the early 2000s—font foundries have not really had to do since the 1980s. The technologies have been developed, promoted, and marketed by software companies, and font foundries have mostly only needed to market the fonts, not the technologies to use the fonts.

Variable font technology is a little different than even OpenType Layout in this respect, because the benefits of the technology to the big software companies is mostly that of data compression—that is how the technology was sold to the managers who signed off on the specificaton and development work needed to develop and support OT variations—, so those companies—with the exception of the Google Fonts team—mostly do not see a need to market the technology to end users, or even to do anything particularly interesting or innovative with it in terms of typeface design and typography. It is enough for them that variable fonts take up less disc space and can download faster.*

This puts the onus on marketing OT variations technology on the people making the fonts, and the onus on developing ways for users to interact with the technology on Web standards and on the makers of individual design apps, rather than e.g. at the operating system level from whence UI and UX can filter down to any software using system APIs for typography features.

What concerns me about some of the comments in this thread is that there seems to be a suggestion that there might be some magic bullets that we could load into fonts, which would significantly alter the challenge of marketing variable fonts within this situation: that tweaking the data structure of the OT variations tables in the hopes that this would encourage different kinds of UI/UX design might make it easier to persuade people to license and use variable fonts. Not only do I not think that the specific change proposed in terms of axis step data belongs at the font data level—since it a) it doesn’t need to be at that level in order to be made available to users in a UI, and b) putting it at that level makes it more restrictive and less flexible than it would be at the UI level—, I also do not think it will change the situation. Marketing technology is hard, especially for people and companies that do not have experience doing it.

_____

* This is also why achieving further development of OT variations technology has been an uphill battle since 2016, with almost nothing being done despite lots of good ideas with very obvious benefits to both makers and users of variable fonts. There are lots of things I want to be able to do in variable fonts that I cannot, because the 2016 iteration. of the technology has holes and limitations. Plans to fill those holes and remove those limitations are advanced, but without buy-in from the stakeholders to actually implement them in software they won’t get formally spec’d.

John Hudson

with the limited and not far from entirely successful experience of marketing of OpenType Layout features in fonts in the early 2000s

Typo: I meant ‘with the limited and far from entirely successful...’

Twenty years on, there are still plenty of users who have zero engagement with OpenType Layout features, and plenty of design schools whose typography programs fail to teach anything about them.

KP Mawhood

John Hudson (edited for John's typo) said:

In March 2020, I was supposed to be participating in a Granshan Type Tech panel on marketing variable fonts, and before it was cancelled due to Covid—before people got used to organising and attending video conference events— I started looking for examples of how the original Linotype machine was marketed and how phototypesetting systems were marketed. Because that’s what needs to be marketed at this stage—a new technology—, which is something that—with the limited and far from entirely successful experience of marketing of OpenType Layout features in fonts in the early 2000s—font foundries have not really had to do since the 1980s. The technologies have been developed, promoted, and marketed by software companies, and font foundries have mostly only needed to market the fonts, not the technologies to use the fonts.

That would be a very interesting panel discussion. Are there any plans to revisit it?

John Hudson said:

There are lots of things I want to be able to do in variable fonts that I cannot, because the 2016 iteration. of the technology has holes and limitations. Plans to fill those holes and remove those limitations are advanced, but without buy-in from the stakeholders to actually implement them in software they won’t get formally spec’d.

That's difficult. 

John Hudson said:

This puts the onus on marketing OT variations technology on the people making the fonts, and the onus on developing ways for users to interact with the technology on Web standards and on the makers of individual design apps, rather than e.g. at the operating system level from whence UI and UX can filter down to any software using system APIs for typography features.

Understood.

John Hudson said (added my own bold):

What concerns me about some of the comments in this thread is that there seems to be a suggestion that there might be some magic bullets that we could load into fonts, which would significantly alter the challenge of marketing variable fonts within this situation: that tweaking the data structure of the OT variations tables in the hopes that this would encourage different kinds of UI/UX design might make it easier to persuade people to license and use variable fonts. 

Let's try to untangle this section. Working with experienced marketing teams, I acknowledge the challenges of this specialism. Marketing is hard, you're right.

"Magic bullets that… significantly alter the challenge of marketing" would require marketing to be centred in the conversation. Instead, contributors are brainstorming product ideas and receiving feedback on those ideas. Rather than seeking to persuade potential licensees, these ideas seek to improve the user experience.

John Hudson said:

Not only do I not think that the specific change proposed in terms of axis step data belongs at the font data level—since it a) it doesn’t need to be at that level in order to be made available to users in a UI, and b) putting it at that level makes it more restrictive and less flexible than it would be at the UI level—

Ok, it's time to press pause on the step axis.

John Hudson said:

Marketing technology is hard, especially for people and companies that do not have experience doing it.

