This was a review for someone who doesn't read slashdot. There wasn't one subtle point. It's well known that google docs require being online and it's also well-known that google is working to change this. Anyone who has spent five minutes examining the capabilities of any of the Google apps knows they fall far short of Microsoft's software. The "power-point killer" designation strikes me as idiotic for a 1.0 version of a new class of software (browser-based presentation creation).
Moreover, suitability is all about what you're presenting. Suppose the reviewer had asked a mathematician to do a comparison of these three presentation packages on the one hand with LaTeX/PDF on the other, for the purposes of giving a mathematical talk. I can tell you from experience that Powerpoint is a joke for this purpose. (I'm not a mathematician but I do include a lot of equations in my slides. LaTeX/PDF rocks.)
Just a few months ago I watched a colleague give a powerpoint presentation and stare in horror at his projected slides because, without realizing it, he had rendered them totally unreadable by using his tablet PC to add last-minute graphics to them (supposedly using the tablet feature as it was intended). You can screw up with Powerpoint too.
This is the very beginning. The interesting thing to speculate about is what the office suite arena will look like three years from now. I'm betting that Google will, at the very least, shake things up a *lot*.
The antitrust complaint about Microsoft has been that everything was bundled in Windows. Isn't this what the antitrust authorities want them to do? Isn't modularity one of the things that's great about Linux? An unbundled commercial OS would allow Microsoft to have a stripped down version to compete with OLPC, and home and business versions that charged by the feature. You don't want eye candy, you don't pay for eye candy. Not a gamer, you don't pay for DirectX. In theory it seems like a great way to tell developers what to work on (you'll see which features folks are willing to pay for) and a way to offer cheap versions of the OS for the folks that just want e-mail/browsing and expensive versions for those who want everything. You're thinking that they'll try to charge a lot, but really, how much worse could it be than what we have now? Most users pay microsoft a lot every few years.
Now this may be a technical and end-user nightmare, I don't know, but trying to develop an a la carte OS makes a lot of sense to me as a business move.
The same thing is starting to happen with cable more extensively than before: you don't want ESPN or the NFL, you shouldn't have to pay for them.
It's clear you aren't a LaTeX user because you've got it exactly backwards. For many users, the main benefit of LaTeX is the ability to focus on the logical structure and content of the document, and not to have to think at all about appearance, which LaTeX handles automatically. The benefits of auto-numbering, robust cross-referencing, and sophisticated bibliographic handling increase as your document gets longer.
Every Word user I know spends a lot of time messing with formatting. Maybe they don't have to, but they do. A WYSIWYG environment encourages you to do that. You don't have to touch formatting in LaTeX. And I can assure you that someone using Winedt, Vim, or Emacs to create a LaTeX document is not thinking about appearance as they're writing. This is why it's so efficient.
You're right that if "If you'd rather be able to just put the words down on the page" then there are a zillion alternatives. But if you want to be able to put the words down in the context of a structured document, the alternatives narrow pretty fast, depending on what you're doing. In many technical fields (not just computer science, I assure you), LaTeX is either a standard or becoming one.
Your comment about color is interesting. When I was talking with publishers I met with a few production folks from one of the big publishing houses. They looked at my sample chapters and one said that they were the best-looking documents they'd seen in a book proposal. Afterwards I asked others about this comment (which astonished me, since I did nothing special --- I didn't know enough to do anything special!) and I was told that they were probably reacting to the even grey-level of the text, what you're referring to as color. Now that this has been brought to my attention, I notice it all the time. TeX/LaTeX is amazing.
Lots of people have pointed out that Outlook is much more than an e-mail client --- it's really an Exchange client, and the killer app there is shared calendaring and, more generally, shared resources. This is far from trivial to implement. If you are wondering why, I'd suggest reading Dreaming in Code, which follows the Chandler project. (Chandler has often been touted as an "outlook killer".)
Love or hate Microsoft, the simple fact is that no one at the moment has a good response to Exchange and Sharepoint.
