I made an italic font without permission

J. Bridges

I took a free Google font and made an ugly italic as my client requested to be used in a PowerPoint. I used Glyphs 3. Its ugly but the client is happy. Should I be arrested? Or turn myself in? 

John Hudson

If it is something you found on Google Fonts, then it is released under the Open Font License (or very occasionally the Apache 2.0 license), so you didn’t make the italic without permission: permission is explicitly granted in the license.

J. Bridges

Thanks John. They are ugly. But they wanted them. 

JoyceKetterer

@James Bridges, to clarify what @John Hudson said:  You can always add to a Google font so long as you change the name.  Your Italic needs a different name from the original.  I think you could also use that same new name for an unchanged version of the upright to be a companion, but I'm not sure. @Dave Crossland?

John Hudson

Joyce, are you referring to OFL ‘Reserved Font Name’? I think only about half the fonts on Google Fonts have reserved names stated, and as I recall GF’s current preference is for fonts not to have a reserved name. When the STI Pub consortium representative was in talks with @Dave Crossland about putting the STIX Two fonts into GF, there was a specific request to remove the reserved name statement.

J. Bridges

That is what I did. Thanks. In the end I convinced them to use an existing serif italic which was more interesting looking and easier to read than a fake roman san serif italic. 

JoyceKetterer

@John Hudson, really?  I thought the primary purpose of the reserved font name clause was to prevent caching issues.  It's a problem for end users more than for designers.  It seems like a bad idea to me to not require name changes.  

jeremy tribby

I think you are both right, joyce and john. avoiding naming collisions with web fonts is one of the stated reasons that reserved font names are included in the OFL. and, as far as I remember, google prefers you not use reserved names.

due to the implementation of cache partitioning in browsers a couple of years ago, there is now less if any value for that particular issue, because fonts are cached per domain that uses the font, not per CDN that serves the font. so users would only be avoiding collisions with themselves. (I think)

dave would know better than me of course but I'd guess the reason google prefers you not use them is unrelated, because the suggestion is older than that - in practice I have really only seen reserved font names introduce confusion about IP/trademarks in what is otherwise a fairly straightforward copyleft license

Peter Baker

My understanding (not sure I remember it exactly) is that the presence of a Reserved Font Name would require that the name be changed if the font was subsetted for a PDF or a web page--subsetting being a kind of alteration. Since GF does a lot of subsetting when serving fonts . . . you see the difficulty. In an operation like GF, it's not desirable to serve fonts with the names obfuscated.

Daniel Benjamin Miller

Peter Baker said:

My understanding (not sure I remember it exactly) is that the presence of a Reserved Font Name would require that the name be changed if the font was subsetted for a PDF or a web page--subsetting being a kind of alteration. Since GF does a lot of subsetting when serving fonts . . . you see the difficulty. In an operation like GF, it's not desirable to serve fonts with the names obfuscated.

I thought that the name only needed to be changed if the font were being distributed as a standalone font file, as opposed to embedded in a PDF or the like (and potentially as a subset). This was the same reason why you could embed the font in a PDF without having to include a full copy of the license in your document — or maybe I have a misunderstanding of the OFL on this point.

JoyceKetterer said:

@John Hudson, really?  I thought the primary purpose of the reserved font name clause was to prevent caching issues.  It's a problem for end users more than for designers.  It seems like a bad idea to me to not require name changes.  

Maybe, but I think it would be better to expect this practically than to force it legally. I think about the idea in the open-source world of someone picking up a project that has been abandoned — I'd rather the new maintainer have the ability to keep the original name (although I guess oftentimes this doesn't happen anyway). Some people on LWN have discussed the practicalities of this clause: https://lwn.net/Articles/552178/.

Peter Baker

Daniel Benjamin Miller said:

I thought that the name only needed to be changed if the font were being distributed as a standalone font file, as opposed to embedded in a PDF or the like (and potentially as a subset). This was the same reason why you could embed the font in a PDF without having to include a full copy of the license in your document — or maybe I have a misunderstanding of the OFL on this point.

Here are a few relevant links, making clear that subsetting (at least as currently done) does require that the font's name be changed when a reserved font name (RFN) has been declared.

1. SIL's FAQ, addressing the question "Is subsetting a web font considered modification?" (Answer: yes; can't use RFN).

2. SIL's page on Web Fonts and Reserved Font Names (see especially the sections on "Importance of Reserved Font Names" and "Subsetting").

3. A (closed) Google Fonts issue addressing the problem of RFN and subsetting.

I wish Dave Crossland would weigh in on this, but I believe this (and possibly also issues around the woff format) is the main reason Google Fonts doesn't allow RFNs.

Dave Crossland

Personal opinion, not the view of my employer, Google:

The OFL RFN makes sense to, and is used extensively by, the license steward, SIL. But personally I find the rfn clause distasteful; it uses the strong arm of copyright to control a softer issue of naming, and the legal systems of the world have trademark law for controlling that, which I think is much more appropriate. 

For instance, when the OFL was drafted, pre web fonts and when all computers used to do work still came as standard with portable disk drives, the rate at which users and other font designers would make modifications to fonts of any kind, was massively reduced compared to today.

Today, as mentioned, web fonts usage commonly involves subsetting modification, or using compression formats which inherently modify the font data; and collaborative development platforms like GitHub have bazillions of users.

These new situations make the RFN very cumbersome in practice, in a way a Registered Trademark is not. 

Eg, most people running rfn projects on GitHub don't set up the dev version with a codename, but retain the reserved name. Then, if you are forking the project to make a contribution, but not making the very first edit to change the name as presented to users, you may likely terminate the license - and create havoc for everyone touching those files after that point. 

If you do know/remember/bother to do so, then making the upstream pull request go to a dev branch so the upstream can merge there and then put the name back before merging to their main branch, is extra friction in a process that a huge amount of effort has been invested in making less... expletive.

In web fonts, if all a user's modifications amount to are Woff2'ing and subsetting, generally I believe font designers are okay with that - and even go so far as to be disappointed to hear they've required the user to rename the font, because they want their work to be known as the family name they chose, unless the visual typeface design is noticably different. 

That's different to what I understand is the SIL motivation, which is that for example they don't want to deal with the support costs and burden of users contacting them about "SIL's" fonts which infact are not the bits SIL put out; for example, if SIL sets the overlap bits correctly in the glyf table, and a user WOFF2's the file and deleted the overlap bits, and then outlined text in that modified font shows overlaps, SIL doesn't want to pay the costs of figuring out the user caused their own problem. I believe majority of type designers don't pay those costs by ignoring the user, or not being fussed about this kind of thing.

If type designers have registered the family name as a trademark, in every scenario where they would be being proactive and relying on the RFN for leverage, I personally think a ® will add just as much leverage. But with more flexibility and no risk of termination.

Google Fonts docs explain that a rfn is dispreferred, but if a ofl copyright author insists, Google will carry the costs of obtaining additional permission, as the OFL suggests. Personally I also find the privilege of those who obtain such permission also distasteful. What about downstream recipients from such privileged third parties? 

Ideally I suppose there ought to be some independent clearing house for requesting and offering and cataloging such permission.... Like the public trademark office databases. But like so many minutia around this stuff, someone ought to do something