Is marketing the scape goat of variable fonts? I feel bad for marketing.

John Hudson said:

I also do not think it will change the situation.

Would it help to talk about that?

Thomas Phinney

KP Mawhood said:

John Hudson (edited for John's typo) said:

In March 2020, I was supposed to be participating in a Granshan Type Tech panel on marketing variable fonts, and before it was cancelled due to Covid—before people got used to organising and attending video conference events— I started looking for examples of how the original Linotype machine was marketed and how phototypesetting systems were marketed. Because that’s what needs to be marketed at this stage—a new technology—, which is something that—with the limited and far from entirely successful experience of marketing of OpenType Layout features in fonts in the early 2000s—font foundries have not really had to do since the 1980s. The technologies have been developed, promoted, and marketed by software companies, and font foundries have mostly only needed to market the fonts, not the technologies to use the fonts.

That would be a very interesting panel discussion. Are there any plans to revisit it?

I should point out that there is an ATypI Tech Talks event just announced, for Sept 1–3. Just started a thread for it: https://typedrawers.com/discussion/4440/atypi-tech-talks

Dave Crossland

Personal opinion only, not the views of my employer, Google

John Hudson said:

Variable font technology is a little different than even OpenType Layout in this respect, because the benefits of the technology to the big software companies is mostly that of data compression—that is how the technology was sold to the managers who signed off on the specificaton and development work needed to develop and support OT variations—, so those companies—with the exception of the Google Fonts team—mostly do not see a need to market the technology to end users, or even to do anything particularly interesting or innovative with it in terms of typeface design and typography. It is enough for them that variable fonts take up less disc space and can download faster.

Indeed. The Google Fonts team was barred by Microsoft from participating in the specification and development work on OpenType 1.8 in 2016 - since the Noto team (which was entirely separate, back then) initiated the work, Google as a company was participating, and the secret cabal mailing list was closed to new individuals.

This is also why achieving further development of OT variations technology has been an uphill battle since 2016, with almost nothing being done despite lots of good ideas with very obvious benefits to both makers and users of variable fonts. There are lots of things I want to be able to do in variable fonts that I cannot, because the 2016 iteration. of the technology has holes and limitations. Plans to fill those holes and remove those limitations are advanced, but without buy-in from the stakeholders to actually implement them in software they won’t get formally spec’d.

A major set of stakeholders are the public projects that implement font technology in libre software, which is not subject to whims of typographically disinterested corporate management. 

Eg https://github.com/be-fonts/boring-expansion-spec

There is also a public spec process at ISO (https://github.com/MPEGGroup/OpenFontFormat) so that once libre demonstrations are available, they will be formally spec'd.

KP Mawhood

Dave Crossland said:

A major set of stakeholders are the public projects that implement font technology in libre software, which is not subject to whims of typographically disinterested corporate management. 

Does it matter if corporate management is typographically disinterested? 

I say this from a perspective of *not* securing buy-in when wrapped up in a passion pitch. In contrast, I did secure a risk owner for font licensing (no one wanted it, passed from between two parties for months) and rollout a global font policy (endless approvals, etc). 

The passion pitch: I couldn't convince corporate management that left-justified letter-spaced (with actual spacing) Urdu* would reflect poorly on a highly publicised initiative to support underprivileged language communities. The product team was interested, but forced to roll out.. something like one language each month? No exceptions for more complex implementation. They had no time. 

I tried to continue this through to a change proposal (internationalisation at scale) with plenty of guidance from higher-ups, but that was also stopped short.

*The Urdu also used Naskh, rather than Nastaliq. It's good to see this highlighted again by this year's Granshan conference. 

https://signs-of-the-times.net/speaker/abeera-kamran/

John Hudson said:

It is enough for them that variable fonts take up less disc space and can download faster… this is also why achieving further development of OT variations technology has been an uphill battle since 2016, with almost nothing being done despite lots of good ideas with very obvious benefits to both makers and users of variable fonts.

I don't really understand how implementation is not valued. Without sufficient user conversions (both type maker and type user), who gains from the technology benefits of data compression. What is the perspective of managers at the big software companies… is there a misunderstanding somewhere?

Dave Crossland

KP Mawhood said:

Dave Crossland said:

A major set of stakeholders are the public projects that implement font technology in libre software, which is not subject to whims of typographically disinterested corporate management. 

Does it matter if corporate management is typographically disinterested? 

It does matter if you want their resource allocation decisions to benefit you - as your case study in Urdu makes clear, so I'm a bit confused what you are really asking here  

John Hudson said:

It is enough for them that variable fonts take up less disc space and can download faster… this is also why achieving further development of OT variations technology has been an uphill battle since 2016, with almost nothing being done despite lots of good ideas with very obvious benefits to both makers and users of variable fonts.

