Hacker News Reader

>This meant: 1. it was in development for 2 years and thus probably stabler to begin with

No. Shorter release cycles don't imply loss of quality. I claim the opposite in fact. Rolling releases are much better for quality. You're releasing smaller more focused features and battle-testing them in the field.

>There is no stabilization phase.

There are always stabilization phases. Not every major release will be an overhaul. Most will be incremental updates.

> No. Shorter release cycles don't imply loss of quality. I claim the opposite in fact. Rolling releases are much better for quality. You're releasing smaller more focused features and battle-testing them in the field.

The jury is out on this. We have seen that longer release cycles with teams that take things seriously can be very successful (mission critical software). We've also seen that short release cycles can be very effective. The context of your software matters a great deal here, as well as the basic mechanics of how your software is delivered. You can't just take the philosophies of web sites and apply them across the board and declare it objective fact:

1. Take for example the fact that with a web service you can deploy quickly to a small subset of users, and grow it as it proves itself. This is quite ideal, but regardless of whether its even practically possible with OS software, its certainly not what Apple does. The combination of a short release cycle with a worldwide release means that "catching your bug" may mean catastrophic data loss for a large amount of customers.

2. With a web service you have the option of rolling back a bad version, or, simply deploying yet another release. Again, whether its possible with OS's or not, Apple certainly doesn't (or can't?) employ this strategy. The best they can do oftentimes is simply remove the update, but everyone that has it installed keeps having the bug until a) your fix is out and b) they actually install said fix.

> There are always stabilization phases. Not every major release will be an overhaul. Most will be incremental updates.

There certainly could be stabilization phases, but the proof is in the pudding and thats not how it feels. All the last releases have seen major additions, and most the regressions I run into either take years to fix (dual monitor support) or still remain increasingly broken (Messages to name just one example). There is a simple reality to having a year to stabilize without the pressure of a PR push of new features vs. balancing your stabilization with coming up with compelling things to slap on a website and show in a keynote.

In theory, but Apple has seemed to copy the MS playbook of releasing software by an arbitrary date—whether or not it is ready. They seems to add more features, instead of fixing bugs. It seems the only time I get a Software Upadate pop up is for Camera RAW, iTunes or a patch for the exploit of the day.

Mac OS X: The 2000-2003 period included three full releases (not counting 10.0, since that was the culmination of several years of work), and didn't have quality problems. The 2003-2011 period had four releases. So I don't see an obvious lesson that one year is sustainable.

iOS: 2007-2011 four updates at one year intervals before the decline in quality that lots of people perceive starting with iOS6. Again, hard to say one year is the culprit.

I do think Apple needs to restrain itself, and work on figuring out which features can be shipped without compromising quality. But whether it needs to do longer releases or do smaller releases is hard to say.

As for Mac OS, when I read Siracusa's review, I was thrilled. There aren't consumer features, but they're doing so many things to make desktop machines more responsive and efficient. It's even more exciting than Snow Leopard.

The period between 2000 and 2003 is kind of hard to draw any conclusions from or the classify as "not having quality problems". That period started with quality problems and saw Apple scrambling to make Mac OS X a reasonable OS to replace OS 9. For example, 10.1 re-introduced disc burning which was missing from 10.0. OS X 10.0 was also riddled with kernel panics and really sluggish. 10.1 continued to have pretty severe crashes and sluggishness, and I remember Jaguar being the first thing that approached actual usability. Panther arguably ended this transition period (where people were still unsure whether to "make the jump") and ushered in the more stable phase I'm referring to (2003 - 2011).

And sure, if Apple only focused on a very sensible amount of changes, then of course a year would be enough. The problem is the contention of a 1 year release cycle and "150 new features!".

I'm not sure the intial problems really sabotage my point. They were able to, year after year, squash bugs while adding substantial functionality that was missing in 10.0.

I guess "not having quality problems" was a poor phrasing. The point is that they weren't suffering from a quick release cycle. But perhaps it's ambiguous.

> when I read Siracusa's review

Are you referring to his Yosemite review? Is it out yet?

No, I was thinking of Mavericks. I was clearly too tired to be commenting yesterday.

There's nothing wrong with yearly release cycles, Apple's problem seems to be that they've tied their OS to the hardware cycle.

It doesn't really matter if the OS is completely done, or still has a fair number of bugs. It MUST release at the same time as the new phone.

Combined with this is the fact that the apps are bundled with the OS. While certain enhancements definitely need operating system support, there's no good reason they couldn't be updating Safari or mail or notes to quash little bugs throughout the year. Instead you have to wait for a .1 release or the next full OS (since they never seem to do .2s).