I agree with the general thrust of your comments except regarding the quality of Word. I'll repeat a comment I made elsewhere in this thread: In my experience Word simply doesn't work very well. I have tried maybe a dozen times over the last 10 years to use it for a project and every time when the project was done I kicked myself. I think the experience you have is highly dependent on what you're doing and how deeply you're delving into the features, and which features you're using. But for highly structured writing I find Word to be a disaster. This is an empirical result, not a theory. If it works for you, I'm glad to hear it.
I'm an academic and a textbook author so I guess you could call me a professional writer. And you're wrong. My two objections to Word have always been that 1) the file format is fragile -- you can not expect to be able to reliably read a 10-year old document and 2) to my continual astonishment, Word simply doesn't work very well.
WRT #1, version upgrades are a nightmare, and I've seen colleagues lose days of work because of file incompatabilities. I know that if you're *very* careful this won't happen, but you shouldn't have to worry about this. And I want reliable access to things I wrote 10 years ago.
#2: The last time I used Word (2003) for a serious project I used styles for different-level headers and so forth. Everything was auto-numbered and auto-formatted and I was pleased and thought to myself that maybe Word was finally usable. Then I inserted a table of contents and doing this stripped the numbering out of all the headings and eliminated all of my bulleted and numbered lists! I simply couldn't believe it. The truth is that the only people I know *personally* who are happy with Word are people who do not use it's features very deeply. There appear to be plenty of people who use Word in a serious way and who are are happy, but I have yet to meet one personally. (And yes, I am at a big university and I am know lots of people:-)
So what do I use? LaTeX and Emacs. I adopted them both specifically when writing my book because I didn't trust Word. It took a long time to become comfortable with both, but it was one of the best decisions I've ever made. My book is published using TeX and I can use the compositor's files to revise the book. I find this works incredibly well.
I always tell PhD students: you are going to a professional writer so use professional writer's tools: LaTeX and BibTeX. (I *never* recommend Emacs, though I personally love it.)
Ballmer's defense in the article is that Vista has proven valuable to corporations and IT staffs. And this is exactly the point. Microsoft is interested in 1,000+ seat adoptions, where the companies have an IT group that can customize the install, create custom templates and VBA-based applications for the office apps, and where everyone uses exchange and sharepoint.
Microsoft products are broken for everyone else because they're not written for everyone else. Once the latest version of Windows and Office permeate the workplace, Microsoft users in the office drives sales of Microsoft products in the home.
I think you're wrong. VPN is a general concept with many implementations. You may be confusing VPN with PPTP (point-to-point tunneling protocol), which is a Microsoft protocol for implementing VPN. There are many robust implementations of VPN that don't involve Microsoft. For example, OpenVPN is supposed to be quite good.
At my university, Cisco-based VPN is the principal way you authenticate from an off-campus IP address to be able to use campus services. My iMac had no problem connecting to my University network, and the University supports Redhat with a RedHat specific setup script. I have no doubt that the University could have supplied a Ubuntu script, and that if I were more knowledgable I could have constructed one myself.
The bottom line remains that Ubunutu didn't work for me (I literally could not follow the documented steps because the program did not work as documented), and the responses I got were "yeah, that's broken."
People always point to sophisticated software being unavailable on Linux. But the majority of users don't use photoshop. I think the problem is that even basic system services are still flaky, so that even if all you want is to use a browser and basic office applications (which work, in my opinion), linux is still too much of a hassle.
My latest example: I can't get VPN to work as documented on Ubuntu 7.04. When I asked about this on the Ubuntu forums, I got the response that "yeah that's broken, but you can do the following on the command line..." Suspend/hibernate doesn't work reliably. There are too many of these rough edges in the basic plumbing, forget the advanced applications.
The good news is that Linux edges closer every year. Things like sound, TV tuner card, and remote support that were a hassle two years ago now work out of the box because of improved support in the kernel, alsa, and lirc. So it's getting there, but IMO it isn't there yet.
Once the interface to the plumbing is working cleanly and correctly, I think we'll see a big uptick in adoption.
