Nightly’s on First; Aurora’s on Second
I wasn’t going to even bother saying anything, but raccettura’s post goaded me into it.
 Let’s put this in some perspective: Apple—user experience and design queen Apple—is to the right of Mozilla’s position on this issue! |
Allow me to succinctly cut through all the cacophony on this: version numbers matter1.
They’ve always mattered. And they will continue to matter.
Why?
Because the first question anyone asks when debugging their own software problem or helping someone else: “What version are you using?”
“Latest” (or the even more asinine “infinite!”) is not even in the universe of a useful answer (and there exist a universe of reasons why it may not even be an accurate one, either).
Version numbers help users find the appropriate release notes for the application sitting in front of them.
Version numbers help users search forums and knowledge bases, and be able to correlate others experiencing a similar problem and use the correct instructions for the version they have.
Version numbers make it possible for savvier users to help themselves, by knowing what to put into the bug database when they search.
Version numbers make it possible for users to help developers, being able to effectively communicate when reporting a bug.
Version numbers are what the press refers to, so they can communicate which feature set their review pertains to.
Version numbers are used by organizations and startups to drive PR and user buzz about new features and product families: “We just released version 2.0! You should really check it out!”
Software labeling—version numbers, as consumers know it—has been a core of communication involving the answer to “Which actual bits are running on my CPU?” since… well the beginning of packaged-and-sold software.
Anyone quoting3 “Nobody asks ‘Which version of Facebook are you running?’” is easily dismissible: it’s a great example of the fallacy of false comparison, but it’s not worth responding to4,5.
In Mozilla’s specific case, this versioning consternation induced by their “rapid release” schedule is a direct result of versioning-as-a-concept being a proxy for something—API compatibility—in the Mozilla platform itself.
It’s important to call this out, because it’s necessary to distinguish between “rapid release causes my extensions to break every six weeks”—to users’ annoyance—and the (non-response) response of “Well, version numbers don’t matter anymore.”6
I see the move to obscure the version number to really be about reducing user “upgrade friction” by making the process as opaque and secretive7 as possible. This makes it easier to use Firefox’s user base as a lever to “move the web,” an argument I’ve previously made. This move is a consistent and logical step toward doing just that.
An interesting thought exercise to consider: what would the response be if Microsoft started updating Visual Studio on random intervals, and refused to divulge the version number installed on your machine when you tried to find it. Given that Mozilla bills the Open Web as a platform, and Firefox the tool to interact with and develop against that platform, it’s not a dissimilar example.
In any event, simply saying version numbers don’t matter anymore, and repeating it over and over like a bird you’d feed crackers to, does not suddenly make it a fact of modern software development.
It illustrates a profound lack of understanding of versioning’s (important) function in the best case.
And disingenuousness in the worst.
_______________
1 You’d probably expect a release engineer to say that, given that version numbers and versioning as a process development area has always been a large part of my job2
2 But that’s not why I’m saying it, entirely anyway
3 I should say “repeating the sentiment,” because I can’t find any Google record of anyone (publicly) having said that; the only reference is back to Laura Thomson’s original post, where the quotation is uncited
4 But I will: software which I download and run on my computer is very different than a web application
5 Facebook, Google Apps, and lots of other websites deploy software that’s broken all the time; people generally don’t have to answer “What version is it?” because, there often isn’t a way to meaningfully interact with those companies about issues in their (web-based) applications; you just wait until it’s fixed
6 In fact, this entire discussion about removing the version number from the UI is probably more acrimonious than it otherwise would be, because Mozilla has not, to date, resolved the issues raised by those negatively impacted by their “rapid [fire] release” decision
7 User experience people would probably choose “unobtrusive”
Amen, brother. Version numbers also indicate that the user/administrator has a choice about what software is available on the machine. In the Cloud Future, such a thing is anathema. Life imitates Neal Stephenson’s art, evidently.
1. I wasn’t goaded you… not intentionally at least.
2. I agree. I admit version numbers are fugly and evil. But they are necessary. Simply put, there is no alternative. Web Apps you run what they give you. Client side apps are user controlled. As a result version discrepancies happen for various reasons.
While it may be, and probably is, me with the problem, I also find it very difficult to know what’s being discussed when everyone is using cutesy names. At least Microsoft and Apple are doing branding with “Vista” and “Lion” (whether that’s an excuse or not…), but I have trouble knowing what’s being discussed when everything goes by the names of toys in movies or Suicidal Salamander.
