"Computer Science is no more about computers than astronomy is about telescopes."
-Edsger Wybe Dijkstra
I wonder how many times will people will bring this up? A cute quote does not an argument make.
In reference to this quote, which I disagree with, its like astronomers choosing to study a hypothetical universe because the real universe, with "constant" light speed is imperfect and not condusive to cleaner formula.
"The reasons universities don't teach people to most effectively code for a given platform is because it will be next to worthless 10 years down the road. You say they need to know "how computers work", but computers never work the same way one year to the next. The hypothetical high level stuff doesn't change."
I never mentioned "platform" and explicitly did not. Believe it or not, computers have not changed very much in the last 20 years. They've gotten bigger and faster, but they basically work the same way. There have been some cool "new" things, like SMP, NUMA, instruction pipelining, etc but computers, themselves, have remained fairly simple devices upon which more complex paradigms are implemented.
The fact that technologies and theories change over the years is nothing new to science, string theory?
Ignoring that "real" computers have real limitations, and that these limitations require advanced algorithms to over come, is not the answer. In reference to Djikstra, it is like an astronomer ignoring the "real" universe because fixed light speed makes his formula more complex. Real life is a bitch.
"I have sitting on my shelf at home Michael Abrash's infamous "Black Book of Graphics Programming". Is any of the code in there relevant to graphics\animation coding today? Not at all. However, the high level concepts it explains are still useful."
I'm not familiar with that book, but if it is any good and it describes how this is done, it would be a great foundation for technology to come.
The whole "computer engineering" vs "computer science" is a red herring. The difference between engineering and science is that engineering is the application of theory where as the science is the research of the theory.
What "hypothetical computer" proponents fail to understand is that there is "real science" in understanding how "real" computers work. To address Dijkstra, it is like astronomers bitching about the speed of light not being infinite, and preferring to work in a hypothetical universe because it is more perfect.
The limitations of computers and designing algorithms that can accomidate real limitations is the difference between playing with toys and real science.
I'll stop being hard on you, but you REALLY need to get a clue about what you don't know. Believe me, this is not intended as an insult, you are a student, you couldn't possibly know.
>>"I can't say I will learn everything I need to know, but I think it will be pretty close"
You best lose that attitude right now. It is wrong on almost every level. Pretty much every word, as ordered in that sentense, is wrong. You don't really learn until you leave school. School gives you foundation so that you are not utterly lost in the real world.
The smartest thing you wrote was this:
>> "Maybe C and assembly have this, I don't know."
You really don't, and that's sad. Whats even more sad is you don't realize that you do not know. For instance, you wrote: "We made one tool that let you see how many times every line of code was executed," that's called a profiler and they've been around for years (decades?), and, yes, they have them for C and assembly.
Listen, I don't want to play bash the nubee all day, while it is sort of fun, it isn't very nice, your arguments are making my point. You claim that java and.net have better debugging tools, but you don't even know what debugging tools are currently available. The little you think you know is be better than the rest that you don't, and that is not an informed way of approaching problems.
Sorry, I have to laugh at this one (at your expense)
>> "I don't think the universities are the problem, at least not in my experience."
Do you *have* any experience?
>> "The final thing I'd like to point out is that one huge issue with software today is it is bug ridden. How easy something is a test makes a big difference in my opinion. Assembly and C will pretty much always be harder to test than languages like Java and C#."
What makes you think that Assembly and C are harder to test? Have you ever used a debugger? I step through code, I can even see the assembler being executed. There are FAR better tools for debugging "real" compiled code than there are interpreted code. If the program is hung, I can stop a program excatly where it is and see my stack and my actual code path. I can even send it a SIG_INT and break into the debugger. With a core dump I can do a post-mortum on a crash.
>> "The "appeal to an expert" fallacy?"
Absolutely not, simply stating that this is not a new argument and it has continued as long as I've been doing this stuff professionally and the same arguments are the same arguments, they weren't right 20 years ago any more than they are today.
>> "It also means that portability becomes ever harder, as well as adaptability to new hardware."
Nonsense, your "VM" will be written in a low level language.