I can't speak about the state of Myth installation *today* but I can speak about it as of one year ago on Debian. It was a PITA. The basic install scripts invoked through apt-get were broken and many of the guides and install tips available online were outdated. Fixing it required a lot of futzing and delving into what the setup was trying to do. I came very close to attempting an install from source.
This was not just my experience -- check out the AMICUS project on sourceforge. OTOH, my impression is that kernel updates may have improved things significantly (pcHDTV cards now are detected automatically, for example).
I say this as someone who now has a correctly-working myth install on two machines. It is a terrific piece of software. But that lightning-shaped scar on my forehead (from banging it against the workbench during my myth install) still throbs every time I see the myth logo.
Keep in mind that Zap2It (and I presume Schedulesdirect) provides more than just the name of a show and the broadcast time. There is a program summary and a program ID sufficient for myth to know which episodes it has recorded. My understanding is that the summary at least is added by the commercial schedule providers.
You're right of course that OTA broadcasters could and should provide free program listings in some common format. However as things have evolved there is so little demand for such a service that there's been no incentive for them to do so.
One assumption in this analysis is that if you write an in-depth article the standard error of its quality will be very low, whereas if you write a blog, the postings will have a high standard deviation. This will in turn lead to a reduction in your perceived value as a source of information if you blog. This argument isn't at all obvious and it depends on assumptions about the quality of your different writings as well as what attracts readers and customers. It also depends on your business model: are you selling writing or services?
Let's say that the long piece you write has a standard deviation that's 1/3 that of the blog posting. (In other words, there's a chance you could write a single piece that damages your brand equity -- Nielsen assumes away this possibility.) If you then write 10 blog pieces, you'll have the same standard deviation for the average as a single long piece. Moreover, the maximum quality of your blog postings will on average be greater than that of your single pieces (because you're drawing from a distribution with a higher standard error). The basic point is that lots of observations may permit folks to infer your quality more accurately. It's not necessary that customers plow through all postings to figure this out --- there are content aggregators (like Slashdot:-) that help separate wheat from chaff.
So what do people evaluate? Your best work? Your average work? The mean quality divided by the standard deviation?
I think Nielsen is correct that you need to think about the impact you're having with what you write, and he may have been correct regarding the advice he gave his world expert, but if you're writing only a few big pieces, you better get them right, or else!
As another poster pointed, out it's quite easy to get a DVD player working under Ubuntu if you don't care about legality, and this is what I found when I googled. I was hoping that I would easily be able to find a US-legal, commercial DVD player for Linux. It sounds like this may be in the works at CNR. Maybe it even exists, but it isn't obvious.
I wasn't clear. User A and B are running. Let's say you're in User A's account. You "switch users". You select User B, enter a password. Sometimes a *second* dialog (with a different look) pops up asking for the exact same password again. So it's not logging in twice that I'm complaining about, it's logging in twice in a row.
Regarding DVDs: I know about the script. But methinks that having to edit the sources file to enable downloading of an illegal script in order to activate what users expect as a standard feature somehow works against this being the year of the Linux desktop.
As TFA suggests, perhaps Linux is poised to take off outside the US. Inside the US, I dunno.
I recently installed Ubuntu 7.04 as a family desktop (dual booting with Windows), on a Dimension 8400. Having reading so much about about what a terrific distribution it is, I decided to experiment. (To provide context, I also have a Debian server that handles backup, slimserver, print sharing, and a Myth backend, and a Debian Myth front end. I'm extremely happy with both.) I've been unpleasantly surprised by Ubuntu as a desktop.
1. Playing DVDs in the US remains a problem. I know that Linspire is going to address this, but this is a huge issue.
2. VPN is a pain. Apparently Network Manager doesn't work right if you have a static IP address! I spent a *lot* of time trying to get VPN to work before I discovered this. Yes, it's a reported bug.
3. Reliable power management, i.e. suspension and hibernation. It's crash city when I suspend or hibernate. Yes I have the latest BIOS. No, I'm not willing to buy a new machine. And yes, I'm sure there are many machines where power management works properly, but I'm also sure there are many machines like mine.
