Daring Fireball: The Case for Going Metro-Only on ARM(daringfireball.net)
daringfireball.net
Daring Fireball: The Case for Going Metro-Only on ARM
http://daringfireball.net/2011/09/the_case_for_going_metro_only
9 comments
After running Windows 8 on my laptop for a while, I'm struck by how poorly-suited Metro is to a traditional (mouse/keyboard) computing experience. My sense is that traditional PCs should boot to the traditional Windows desktop and tablets should only have Metro. It would be seen as a nice benefit that traditional x86 computers can ALSO run Metro apps but I don't think anyone will lament the fact that you can't run Excel on your tablet...
"It would be seen as a nice benefit that traditional x86 computers can ALSO run Metro apps but I don't think anyone will lament the fact that you can't run Excel on your tablet..."
Counterpoint: The next IMHO-blindingly-obvious step in computer evolution is that your tablet/cell phone becomes your desktop when plugged into a probably-USB-based dock that adds a keyboard, mouse, and real desktop monitor (or more). Tablets that can't run real OSes days are numbered.
Having a well-delineated tablet UI layer is good. We've got over a decade of proof that just smashing a conventional desktop onto a tablet doesn't work. But I would not want to massively commit to the idea that tablets will never run Excel. They will... just not in their touch-only mode.
It's all about how committed you end up to it, though; in the short term, if only Metro is on the tablet of 2012, that's OK, as long as you can pivot, potentially as soon as early 2013. (Prototypes of this ought to exist somewhere now if MS is going to pull this off.) But if you've written this requirement deep into the tech stack, you're in trouble.
Counterpoint: The next IMHO-blindingly-obvious step in computer evolution is that your tablet/cell phone becomes your desktop when plugged into a probably-USB-based dock that adds a keyboard, mouse, and real desktop monitor (or more). Tablets that can't run real OSes days are numbered.
Having a well-delineated tablet UI layer is good. We've got over a decade of proof that just smashing a conventional desktop onto a tablet doesn't work. But I would not want to massively commit to the idea that tablets will never run Excel. They will... just not in their touch-only mode.
It's all about how committed you end up to it, though; in the short term, if only Metro is on the tablet of 2012, that's OK, as long as you can pivot, potentially as soon as early 2013. (Prototypes of this ought to exist somewhere now if MS is going to pull this off.) But if you've written this requirement deep into the tech stack, you're in trouble.
"The next IMHO-blindingly-obvious step in computer evolution is that your tablet/cell phone becomes your desktop when plugged into a probably-USB-based dock"
The idea is obvious, yes, but USB? Thunderbolt is, IMO, much more likely to be _the_ connector. It has the speed to do the video that USB (even USB 3) lacks (power requirements may be a problem, but I think USB 3 has the same problem)
I also think that the parties building those docks will want to include a 'cloud connection', for which they will charge whatever they can, because, in general, hardware is not a business to make money in anymore. I am not sure whether they will get away with that. One thing that will depend on is what the economy will be doing.
The idea is obvious, yes, but USB? Thunderbolt is, IMO, much more likely to be _the_ connector. It has the speed to do the video that USB (even USB 3) lacks (power requirements may be a problem, but I think USB 3 has the same problem)
I also think that the parties building those docks will want to include a 'cloud connection', for which they will charge whatever they can, because, in general, hardware is not a business to make money in anymore. I am not sure whether they will get away with that. One thing that will depend on is what the economy will be doing.
The next IMHO-blindingly-obvious step in computer evolution is that your tablet/cell phone becomes your desktop when plugged into a probably-USB-based dock that adds a keyboard, mouse, and real desktop monitor (or more).
Thanks for publicly mentioning that (seriously): Your mention in itself is now "prior art" for pretty much anyone who might later try to patent the idea. (IIRC, Arthur C. Clarke described something much like that in one of his Rama novels back in the late 1970s or early 1980s.)
Thanks for publicly mentioning that (seriously): Your mention in itself is now "prior art" for pretty much anyone who might later try to patent the idea. (IIRC, Arthur C. Clarke described something much like that in one of his Rama novels back in the late 1970s or early 1980s.)
It's been mentioned many times before, including by ESR in 2008: http://esr.ibiblio.org/?p=614. Not that I expect that to stop the patent office from handing a lawsuit factory to the next Lodsys.
This sounds just very similar to the route Apple has chosen (though their trackpad tech + latest OSX blurs the lines a bit).
If iOS apps could exist as say, dashboard widgets, there would be no difference (Apple probably thinks this is a silly and sales-reducing idea).
If iOS apps could exist as say, dashboard widgets, there would be no difference (Apple probably thinks this is a silly and sales-reducing idea).