More on point, Mozilla seems to be digging in its heels and behaving badly, if not at this point, surprisingly. It’s this bad behavior more than anything that’s going to bite them on the ass, because when you have nowhere to go, but you’re feeling abused, you leave anyway, and usually with friends and family in tow (as a wise man said recently).
I wonder if a fork isn’t coming. It seems unlikely to me though given that the non-Microsoft world seems to be leaning in a Webkit direction and Chrome’s not doing so bad bringing it to Windows users either. Even if Mozilla wasn’t so busy shooting itself in the face, there was an unfavorable tide underway.
Version numbers are beautiful, angelic *and* necessary.
Any Mozillian wasting one calorie of energy on this bullshit topic instead of something as critical as MemShrink should really take a good hard look at themselves and wonder, WTF am I really contributing to the cause?
Dotzler must be going for the most morose and irrelevant topics record @ Mozilla. He’s got some good competition from the past so let’s all brace ourselves.
Since when does management equate to vague high-level thinking that slips through into the real world, rather than looking after the people who really do the work or just getting out of the way?
When Firefox forks, which it probably will in no more than three years time if they keep up like this, the web will be hurt a lot. Devs will flock towards the new fork, and now Firefox lacks manpower, imagine when it splits…
The future of the web doesn’t look to bright.
Asanine, the word is pronounced Asanine.
After I left https://bugzilla.mozilla.org/show_bug.cgi?id=678775#c35 , Asa explained his reasoning to me. I found these three arguments most convincing:
1) For users who just want to know whether they’re up to date, saying “Firefox is up to date (checked 7 seconds ago)” is more useful than showing a version number. We should make this change regardless of we drop version numbers from About. (To make this work consistently, we’ll need additional strings, such as “Firefox is not up to date because you’re on the abandoned PPC architecture; sucks to be you”.)
2) Once we do (1), the version number becomes a source of noise and confusion, at least for users who are just trying to see whether they are up to date. An especially bad cases is when we release 5.0.1 for Mac only: we don’t want Windows users to see “5.0″ in About, compare against some news article, and believe they’re out of date.
3) The troubleshooting and bug-reporting cases are better served by about:support, where users can grab not only the Firefox version number but also the OS version number, a list of extensions, recent crash reports, and graphics driver info. (Whether users will actually look in the Help menu, and successfully copy the relevant information, is another question.)
4) Bugzilla and website-feedback forms already capture the user-agent string.
5) You can always use https://wiki.mozilla.org/RapidRelease/Calendar to go from “User said on 2011-08-01 that they were using Firefox Aurora” to “User was probably on 7 Aurora”.
I’m not entirely convinced that the change is a good one, but I don’t think Alex Limi and Asa Dotzler are suggesting it out of delusional zealotry.
I’d really like to see Mozilla next year switch to year-month releases. The release of Firefox 12 or 12.04 on say 24 April, 2012 followed by Firefox 12.06 six weeks later would be a nice, accurate way to represent version numbers. The version numbering would also reduce the important of the version number because it’s just the month of the release. I especially like the consistency in version number history; Firefox 12 neatly follows after Firefox 11.
@Jeff He does not have a choice when it comes to supported versions, because there is only one.
@Robert Accettura: But that version discrepancies happen is a problem since there is only one supported one.
@Jeremy Bicha: This should works, unless a something gets delayed for some unfortunate reasons. What then? Release 12.04 in May?
Hey preed, that Facebook quote was said by one of the keynote speakers at OSCON this year. The whole quote was something like “Version numbers are obsolete. Nobody asks ‘What version of Facebook are you running?’”
I did not take detailed citation information, but I’m pretty sure it was this talk:
http://www.oscon.com/oscon2011/public/schedule/detail/21244
(there’s a video so you can check if you want, tell me if I’m wrong).
I think the point of statements like that is that apps have blurred the line: they feel like client software and operate in a similar silo, but in many other ways behave like webapps. And then of course there’s Chrome.
FWIW, I don’t think version numbers are ever going away, even if that bug goes ahead.
@Jesse Ruderman: I thought that the reasoning is something like this – it actually makes sense. But users don’t see this reasoning, there is clearly a huge communication problem. I also think that rapid releases created lots of distrust which explains the extreme reaction to this suggestion. Which is why I suggested to postpone this change in my blog post.
Whew! Lots of great discussion here; let’s see if I can respond to everyone:
@Jeff,
Version numbers are doubleplusungood.