4. The general polish of the Gnome interface is low compared to Windows and OS X. (Yes, I've also looked at KDE.) When I switch users, why do I have to log in twice?
These strike me as all pretty basic issues. I haven't tried to find problems. I've just tried to get the Ubuntu desktop working as a functional equivalent of the Windows desktop. I couldn't do it.
I do see huge progress relative to 5 years ago, but I also see a long way to go.
Perhaps you weren't a user several years ago when Intuit restricted Turbo Tax use to a single machine. They got hell for it. I wrote a letter to the CEO complaining about it and got a phone call from Intuit! Here was Intuit's take on the situation (a statement in May 2003 from Intuit):
"All in all, this was a solid tax season for us," said Bennett. "We grew faster than the industry and were successful in driving customers to new, higher-end offerings, though product activation didn't perform as we'd expected. Intuit has a long heritage of doing right by customers, and some of our customers didn't have the great experience they expect from Intuit. In addition, we didn't get the revenue and profit growth we expected. Therefore, we've decided to discontinue product activation next season." Bennett noted that next year, the company would include in-product technology to unlock marketing and trial versions of TurboTax products.
The critical issue will be the response of businesses who maintain web sites, not Microsoft. It will come down to dollars and cents for the business. If a web site is inaccessible to you because it's using a non-standard technology, complain about it. If there are enough complaints from the right customers, the businesses running the sites will change or microsoft will help the businesses reach linux users.
Perhaps the Novell deal will give Microsoft an incentive to support Linux.
Slashdot Mirror
This was a review for someone who doesn't read slashdot. There wasn't one subtle point. It's well known that google docs require being online and it's also well-known that google is working to change this. Anyone who has spent five minutes examining the capabilities of any of the Google apps knows they fall far short of Microsoft's software. The "power-point killer" designation strikes me as idiotic for a 1.0 version of a new class of software (browser-based presentation creation).
Moreover, suitability is all about what you're presenting. Suppose the reviewer had asked a mathematician to do a comparison of these three presentation packages on the one hand with LaTeX/PDF on the other, for the purposes of giving a mathematical talk. I can tell you from experience that Powerpoint is a joke for this purpose. (I'm not a mathematician but I do include a lot of equations in my slides. LaTeX/PDF rocks.)
Just a few months ago I watched a colleague give a powerpoint presentation and stare in horror at his projected slides because, without realizing it, he had rendered them totally unreadable by using his tablet PC to add last-minute graphics to them (supposedly using the tablet feature as it was intended). You can screw up with Powerpoint too.
This is the very beginning. The interesting thing to speculate about is what the office suite arena will look like three years from now. I'm betting that Google will, at the very least, shake things up a *lot*.
TFA says it was a New York Post dated 9/16/07, with the drive having a manufacture date of 9/4/07. I didn't see a date for the actual event.
You think they had copies of the New York Post at the hard drive factory in Malaysia??
:-)
OTOH, maybe someone there subscribes just to throw investigators off the trail
The antitrust complaint about Microsoft has been that everything was bundled in Windows. Isn't this what the antitrust authorities want them to do? Isn't modularity one of the things that's great about Linux? An unbundled commercial OS would allow Microsoft to have a stripped down version to compete with OLPC, and home and business versions that charged by the feature. You don't want eye candy, you don't pay for eye candy. Not a gamer, you don't pay for DirectX. In theory it seems like a great way to tell developers what to work on (you'll see which features folks are willing to pay for) and a way to offer cheap versions of the OS for the folks that just want e-mail/browsing and expensive versions for those who want everything. You're thinking that they'll try to charge a lot, but really, how much worse could it be than what we have now? Most users pay microsoft a lot every few years.
Now this may be a technical and end-user nightmare, I don't know, but trying to develop an a la carte OS makes a lot of sense to me as a business move.
The same thing is starting to happen with cable more extensively than before: you don't want ESPN or the NFL, you shouldn't have to pay for them.