>> "It's about algorithms. Computers just happen to be the most convienent means for trying them.."
That is the most ignorant statement I have ever read, and it is likely you take offense to my previous statements because they are aimed directly towards your "education." Have you ever written graphic manipulation and animation code? Without full understanding of HOW your computer works, your system won't work acceptably. You need to be aware of virtual memory pages especially when you need to do things like matrix transforms of large blocks of memory. There are whole papers dedicated to the study of this topic. It can mean the difference between smooth graphics and horribly choppy ones.
Obviously, there are classes of applications which do not require being tailored to real-world computers, and in these one should use a higher level language if convenient. That being said, one MUST have the knowledge of how computers work to be able to best judge which is correct.
>> "With the trend towards VM's and virtualization, that "hypothetical" computer comes ever closer."
We will *never* have a hypothetical computer any more than we will have a paperless office or a 4 day work week. As long as there is benefit to coding to the computer, people will take this advantage against their competition. You tell the game industry to use Java!
One may observe this is is absolutely true, your post makes this perfectly clear. (OK, enough insulting)
Nice strawman argument about the assembler, it has, of course, no impact on this discussion.
"More like, 'software sucks today for the same reason it always has -- fossized thinkers can't change to make things easier for those who necessarily follow them.' Ego, no more."
My favorite Einstein quote is "It should be as simple as possible, but no simpler." There is a point at which over simplifying something inherently complex is counter productive.
High level languages have always had a place between fundimentally more complex concepts and mere user interfacing. What I have a problem with is the ignorance of "real" computers in supposed "computer scientists," who, in the course of their careers, will make poorly informed decisions and cause software to suck even more.
We can't forget that software which we write must eventually run on a real computer.
The first mistake: Confusing "compile" performance with execution performance. The job of maping C/C++ code to machine code is trivial.
I've been programming professionally for over 20 years, and for those 20 years, the argument is that computers are now fast enough to allow high level languages and we don't need those dirty nasty assemblers and low level languages.
What was true 20 years ago is still true today, well written code in a low level language tailored to how the computer actually works will always be faster than a higher level environment.
The problem with computer science today is that the professors are "preaching" a hypothetical computer with no limitations. Suggesting that "real" limitations of computers are somehow unimportant.
If computer science isn't about computers, what is it about? I haate that students coming out of universities, when asked about registers and how would they write a multiply routine if they only had shifts and adds, ask "why do I need to know this?"
Software sucks today because software engineers don't understand computers, and that's why languages and environments like Java and.NET will make software worse.
Who says some developers can't get together and just start "www.reallyfreedb.org?" We can download July version of the dataase, host the software, and continue on as if nothing happened.
Anyone interested?
The positive aspect of this is that clueless users may be protected from identity theft.
The downside, and hopefully not the "real reason," is the data collection you speak of. However, learning 192.168.254.0 as a subnet is no great illumination.
It is only if the router has a default password and the officer knows the make and model, or tries several and gets lucky, can they find the native IP address behind the router.
Tell the governor you are boycotting New Hampshire Tourism and products because these abusive and bullying tactics being used by the police are unacceptable in a state that use's the slogan "Live Free or Die."
I called the Nashua Police and asked if the article was true. They said it was. I said that is outrageous that they would do that and that the wiretapping act has a very strong wording that implies that it is for the "interception" of communications.
I asked if they had been contacted by the ACLU yet, at which point the officer hung up on me.
They are a corrupt bunch of bastards. A guy installs a security camera on his property because of crime. A camera readily perchased from Walmart. Because he excersizes his rights to require a warrent and not just let them in, and has evidence that they were abusive to him, they loosely interpret a statute intended to protect the privacy of electronic communications to include home security.
Bastards!!! Fucking bastards. What happened to the constitution? Jesus fucking christ, will someone step up and stop this shit?
I won't post the link, because SCO shouldn't boast "users" because of this, however.....
Some nice items: "You must be a qualified developer with the.Net and Java development platforms, having developed on those platforms. You must be able to demonstrate proficiency with these platform languages and may be asked to submit an example of Java application you or your company has developed."