I don't really understand how implementation is not valued. Without sufficient user conversions (both type maker and type user), who gains from the technology benefits of data compression. What is the perspective of managers at the big software companies… is there a misunderstanding somewhere?

I think its useful to see VF implementation through a "stepped levels" model: 

Level 1: Variable fonts start out as not supported at all, and when loaded, only the default style can be accessed as a fallback. This scenario is now very rare, and usually only occurs because of a problem with the font file itself, so that it appears to apps like a single, non-variable font style.

Level 2: Variable fonts are supported only as containers for static fonts. In this case, users can not set any axis value, but can access the named instances. Current examples include Apple’s Keynote, Numbers, and Pages apps, and Microsoft’s Word and Powerpoint; and operating system level support, which allows UIs to specify specific instances and load them from a VF instead of a set of static fonts. Importantly, this requires zero UX changes; the mental model of fonts is completely unchanged. It is the pure "to compress" benefit, only.

Level 3: Variable fonts are supported with direct control over axes (usually shown as sliders and number inputs). Many design apps now allow for this, including Figma, Sketch, and Adobe’s Photoshop, InDesign, and Illustrator apps. v-fonts.com/support actively maintains a list of supporting software. This is the "to express" benefit. It breaks the mental model of static fonts for many users, and a common response from many print-orientated graphic designers is "why would I want this? It seems like a useless toy"; but new-media orientated designers (especially those dealing with motion graphics, or multi-modal design) tend to "get it".

Level 4: Variable fonts are supported with full integration into the typography engine. For example, the Optical Size (opsz) axis is applied automatically based on the font size in points (InDesign good, browsers bad), and with a way to change the ratio (not seen this yet); or, when adding a stroke to text, the app correctly applies the stroke only to only the outer shape of each glyph - and not the inner contours used to construct it (Photoshop and InDesign good, browsers, Illustrator and XD and Figma bad). Most importantly, within an app’s scripting or type preferences, users can define contextual rules to apply axes values to character or paragraph styles, rather than only setting axes directly in a GUI. For example, set the Grade axis value based on light mode or dark mode, which is easy in CSS with `@media (prefers-color-scheme: dark) { ... }`. This is the "to finesse" benefit. Only savvy typographers who are familiar with fine tuning their typography "get it"; for example, those who know how to set letter-spacing and font-weight to approximate optical size specific designs with static font families.

Since the "to compress" benefit is straightforward to quantify - take 2.0Mb of X static font family data and replace it with 200kb of X VF family data, or whatever - and it doesn't require UX changes, it has been easy to get software engineering team managers to sign off on building Level 2 support.

But building Level 3 or 4 support requires a certain level of inspiration and interest in supporting the latent unmet needs of sophisticated users - qualitative user testing of mass users will likely reveal a (large) "who cares" attitude. 

KP Mawhood

Dave Crossland said:

KP Mawhood said:

Dave Crossland said:

A major set of stakeholders are the public projects that implement font technology in libre software, which is not subject to whims of typographically disinterested corporate management. 

Does it matter if corporate management is typographically disinterested? 

It does matter if you want their resource allocation decisions to benefit you - as your case study in Urdu makes clear, so I'm a bit confused what you are really asking here  

Sorry for the confusion. Let's try again – I messed up the Urdu because I didn't reframe in terms relevant to corporate management. No reason for them to care about typographic practice, but rather the value to the business.

I hadn't learnt the right vernacular… and I had not understood the company’s strategy and how to effectively partner with that strategy.

Since the "to compress" benefit is straightforward to quantify - take 2.0Mb of X static font family data and replace it with 200kb of X VF family data, or whatever - and it doesn't require UX changes, it has been easy to get software engineering team managers to sign off on building Level 2 support.

But building Level 3 or 4 support requires a certain level of inspiration and interest in supporting the latent unmet needs of sophisticated users - qualitative user testing of mass users will likely reveal a (large) "who cares" attitude. 

Thanks for taking the time to articulate the four levels.

Font licensing is also straightforward to quantify – in terms of elevated risk conditions as well as the long-term cost of those risks. As it did with CTOs, the heightened attention to risk management has broadened the role of general counsels over time. In licensing and technology, we often have a seat at the table – it's easy to achieve sign off.

But building Level 3 or 4 support requires a certain level of inspiration and interest in supporting the latent unmet needs of sophisticated users - qualitative user testing of mass users will likely reveal a (large) "who cares" attitude. 

Very true. Equally, who cares about data savings on a font file, or reading a license agreement? It's not the masses. Who has the inspiration or interest on C-suite teams… Figma is young enough to have its founders (and original vision) at the helm.

When invested in user research, user experience can be straightforward to quantify. It similarly has a problem with maturity levels, as a barrier to implementation.

https://www.nngroup.com/articles/ux-maturity-model/