It's clear you aren't a LaTeX user because you've got it exactly backwards. For many users, the main benefit of LaTeX is the ability to focus on the logical structure and content of the document, and not to have to think at all about appearance, which LaTeX handles automatically. The benefits of auto-numbering, robust cross-referencing, and sophisticated bibliographic handling increase as your document gets longer.
Every Word user I know spends a lot of time messing with formatting. Maybe they don't have to, but they do. A WYSIWYG environment encourages you to do that. You don't have to touch formatting in LaTeX. And I can assure you that someone using Winedt, Vim, or Emacs to create a LaTeX document is not thinking about appearance as they're writing. This is why it's so efficient.
You're right that if "If you'd rather be able to just put the words down on the page" then there are a zillion alternatives. But if you want to be able to put the words down in the context of a structured document, the alternatives narrow pretty fast, depending on what you're doing. In many technical fields (not just computer science, I assure you), LaTeX is either a standard or becoming one.
Your comment about color is interesting. When I was talking with publishers I met with a few production folks from one of the big publishing houses. They looked at my sample chapters and one said that they were the best-looking documents they'd seen in a book proposal. Afterwards I asked others about this comment (which astonished me, since I did nothing special --- I didn't know enough to do anything special!) and I was told that they were probably reacting to the even grey-level of the text, what you're referring to as color. Now that this has been brought to my attention, I notice it all the time. TeX/LaTeX is amazing.
Lots of people have pointed out that Outlook is much more than an e-mail client --- it's really an Exchange client, and the killer app there is shared calendaring and, more generally, shared resources. This is far from trivial to implement. If you are wondering why, I'd suggest reading Dreaming in Code, which follows the Chandler project. (Chandler has often been touted as an "outlook killer".)
Love or hate Microsoft, the simple fact is that no one at the moment has a good response to Exchange and Sharepoint.
I agree with the general thrust of your comments except regarding the quality of Word. I'll repeat a comment I made elsewhere in this thread: In my experience Word simply doesn't work very well. I have tried maybe a dozen times over the last 10 years to use it for a project and every time when the project was done I kicked myself. I think the experience you have is highly dependent on what you're doing and how deeply you're delving into the features, and which features you're using. But for highly structured writing I find Word to be a disaster. This is an empirical result, not a theory. If it works for you, I'm glad to hear it.
I'm an academic and a textbook author so I guess you could call me a professional writer. And you're wrong. My two objections to Word have always been that 1) the file format is fragile -- you can not expect to be able to reliably read a 10-year old document and 2) to my continual astonishment, Word simply doesn't work very well.
:-)
WRT #1, version upgrades are a nightmare, and I've seen colleagues lose days of work because of file incompatabilities. I know that if you're *very* careful this won't happen, but you shouldn't have to worry about this. And I want reliable access to things I wrote 10 years ago.
#2: The last time I used Word (2003) for a serious project I used styles for different-level headers and so forth. Everything was auto-numbered and auto-formatted and I was pleased and thought to myself that maybe Word was finally usable. Then I inserted a table of contents and doing this stripped the numbering out of all the headings and eliminated all of my bulleted and numbered lists! I simply couldn't believe it. The truth is that the only people I know *personally* who are happy with Word are people who do not use it's features very deeply. There appear to be plenty of people who use Word in a serious way and who are are happy, but I have yet to meet one personally. (And yes, I am at a big university and I am know lots of people
So what do I use? LaTeX and Emacs. I adopted them both specifically when writing my book because I didn't trust Word. It took a long time to become comfortable with both, but it was one of the best decisions I've ever made. My book is published using TeX and I can use the compositor's files to revise the book. I find this works incredibly well.
I always tell PhD students: you are going to a professional writer so use professional writer's tools: LaTeX and BibTeX. (I *never* recommend Emacs, though I personally love it.)
Ballmer's defense in the article is that Vista has proven valuable to corporations and IT staffs. And this is exactly the point. Microsoft is interested in 1,000+ seat adoptions, where the companies have an IT group that can customize the install, create custom templates and VBA-based applications for the office apps, and where everyone uses exchange and sharepoint.