As followers of IngSoc, we both know the only version of the software that was every released was “latest”.
And it was bug-free. And the features doubleplusplusgood.
@Robert,
I meant “goading” in the best possible sense… more along the lines of “Well, if he’s gonna share his thoughts on it, I certainly am!”
I think your phrasing hits the nail on the head: client side apps being user-controlled is a large (unsaid, because it sounds bad when you verbalize it) part of this debate; removing the language to discuss which bits are running on your local CPU is part of moving toward a world where you are no longer allowed that control; don’t fret: the corporations will let you know what you’re allowed to do with your hardware.
All we’re hearing from vendors moving in this direction is “We want you on the latest software.”
As a user, what about what I want?
@Ron,
I doubt a fork is coming; Mozilla Corporation controls the most important aspect of Firefox—the branding trademarks and copyrights—and while it may be easy to “fork the code,” that analysis is simplistic in (as it’s been argued to me, and as much as I hate to, I probably agree) that the majority of users won’t use a “knock-off Firefox.”
I think Mozilla Corporation’s behavior would have to be blatantly worse to cause people to leave, and even so, they will move to Chrome; that’s a lot easier than forking.
@pd,
I detest (and refuse to use) the term “Mozillian” ever since Mark Surman told me I had to pay $5 to be one.
To your other point: open source tends to embrace bike shed conversation, despite the fact that they’re the most pointless of conversations to have in that type of organizational context. At some point, a decision will be made, and I think this discussion is mostly an exercise in “Get all the bitching out, people, because the decisions is already made, and we need to have something to point to when people complain about it in the future to say ‘We already talked about that.’”
(This isn’t a new pattern at all from a Mozilla governance perspective.)
@Tiango,
See my comments above about forking.
I doubt it’ll happen, but if it ever does, I seriously doubt developers will “flock” to the new fork. I would assert that Mozilla (and certainly Firefox) don’t have many (if any) “community developers” anymore; they’ve either been hired to work for Mozilla Corporation (and those people won’t be working on a fork), or those that weren’t (for whatever reason) have been marginalized to the point they’ve left as well.
@Jesse,
1) Great; I totally support adding that. It doesn’t speak to why the version number needs to be removed.
2) A source of noise and confusion? Just how moronic does Mozilla really consider Firefox users to be? (Regarding 5.0.1 for the Mac, removing the version number is attempting to address poor program and release management [and possibly release engineering; I'm not totally familiar with the details] by futzing with strings; Mozilla should solve these problems instead of futzing with the strings.)
3) I knew someone was going to bring up about: pages; “whether users will actually look in the Help menu” is the seed of the counterargument…
4) Because I always file bugs from the exact browser that has the bug. (And it always detects said browser correctly, too!)
5) So… user experience people are telling me that if I’m JUST TRYING TO FIGURE OUT WHICH DAMN VERSION OF THEIR SOFTWARE I’M USING, I need to correlate the date of the post and cross-reference it with a wiki, or some other data source? Really? That’s a better user experience? Really? (Honestly, this is where so-called “user experience”-people totally lose me; their reasoning, sometimes, makes no logical sense whatsoever.)
As for “delusional zealotry,” I don’t find such language particularly useful, for a couple of reasons: a. I don’t know (nor do I particularly care) about the personal or emotional motivations driving this change; I’m not speaking to that. I’m speaking to the merits of the idea; and b. I want nothing I’ve said to be construed as me making such a claim; another common pattern is the second someone thinks they smell a whiff of “trolling”, they decry that person “a hater,” and forcibly shut down the discussion, even if 98% of what they said was based in an objective argument.
I refuse to play that game.
@Jeremy,
Monthly releases would be a stretch for Mozilla, I think, given the experience (and issues) with “rapid release.”
(There’s also the bit called out below about what do they call it if it slips.)
Having said that, I don’t think releasing faster is what Mozilla should focus on; this whole “using users as pawns to move the lever that will move The Web”-argument I’ve made aside, as my mom always told me, “Don’t try to be someone else.”
Mozilla is trying to be someone else, and it’s unfolding the way you’d generally expect that to.
@Beno,
You keep referring “supported versions,” as if that had any relevance to this discussion.
That’s a completely orthogonal point.
@Laura,
Thanks for pointing me at the video!
The quotation—”I’m going to steal a punchline from a good friend who once told ‘OK, What version of Facebook are you running?’ to let me understand that even basic concepts like versions of software don’t apply anymore.”—is here.