Well, there's nothing wrong with copying something that works. Sometimes others get things right first and the best move is to follow. Innovation at the expense of functionality is stupid.
I completely concur. I am beginning to suspect another release flop ala Windows Me and Windows Vista... if Microsoft default to Metro on traditional workstations. I say 'if' because they're still a bit vague on their roadmap. There's lots of hype in the air now, but once the dust begins to settle I think more people will come to this realization.
EDIT: Changed Windows 2000 to Windows Me. Windows 2000 was generally considered to be a successful release.
EDIT: Changed Windows 2000 to Windows Me. Windows 2000 was generally considered to be a successful release.
I agree. It's almost the exact reverse of what they tried to do with Tablet PCs before the iPad: wedge a fundamentally inappropriate UI into a place that demands something else.
Apple also seems to be headed in the same direction with its introduction iOS features into Lion: http://blog.byjoemoon.com/post/3556631202/touch-ui-is-not-th...
Apple also seems to be headed in the same direction with its introduction iOS features into Lion: http://blog.byjoemoon.com/post/3556631202/touch-ui-is-not-th...
I respectfully submit that each and every DF post isn't worth front-page attention on HN.
https://news.ycombinator.com/item?id=3019147
Edit: The link is a request for PG to blacklist DF. I honestly don't think the submissions add to the overall quality of HN.
https://news.ycombinator.com/item?id=3019147
Edit: The link is a request for PG to blacklist DF. I honestly don't think the submissions add to the overall quality of HN.
I think people upvote it for the conversation. People want to discuss Gruber, and since he doesn't have comments on his blog, they come to HN.
Gruber can be very partisan, but this is one of his better posts, written respectfully and intelligently with none of the partisan slant people often complain about. IMO, this is an example of why he shouldn't be banned on HN.
A couple of points:
* People forget that Microsoft is used to selling Windows that runs on multiple architectures. They had(have?) Itanium, DEC Alpha, PowerPC, and MIPS in the past. Of course those were targeted at professionals, not consumers.
* I don't think MS really cares that much about x86 native apps running on ARM. They care about desktop managed apps running on ARM. Why? LOB apps. LOB apps are managed and they want to support ARM tablet users having access to these.
* Related to the above point, there are actually few native x86 apps that really matter. Look at the Amazon top 20 selling SW list. It's basically: Office, Windows, Intuit, Adobe, and antivirus. Microsoft ships half of the top 20. They'll be shipping antivirus, so effectively shipping 2/3rd of the top 20. The only apps in the top 20 they'd need to get are Intuit (Quickbooks/Quicken/TurboTax) and Adobe Premiere XYZ. MS can probably have at launch 95% of the cycles consumed by native x86 apps on ARM.
With that said, for once, I do think Gruber is generally correct.
* People forget that Microsoft is used to selling Windows that runs on multiple architectures. They had(have?) Itanium, DEC Alpha, PowerPC, and MIPS in the past. Of course those were targeted at professionals, not consumers.
* I don't think MS really cares that much about x86 native apps running on ARM. They care about desktop managed apps running on ARM. Why? LOB apps. LOB apps are managed and they want to support ARM tablet users having access to these.
* Related to the above point, there are actually few native x86 apps that really matter. Look at the Amazon top 20 selling SW list. It's basically: Office, Windows, Intuit, Adobe, and antivirus. Microsoft ships half of the top 20. They'll be shipping antivirus, so effectively shipping 2/3rd of the top 20. The only apps in the top 20 they'd need to get are Intuit (Quickbooks/Quicken/TurboTax) and Adobe Premiere XYZ. MS can probably have at launch 95% of the cycles consumed by native x86 apps on ARM.
With that said, for once, I do think Gruber is generally correct.
WRT your first point, Microsoft used to sell Windows on multiple architectures. They abandoned Alpha, PowerPC, and MIPS in 1999. http://en.wikipedia.org/wiki/Windows_NT#Supported_platforms A big part of the problem was that software vendors did not make non-x86 versions of their software and subsequently users were confused and disappointed that the software that they wanted to run on their (e.g. Alpha) computer either wouldn't run or had to run in an emulator and thus ran slower than on an x86 machine.
Microsoft supports two architectures currently: x86/Itanium and ARM with WinCE/WinMobile/WP7. There is no real synergy between the two currently, e.g. the "Office" apps that run on ARM are totally different beasts than what run on x86. The primary crossover is "mindshare" - i.e. users recognize the names.
Microsoft supports two architectures currently: x86/Itanium and ARM with WinCE/WinMobile/WP7. There is no real synergy between the two currently, e.g. the "Office" apps that run on ARM are totally different beasts than what run on x86. The primary crossover is "mindshare" - i.e. users recognize the names.
I think I get the whole reasoning but I really disagree with one of Microsoft's decisions on this.
