A couple weeks ago, a customer reported that my fonts didn’t work in Microsoft Office. I looked up his order. He had bought the fonts in 2015. I asked: have they not worked for five years? Oh no, he said, they were working fine until September 28.
So what happened on September 28?
In principle, software is boundlessly flexible—just edit the source code and recompile. But for much of its early history, in practice it was not. Why? Because distributing software still meant manufacturing some kind of physical object as its container—like a floppy disk or device or cartridge or CD-ROM or even a book. Thus, programmers had an extremely strong incentive to get things right the first time. In most cases, there would be no second chance.
The internet, of course, changed all that. The first commercial program I remember being distributed exclusively online was the Netscape Navigator web browser in 1994. Another novelty of Navigator: it was the first time “beta” software—formerly a private testing phase—was made available to the public. Netscape had quickly figured out that because they could issue updates at no cost, there was no longer any need to put as much time into polishing the software. Ship it now; fix it later.
Since then, we’ve climbed the ladder to what is today monumentally marketed as the cloud. But really, it’s the same idea as in 1994: why finish today what you can put off till tomorrow? In recent years, Adobe and Microsoft, who make typesetting software used by almost all my customers, have converted their flagship software packages into cloud offerings.
I could list numerous reasons why this trend is dispiriting. Here’s the simplest: having unlimited flexibility in issuing software updates has made these behemoth software companies lazier than ever, and their software less reliable.
The need for continual updates is often framed in terms of “security”—the idea, I guess, is that some conjectural horde of hackers will take your computer hostage unless you continue to pay for updates. But consistent with the pattern of any protection racket, the most likely risks to the stability of your computing experience arguably come from Adobe and Microsoft themselves.
Before the current era, it used to be that Adobe & Microsoft would release software with bugs. But major releases were far apart, so the bugs would persist for a long time. As a solo font vendor, I could devise workarounds. Everyone could get back to doing their jobs and ignoring the software.
Cloud updates, however, have changed the rhythm. In principle, bugs can be fixed faster than before. In practice, new bugs are also introduced faster. For instance, a year ago I got reports from a few customers that the capital I in my fonts had started disappearing from PDFs generated with Adobe software (?!?!) After a long conversation with a skeptical Adobe engineer, I produced enough evidence for him to sheepishly concede that yep, it was an Adobe bug.
As for my customer who started having problems on September 28? I theorized that there was probably a buggy Microsoft Office update that went out to his computer that day. (True.) And that rolling back to the previous version would cure the problem. (Also true.)
But as a software vendor who wants customers to be minimally inconvenienced by software chores, these workarounds are unsatisfying. If these companies push out a cloud update that suddenly breaks my fonts, customers contact me first. At that point, I have no good advice except “wait for the next update” (until then, things are still broken) or “roll back to the previous update” (but this can be laborious).
So to those who subscribe to cloud services, a gentle word of caution. I’m not suggesting you should forgo updates to make my life easier. Just be aware of the alternate point of view: why fix what isn’t broken? Personally, I turn off all automatic updates. If I rely on a certain program for my work, I update it gingerly and only when necessary. Ultimately, cloud updates are cheap only if your time has no value.