Your citation of it is lacking some critical context, though; surrounding that statement, he says: “This is what’s happening in the cloud, because you see the cloud changes a lot of things; a lot of the yard sticks we had before the cloud existed are not there anymore. [quotation above.] And actually, in the cloud, you have all the technologies…”
Unless you’re making the claim that Firefox is now somehow The Cloud(tm), then I stand by my claim that it’s a total false analogy.
As raccetura pointed out, client applications are, by definition, controlled by the user: they install/(can try to) uninstall it.
There is a definite move to treat users like children, and force them to run whatever companies have decided they want them running once the user takes the initiative to install it. I find that distasteful on a number of levels, and contradictory to the claim (which Mozilla makes constantly) “We’re all about the user.”
I understand certain people will never want to see or hear about upgrades, and “always want the latest”; they don’t want to see 30 “time to upgrade” boxes (in some sense, Firefox, along with Adobe’s Flash Updater created these problems in the first place, along with Windows Update). But there are perfectly reasonable ways to address these issues while still allowing other users who want to make informed decisions about these sorts of things do so.
To pretend that everything is “cloudy” and “blurring together,” or that you “JUST CAN’T LET USERS HAVE A CHOICE, BECAUSE THEY’LL MAKE THE WRONG ONE” is where we start to tread in to the disingenuous realm I was referring to above.
Part of the (underplayed) implication when people present this false dillema is that they seem to be assuming that the software they release is so perfect, everyone will always want the latest. Anecdotally, many users consider certain products—web browser, email reader, etc.—to be “done”, and are entirely happy with what they have. They are, in fact, frustrated that they keep getting updates that break of change functionality they’ve come to rely on, and often times, this is due to poor quality.
For what it’s worth, this speaker doesn’t garner much credibility with me; a few minutes later, he says “I actually argue that in the cloud, the [open] source code is the Terms of Use and SLA much more than the bits.” Around this point, he discusses how it doesn’t matter what stack, open or closed, is producing the JSON feed you’re consuming.
This is a delicious twisting of what “open source” and “openness” really means, in a way only a Microsoft employee could contrive.
The Facebook analogy doesn’t make sense to me. At any given time, there is exactly one version of FB. Users have no influence. But Firefox is distributed software, and each copy has its own update cycle. That’s a fact, like it or not. That’s why we need to talk about version numbers, to allow for precise support and problem analysis.
Kai
To start with a side note, Jeremy Bicha didn’t say “monthly releases”, he said “year-month releases”. That is, releases numbered YY.MM, no matter how frequent they are. For example, Ubuntu 11.04 was released last April, and the next version will almost certainly be called 11.10 because it’s scheduled for mid-October. That also answers Bono’s question about what happens if it slips: it happened once with Ubuntu, where instead of 6.04 we released 6.06.
Having said that, I would strongly discourage any other project from adopting that scheme. Not an Ubuntu release goes by without magazines, news sites, and users in general misunderstanding the version number, for example thinking that “11.04” is just a point-release of some earlier Ubuntu 11 that they missed. (And every day at the office I walk past a prize plaque that LinuxWorld awarded to us for “Ubuntu 7.08” … a version that never existed.)
I’ve experienced the Web app side of this argument too: Canonical’s Launchpad Web site used to make a big deal about its own version number. I argued against that, for exactly the inverse of the reasons you’ve presented here. You (almost always) can’t choose which version of a Web site you’re using, so it doesn’t help anyone to know which version you’re using. Partly for that reason, Web sites should improve incrementally rather than in bursts; and they should advertise those incremental improvements rather than versions. The only time someone should need to know a Web site’s version number is if they’re developing against that Web site’s API.
Which brings me finally to emphasize a point you touched on briefly: “versioning-as-a-concept being a proxy for something — API compatibility — in the Mozilla platform itself.” The more something acts as a platform, the more important its version number is. Version numbers are more important for Facebook developers than Facebook users. They’re more important for Web browsers than they are for games. They’re more important for games that have mods than games that don’t. And they’re more important for browsers that have extensions than browsers that don’t.
Also worth mentioning that Facebook “updates” aren’t the user’s problem. Firefox updates are. If Facebook updates their service, it doesn’t break numerous things on the user’s side, except perhaps for Greasemonkey scripts, which unlike Firefox addons, Greasemonkey scripts are not part of the Facebook architecture.