* ARM is Metro only
* x86 is old + Metro
* All Metro apps must be purchased in the App Store
I understand the last one from an ARM perspective. It stops any confusion if the app with run on ARM. If I can see it in the App Store then it will run. I just think it is a bit of overreaction / greedy to require all x86 Metro apps to be in the App Store. It just seems like a really bad way to mess up adoption of Metro on x86.
* ARM is Metro only
* x86 is old + Metro
* All Metro apps must be purchased in the App Store
I understand the last one from an ARM perspective. It stops any confusion if the app with run on ARM. If I can see it in the App Store then it will run. I just think it is a bit of overreaction / greedy to require all x86 Metro apps to be in the App Store. It just seems like a really bad way to mess up adoption of Metro on x86.
Not trying to be nitpicky, but capitalized App Store is the Apple owned store. It is app store if meant generally or Windows Marketplace if speaking of the Microsoft branded store.
This whole "not confusing people" thing seems like a nifty idea, but I think we're talking about a different company.
Is there any reason to believe that Microsoft doesn't want to confuse consumers or that not confusing consumers is in any way a winning strategy for them? Yes, I'm being sarcastic, but it seems to me that not confusing customers is Apple's schtick. Microsoft does very well with a different business model that historically has included:
WinCE WinMobile PocketPC etc. even Windows Phone 7 has gone through several name variations.
Windows Vista and 7 both offered a bewildering range of SKUs.
Version naming and numbering schemes that make no sense to anybody. (Although Final Cut Pro X is an admirable attempt by Apple to compete here, Pro having been Apple's traditional replacement for 3.)
And don't forget that Microsoft's hardware partners are all past masters of incomprehensible product lines.
And can anyone figure out which Intel CPU is which without using google today?
Is there any reason to believe that Microsoft doesn't want to confuse consumers or that not confusing consumers is in any way a winning strategy for them? Yes, I'm being sarcastic, but it seems to me that not confusing customers is Apple's schtick. Microsoft does very well with a different business model that historically has included:
WinCE WinMobile PocketPC etc. even Windows Phone 7 has gone through several name variations.
Windows Vista and 7 both offered a bewildering range of SKUs.
Version naming and numbering schemes that make no sense to anybody. (Although Final Cut Pro X is an admirable attempt by Apple to compete here, Pro having been Apple's traditional replacement for 3.)
And don't forget that Microsoft's hardware partners are all past masters of incomprehensible product lines.
And can anyone figure out which Intel CPU is which without using google today?
I think the easiest way to minimize confusion would be to market ARM-based Windows machines as “Metro only”. Intel gets classic Windows and Metro apps, ARM gets Metro.
I don't think this would be that much less confusing for lay users. I would imagine MS will rely on pushing 3rd party software through their App Store, and only exposing compatible software. I imagine this will work better to minimize end-user confusion and to enforce developers doing real ports of their applications instead of just recompiling.
I don't think this would be that much less confusing for lay users. I would imagine MS will rely on pushing 3rd party software through their App Store, and only exposing compatible software. I imagine this will work better to minimize end-user confusion and to enforce developers doing real ports of their applications instead of just recompiling.
By not allowing legacy apps on ARM microsoft would not have any leverage in the tablet space. If I have a choice between iOS and Windows ARM where the former had a ton of apps and latter doesn't have many, then why would ?I choose windows? It's the same Windows Phone story again. No matter how superior it is to the iPhone, it will lose out for the lack of apps.
Microsoft knows this. I won't be surprised if they have more nd more support for legacy API and applications on ARM over time.
Microsoft knows this. I won't be surprised if they have more nd more support for legacy API and applications on ARM over time.
I believe Microsoft has already stated that only Metro apps will run on ARM (no legacy x86 apps):
http://www.infoworld.com/t/microsoft-windows/will-windows-8-...
Steve Sinofsky: "We've been very clear since the very first CES demos and forward that the ARM product won't run any X86 applications. We've done a bunch of work to enable that -- enable a great experience there, particularly around devices and device drivers... [W]hat we announced yesterday for the first time was that when you write a Metro style application, all the tools are there to enable you in any of the languages that we support to automatically support ARM or X86. I think that's the key part of everything that we'll run."
Why does the argument that by some magic you'll be able to purchase an iPad-like Windows 8 tablet and run full blown desktop Microsoft Office keep cropping up?
There will probably be a version of MS Works that is written for the Metro touch interface that will work on ARM, but the performance of an x86 desktop application running in an emulator on ARM would be atrocious and everybody knows it.
http://www.infoworld.com/t/microsoft-windows/will-windows-8-...