Microsoft products are broken for everyone else because they're not written for everyone else. Once the latest version of Windows and Office permeate the workplace, Microsoft users in the office drives sales of Microsoft products in the home.
It's all about the corporate customers.
I think you're wrong. VPN is a general concept with many implementations. You may be confusing VPN with PPTP (point-to-point tunneling protocol), which is a Microsoft protocol for implementing VPN. There are many robust implementations of VPN that don't involve Microsoft. For example, OpenVPN is supposed to be quite good.
At my university, Cisco-based VPN is the principal way you authenticate from an off-campus IP address to be able to use campus services. My iMac had no problem connecting to my University network, and the University supports Redhat with a RedHat specific setup script. I have no doubt that the University could have supplied a Ubuntu script, and that if I were more knowledgable I could have constructed one myself.
The bottom line remains that Ubunutu didn't work for me (I literally could not follow the documented steps because the program did not work as documented), and the responses I got were "yeah, that's broken."
People always point to sophisticated software being unavailable on Linux. But the majority of users don't use photoshop. I think the problem is that even basic system services are still flaky, so that even if all you want is to use a browser and basic office applications (which work, in my opinion), linux is still too much of a hassle.
My latest example: I can't get VPN to work as documented on Ubuntu 7.04. When I asked about this on the Ubuntu forums, I got the response that "yeah that's broken, but you can do the following on the command line..." Suspend/hibernate doesn't work reliably. There are too many of these rough edges in the basic plumbing, forget the advanced applications.
The good news is that Linux edges closer every year. Things like sound, TV tuner card, and remote support that were a hassle two years ago now work out of the box because of improved support in the kernel, alsa, and lirc. So it's getting there, but IMO it isn't there yet.
Once the interface to the plumbing is working cleanly and correctly, I think we'll see a big uptick in adoption.
Thunderbird 1.5 did a great job of converting my (pop) mail from Eudora 7. I had thousands of messages and 100s of megabytes of attachments.
I can't speak about the state of Myth installation *today* but I can speak about it as of one year ago on Debian. It was a PITA. The basic install scripts invoked through apt-get were broken and many of the guides and install tips available online were outdated. Fixing it required a lot of futzing and delving into what the setup was trying to do. I came very close to attempting an install from source.
This was not just my experience -- check out the AMICUS project on sourceforge. OTOH, my impression is that kernel updates may have improved things significantly (pcHDTV cards now are detected automatically, for example).
I say this as someone who now has a correctly-working myth install on two machines. It is a terrific piece of software. But that lightning-shaped scar on my forehead (from banging it against the workbench during my myth install) still throbs every time I see the myth logo.
Keep in mind that Zap2It (and I presume Schedulesdirect) provides more than just the name of a show and the broadcast time. There is a program summary and a program ID sufficient for myth to know which episodes it has recorded. My understanding is that the summary at least is added by the commercial schedule providers.
You're right of course that OTA broadcasters could and should provide free program listings in some common format. However as things have evolved there is so little demand for such a service that there's been no incentive for them to do so.
One assumption in this analysis is that if you write an in-depth article the standard error of its quality will be very low, whereas if you write a blog, the postings will have a high standard deviation. This will in turn lead to a reduction in your perceived value as a source of information if you blog. This argument isn't at all obvious and it depends on assumptions about the quality of your different writings as well as what attracts readers and customers. It also depends on your business model: are you selling writing or services?
:-) that help separate wheat from chaff.
Let's say that the long piece you write has a standard deviation that's 1/3 that of the blog posting. (In other words, there's a chance you could write a single piece that damages your brand equity -- Nielsen assumes away this possibility.) If you then write 10 blog pieces, you'll have the same standard deviation for the average as a single long piece. Moreover, the maximum quality of your blog postings will on average be greater than that of your single pieces (because you're drawing from a distribution with a higher standard error). The basic point is that lots of observations may permit folks to infer your quality more accurately. It's not necessary that customers plow through all postings to figure this out --- there are content aggregators (like Slashdot
So what do people evaluate? Your best work? Your average work? The mean quality divided by the standard deviation?