A problem with yearly releases is that Apple only seems to give support out for the most ~3 recent versions. (At least in OSX world; I believe there is still no shellshock patch for 10.6.) So do you have support for ~4 years, or ~10 years?

>I'd just like to point out that this is likely not to get better with this (arbitrary) decision to release a new OS every year.

Release dates are arbitrary and inconsequential. Scope and freeze dates are what matters.

>Couple this with the stark reality that Apple has simply run out of ideas in terms of software. Every new version of OS X boils down to: 1. arbitrary UI tweak (forcing developers to refresh), 2. Gimmick features in Mail.app/Safari (RSS in Mail/Safari, Postcards in Mail, yet another 3d effect to re-arrange your tabs in Safari, annotations in Mail, etc etc etc), and 3. regressions of features that worked for years.

Siracusa begs to differ.

> Couple this with the stark reality that Apple has simply run out of ideas in terms of software

I dunno. Continuity seemed to me to be novel and useful when it was presented. Healthkit is new, and not really "just a refinement".

Continuity is really interesting since it really highlights just how far we've gotten from the original promise of seamless integration between Apple products. For starters, its taken years to get to this feature (with incredibly confusing pitstops in the middle, such as the existence of a feature called Airdrop on Mac AND iPhone that didn't work between them....).

All in all though, continuity is another disappointment to me. We are still stuck in app-land. Handover is app-to-app. So when I'm listening to a podcast in Safari, I won't be able to handover to Overcast on my phone. I actually have to wait for marco arment to write a desktop podcast player, and worse, I have to use it, to get this basic functionality. Again, an aggressive misunderstanding of what it means to have true continuity: reading the same data on two devices, not using the same app on both. But Apple thinks wallet first, and "App" is their ecosystem's bread and butter, and so this way it will stay.

This is particularly depressing since I remember, 10 years ago, having (what I believe to be) much more impressive interconnectivity between my Mac and my Nokia phone through Saling clicker. When someone called me, my music would pause. I could control my mac with a really cool remote. Instead, we seem to be living in a parody of that world where now if someone calls me 10 devices in my house go off since Apple "connected" them all together to ring simultaneously, its maddening.

Apple (and others) have clearly double downed on an app-centric metaphor. This is obvious from App Store and home screen only showing apps--interestingly they did add recent people to the multitasking menu). They're also struggling with trying to reinvent documents (and haven't done much). Services/Workflow-centric and people-centric are the other approaches I've seen attempted[1] with only minor success. I don't blame Apple, none of the other approaches have gained any traction and Apple has really streamlined app centric management: centralized finding apps, purchasing, encapsulating installing/uninstalling, updating, etc. All of those things were pretty inconsistent and terrible before OSX/iOS.

I also wished they refined more of the Bluetooth integration that OSX had 10 years ago (sending/receiving txt messages, using it as a remote, etc). Sadly, it only now seems to be coming back and in an iPhone-specific way. But other than the iPhone tie-in, what would be different? If you had your phone Bluetooth paired to multiple devices, wouldn't they all buzz? If you don't want them to buzz, can't you just disconnect them (as you would un-pair bluetooth)? If you had a remote tethered to multiple devices, would "Play" trigger them all at once?

[1] http://www.hanselman.com/blog/WhyIsntPeopleCentricUIDesignTa...

> It used to be that we'd have a new release roughly every 2 years (from 2003 - 2011). This meant: 1. it was in development for 2 years and thus probably stabler to begin with, and 2) there were then 2 years to iron out the bugs.

Doesn't this discussion boil down to the trade-offs between slower and more rapid release cycles? For example, releases after 2 years of development are far more complex, with all the disadvantages that entails. Technology reaches users more slowly (solutions to some bugs/features could take 3 years or more) which also delays feedback to developers. etc.

What's funny is that Apple seems to have found the anti-sweetspot for its release cycles.

For example, for an app like Safari, it would make so much more sense to release very frequently (ala Chrome/Firefox) separate from the yearly OS cycles. Instead, new web standards features/JS improvements/what-have-you are inexplicably tied to fluff features like 3d tabs. On the other hand, features that would make more sense to not mutate constantly on your user get changed way too frequently (for example Mail, which quite frankly doesn't need constant updating, yet still broke Gmail for so many people on Mavericks release and should have really been caught beforehand). This strikes a balance that lacks the benefits of continuous deployment AND slow sober releases.

I think the opposite is true. iOS7/8 was a giant transition that was staged over two releases. I suspect the next few years will be much less major reworking and much more tuning up and adding capabilities.