Steve Sinofsky: "We've been very clear since the very first CES demos and forward that the ARM product won't run any X86 applications. We've done a bunch of work to enable that -- enable a great experience there, particularly around devices and device drivers... [W]hat we announced yesterday for the first time was that when you write a Metro style application, all the tools are there to enable you in any of the languages that we support to automatically support ARM or X86. I think that's the key part of everything that we'll run."
Why does the argument that by some magic you'll be able to purchase an iPad-like Windows 8 tablet and run full blown desktop Microsoft Office keep cropping up?
There will probably be a version of MS Works that is written for the Metro touch interface that will work on ARM, but the performance of an x86 desktop application running in an emulator on ARM would be atrocious and everybody knows it.
There are three types of Win8 apps: Metro, Win32/ARM, and Win32/x86. The question is whether Win32/ARM apps will be allowed.
Why does the argument that by some magic you'll be able to purchase an iPad-like Windows 8 tablet and run full blown desktop Microsoft Office keep cropping up?
Because people like Office, MS makes a lot of money from Office, and MS actually showed full-blown desktop Office running on ARM a few months ago.
Why does the argument that by some magic you'll be able to purchase an iPad-like Windows 8 tablet and run full blown desktop Microsoft Office keep cropping up?
Because people like Office, MS makes a lot of money from Office, and MS actually showed full-blown desktop Office running on ARM a few months ago.
Microsoft has already shown a version of Office running on ARM natively. For lot of current Windows app written in .NET or standard Windows API, the transition should be easy so long as MS maintains the same API's across architecture. Only problem would be for old legacy software where company that maintained it is out of business, or software that heavily relies on kernel specific functionality, etc
The problem is a lot of legacy apps won't run on ARM anyway because they are compiled to x86/x64. This just avoids people buying an ARM laptop and expecting their copy of Office 2007 to work.
There should be some legacy apps that will run. Especially if non-phone ARM devices get the full .net framework.
For someone who tends to be dismissive of Microsoft, Gruber sure seems to be writing about them a lot lately.
I'd say he may be pretty dismissive of Microsoft in general, but I think nearly anyone that has used or seen Windows Phone 7 is thinking it's a great thing.
But that actually comes down to the same thing. The reasons Gruber likes WP7 is because it cut the cord with all previous versions. It's a fresh start. Which Gruber feels was needed. I actually agree with him on this point.
Now, where the problem comes with regard to Windows 8 is simply that there is no cutting the cord with past versions. It still has all that legacy support in it and keeps getting more convuluted. Even more now that there is support for two different processor families.
Windows 8 would be a great chance to slice away some of the old legacy crap.
This is simply an example of where Apple has really excelled when they've been in a similar situation. Not that the x86/ARM thing is exactly the same. Apple's was a transition, while Microsofts is in Addition to. But Apple provided tools to make the user experience seamless while Microsoft is making it more difficult for average users to understand.
But that actually comes down to the same thing. The reasons Gruber likes WP7 is because it cut the cord with all previous versions. It's a fresh start. Which Gruber feels was needed. I actually agree with him on this point.
Now, where the problem comes with regard to Windows 8 is simply that there is no cutting the cord with past versions. It still has all that legacy support in it and keeps getting more convuluted. Even more now that there is support for two different processor families.
Windows 8 would be a great chance to slice away some of the old legacy crap.
This is simply an example of where Apple has really excelled when they've been in a similar situation. Not that the x86/ARM thing is exactly the same. Apple's was a transition, while Microsofts is in Addition to. But Apple provided tools to make the user experience seamless while Microsoft is making it more difficult for average users to understand.
The BUILD conference was a nice coming out party for Windows 8, and it's given people the opportunity to be excited about something non-Apple for a change.
If he wasn't covering it, you'd hear cries of "Typical Gruber, ignoring all the interesting things going on outside of Apple."
If he wasn't covering it, you'd hear cries of "Typical Gruber, ignoring all the interesting things going on outside of Apple."
Actually, reading this made me think that Gruber is getting better again. DF used to be great when it was Apple vs Microsoft/OSX vs Windows and not so good now it's Apple vs Google/iOS vs Android.
He knows how to write about the former, less so about the latter.
He knows how to write about the former, less so about the latter.
Fortunately for the rest of us, Gruber's writing is much more insightful than yours.
Today we are going to have a civilized discussion of a Gruber article on HN.
For those that want to discuss the article, comment elsewhere!
For those that want to attack or defend Gruber, just upvote the appropriate account below.
(If you want to see who is winning, click the appropriate account name. Both accounts started with 1 karma and no other comments or submissions.)
For those that want to discuss the article, comment elsewhere!
For those that want to attack or defend Gruber, just upvote the appropriate account below.
(If you want to see who is winning, click the appropriate account name. Both accounts started with 1 karma and no other comments or submissions.)