I think Nielsen is correct that you need to think about the impact you're having with what you write, and he may have been correct regarding the advice he gave his world expert, but if you're writing only a few big pieces, you better get them right, or else!
As another poster pointed, out it's quite easy to get a DVD player working under Ubuntu if you don't care about legality, and this is what I found when I googled. I was hoping that I would easily be able to find a US-legal, commercial DVD player for Linux. It sounds like this may be in the works at CNR. Maybe it even exists, but it isn't obvious.
Thanks for the suggestions.
I wasn't clear. User A and B are running. Let's say you're in User A's account. You "switch users". You select User B, enter a password. Sometimes a *second* dialog (with a different look) pops up asking for the exact same password again. So it's not logging in twice that I'm complaining about, it's logging in twice in a row.
Thanks. There may be something appropriate there, it's hard to tell. The CNR warehouse doesn't have the best descriptions in the world.
Every time I've looked --- including your link --- CNR says "coming soon". Is there something I'm missing?
Regarding DVDs: I know about the script. But methinks that having to edit the sources file to enable downloading of an illegal script in order to activate what users expect as a standard feature somehow works against this being the year of the Linux desktop.
So you mean to tell me that you think I would lug a Dimension 8400 (medium-sized tower case) along on 3 or 4 hours flights? :-)
As TFA suggests, perhaps Linux is poised to take off outside the US. Inside the US, I dunno.
I recently installed Ubuntu 7.04 as a family desktop (dual booting with Windows), on a Dimension 8400. Having reading so much about about what a terrific distribution it is, I decided to experiment. (To provide context, I also have a Debian server that handles backup, slimserver, print sharing, and a Myth backend, and a Debian Myth front end. I'm extremely happy with both.) I've been unpleasantly surprised by Ubuntu as a desktop.
1. Playing DVDs in the US remains a problem. I know that Linspire is going to address this, but this is a huge issue.
2. VPN is a pain. Apparently Network Manager doesn't work right if you have a static IP address! I spent a *lot* of time trying to get VPN to work before I discovered this. Yes, it's a reported bug.
3. Reliable power management, i.e. suspension and hibernation. It's crash city when I suspend or hibernate. Yes I have the latest BIOS. No, I'm not willing to buy a new machine. And yes, I'm sure there are many machines where power management works properly, but I'm also sure there are many machines like mine.
4. The general polish of the Gnome interface is low compared to Windows and OS X. (Yes, I've also looked at KDE.) When I switch users, why do I have to log in twice?
These strike me as all pretty basic issues. I haven't tried to find problems. I've just tried to get the Ubuntu desktop working as a functional equivalent of the Windows desktop. I couldn't do it.
I do see huge progress relative to 5 years ago, but I also see a long way to go.
Perhaps you weren't a user several years ago when Intuit restricted Turbo Tax use to a single machine. They got hell for it. I wrote a letter to the CEO complaining about it and got a phone call from Intuit! Here was Intuit's take on the situation (a statement in May 2003 from Intuit):
"All in all, this was a solid tax season for us," said Bennett. "We grew faster than the industry and were successful in driving customers to new, higher-end offerings, though product activation didn't perform as we'd expected. Intuit has a long heritage of doing right by customers, and some of our customers didn't have the great experience they expect from Intuit. In addition, we didn't get the revenue and profit growth we expected. Therefore, we've decided to discontinue product activation next season." Bennett noted that next year, the company would include in-product technology to unlock marketing and trial versions of TurboTax products.
The critical issue will be the response of businesses who maintain web sites, not Microsoft. It will come down to dollars and cents for the business. If a web site is inaccessible to you because it's using a non-standard technology, complain about it. If there are enough complaints from the right customers, the businesses running the sites will change or microsoft will help the businesses reach linux users.
Perhaps the Novell deal will give Microsoft an incentive to support Linux.