Documentation: Explain we only support current version & push updates

[davelab6]

Per JulietaUla/Montserrat#60 (comment) and #1307 we should update our about page (and FAQ) to tell users that

1. we push family updates that improve the quality of the fonts but may change their visual design in subtle ways in ways you may not expect;
2. while an old version's files may be available if accessed directly from the gstatic CDN, they should not be directly linked to or relied upon, because they could disappear at any time - only the CSS API is supported; and
3. if people really care about controlling updates, they should self-host; likely with something like an NPM package, which gives finer version control and can track the actual font files' internal versioning scheme.

[ctolkien]

but may change their visual design in subtle ways;

In major ways:

Also ref: JulietaUla/Montserrat#39

[danyj]

If this is subtle JulietaUla/Montserrat#60 (comment) than we all should quit our jobs and hire @davelab6 for our projects.

[DanielMcPhee]

@ctolkien Now there's a smoking gun if ever I've seen one...

[esistgut]

What happens when the changes are not "subtle"? Who does decide when a change must be labeled as "subtle" or "breaking"? Do every single designer who publish fonts on Google Fonts has this power to break every single site linking to his/her font?

if people really care about controlling updates, they should self-host.

Can you please elaborate on when people are not supposed to care about major changes like the Montserrat update?

[alexpana]

What happens when the changes are not "subtle"? Who does decide when a change must be labeled as "subtle" or "breaking"?

Apparently, it's @davelab6 and he doesn't give a shit. You'll just have to self-host.

[davelab6]

What happens when the changes are not "subtle"?

Then the project is "forked" into a separate family. Eg https://fonts.google.com/selection?selection.family=Exo|Exo+2|Shadows+Into+Light|Shadows+Into+Light+Two

Who does decide when a change must be labeled as "subtle" or "breaking"?

👋

I've updated the above issue to make it clearer; "may change their visual design in subtle ways in ways" is now "in ways you may not expect."

Do every single designer who publish fonts on Google Fonts has this power to break every single site linking to his/her font?

No; if the upstream designer offers an update that the Google Fonts team does not want to onboard for any reason, then Google Fonts won't pick it up.

Relatively few of the fonts available in Google Fonts are actively developed by their designers; most of the projects only advance when the Google Fonts team commissions improvements.

if people really care about controlling updates, they should self-host.

Can you please elaborate on when people are not supposed to care about major changes like the Montserrat update?

The amount people care is related to how much they believe a change is major.

I personally do not think the change here is "major"; I believe the typeface design is recognizably the same between Montserrat v6 and v7, while the typeface designs Exo and Exo 2 are recognizably different.

This is a judgement call, and it is literally my job to make apply my judgement to these decisions. We have different opinions about what is subtle and what is a major redesign (that legitimates publishing as a second family.) It is a subjective judgement; in my opinion, this update was a wonderful improvement to the family. Making any improvement necessarily means changing something. If I had published v6 as "Montserrat 2" (and "Montserrat Alternates 2") then we'd have had 4 families, and if I had then published v7 as "Montserrat 3" then we'd have 6 families... all versions of Montserrat, all varying in ways that non-designers can't tell apart.

I appreciate that professional designers can tell them apart, and that the differences can be meaningful (like if your site is about selling sports shoes vs lingerie.) I appreciate that some agencies, their clients, or other kinds of users won't have time or budget to update their CSS from 400 to 500, and their visual design has "changed out from under" them.

But, I believe most users won't even notice anything changed, except that their text now reads more comfortably. Google Fonts is for everybody, not only professional designers.

And I have an eye on the long term. The default (400 roman) style of Montserrat "should" be the weight that it is now, and for the last 6 years it has been "off." It is used on X-many sites today, but it will be on X+Y-many sites in the future, and those all sites should have the best quality version of the typeface. Now they do.

Stepping back, I think that the visual/graphic design culture is rooted in print media (media in the Marshall McLuhan sense); in print media, it is physically impossible to change the typeface after the document is printed. When reprinting something, it is very expensive to introduce a newer font to an older documents, such that introducing changes to the typeface causes reflow (especially to make longer text runs.) Print page layouts are brittle, and checking and redesigning everything will be expensive... so it just doesn't happen.

Whereas, in digital media, documents are abstracted from print, and in hyper-text, decomposed to separate style from content. This breaks many of the almost-unconscious assumptions that design culture inherits as a legacy from print media. The whole "web standards movement," the CSS Zen Garden, Adobe Flash vs Responsive HTML5, was part of this; print culture was reflected in the Flash's absolute pixel positioning, which many designers with a background in print found comforting, and correspondingly the heterogenous rendering involved in web standards was discomforting. Letting the browsers' chips fall where they may.

In digital media, it is possible for hyperlinked components of a web document to change "out from under" the document; the positive power of this was popularized by CSS Zen Garden. This contradicts print design cultural assumptions about what is possible; it is not "normal" in print culture, for sure. But web culture is fundamentally about isolating "style" and tolerating page-layouts changing, and building in affordances.

As a concrete example of when people have not cared about "major" changes like the Montserrat update, on Tuesday we also pushed what I think is a "subtle" update to Merriweather - used on 3M sites - where the typeface was also completely redrawn and updated. I haven't heard a single thing about it from anyone, on Github or Twitter. The redesign causes text reflow to about the same degree as Montserrat, but in the other direction:

IE 11 Rendering

The change in the weight of the Regular is much less, of course; but keep in mind that we updated 25 families on Tuesday. The only family I’ve seen comments on is EB Garamond, where we addressed a common complaint about old style numerals being default, and pushed an update that makes them lining numerals and places the old style ones in a stylistic set OpenType feature.

Last year we updated Josefin Sans, while Josefin Slab remains untouched. Initially they had the same stroke weights in each style, and today you can compare the weight difference in the two Regulars - there the problem was the other way, that the Josefin Slab regular is much too light for a Regular. At the same time, also commissioned a 100% redraw of the Pacifico typeface. When we went live, not a peep.

If you want to use Google Fonts but anticipate our changes, you can watch this Github repo and see when pull requests are merged. I tend to leave font update PRs open for a while to get comments, and there tends to be a delay between the PR merge and the actual push to the production Fonts API.

Ultimately for me, the fonts are available under licenses that respect your freedom to go your own way. Part of why I got into the fonts business is that I was dissatisfied about using fonts that (a) had problems and (b) the designers said they would never fix those problems, and there was nothing I could do about it. "License says no." Here, the license says yes :)

I've discussed the various proposals this week for versioning with the whole GF team, and currently we don’t plan to offer any reliable access to previous versions, or opt-in features for version pinning, etc. We want everyone on the latest fonts.

When @serphen says that it is possible to access specific older versions, this is not a good idea... You might be able to access those versions today, but this is unsupported and may completely break - as in, those gstatic URLs may begin 404ing at any time.

Ultimately: People choose to rely on Google Fonts to take care of all the minutia of web fonts, and we believe auto-upgrading all users to the latest versions, that have improved quality and bug fixes, is the responsible thing to do.

[laerm0]

Well put, @davelab6.