So, they can say you are not qualified because they have given no criteria about what is or what is not qualified.
You need this much machine:
"a. Memory 768MB, HIGHLY recommend 1GB plus b. Windows XP Pro or Windows 2003 server (Windows XP Home will not work) c. Processor speed - faster the better, at minimum should be Pentium (P4 class) 1.8 - 2.0GHz plus. d. If firewall software is installed, it must be configurable. e. Need to be able to disable anti virus software"
This would require having Windows XP Pro or 2K3 Server, no thanks.
It would cost more than $1000 to get to vegas, stay at a hotel, make sure you have the software & hardware needed.
I am depressed that the current state of science, politics, and media allows, actually promotes, a distrust in science and by being "fair and balanced," gives equal weight to opposing arguments regardless of merrit.
With science, it takes a lot of knowledge and study to fully understand complex topics. Well crafted arguments that depend on people ignorant of the body of facts can sound just as, or more, convincing as those dependent on the actual facts.
Finally, the bullies have found a way to beat the nerds, make truth and facts irrelevent and smart people can no longer win the debate. We are rules by idiots, and they have changed the rules so idiots win.
Short and simple, interpreted/virtual machine code has its place, but it does not make porting easier. Simple programs can be written in *any* language and be easily portable, obviously, some more than others, but all pretty easy.
Once you get in to complexity: memory constraints, performance constraints, large processing, and so on, interpreted and vitual machine systems directly impact the viability of the code. A virtual machine or interpreted system can use more than an order of magnitude more resources than a well coded C/C++ program. If your program is a couple megabytes, well, 20 or so megabytes is not too bad for Java. If your program takes a gig of RAM, then 10 Gig may make your program unworkable.
Then there's porting, don't get me started. Most of my time coding in "portable" java is figuring out the differences in the different virtual machines.
Like I said, an easy program is easy, almost any language and environment will do. More complex programs require more direct control over the actual processing.
As a open source developer, I declare that I will refund 100% of the purchase price if you find a bug in my code.
However, use of this software is provided strictly on an "as is" basis. The user assumes all risk and responsibility for determining the fitness of this software for their application.
The problem with MySQL is that it isn't a good enough SQL dataase to actually run standard SQL benchmarks. I think, maybe, with 5.1 it may actually be able to run something like TPC, but I'm not sure.
Arbitrary benchmarks are meaningless, how many inserts or queries a db can do is meaningless unless something is known about the data and the condidtions. If either the data or conditions of the test are bogus, then the benchmark is bogus.
MySQL is very VERY good at crafting bogus benchmarks that make it look good.
Sorry for the flaim bait, but I can't read something like this without correcting certain errors!
MySQL is a very poor choice for high volume web sites that have a fair number of updates/inserts/deletes as, unless you are using InnoDB, you will be locking the tables and killing performance. If you are going to use InnoDB, you may as well use PostgreSQL for better performance.
I'm sure we've all seen grocklaw crash once or twice because of MySQL.
This is the sort of post that drives me nuts. Because I have a different opinion, you make assumptions about where those opinions comes from.
(1) I am very proficient in Windows. That's your first mistake. One can be proficient in Windows and feel like they are in a prison cell. Windows and Linux are different and things that Linux does Windows does not, and vice vera, but Linux does more of what Windows does.
(2) Choosing a platform I have chosen Linux because Windows is too unstable. This does not disqualify me from unbiased judgement, I have made my unbiased judgement and have chose which I wish to use.
(3) Qualifications I have been a Windows developer since version 1.03 back in the 80s. I have personally, that's my home address, beta tested Windows 3.0, 3.1, 3.1WFW, NT, MSC and MSVC etc. I have published Windoes programming articles.
(4) UNIX I have used UNIX since 1985 using some of the first Sun machines.
(5) Setup is not usability Your example about the cannon printer, so what, for every example you can site about Linux, I can probably site a similar experience on the various Windows version. SETUP is NOT USABILITY. If you bought that cannon printer and it didn't have a driver for your Windows ME laptop, you'd be pissed off at cannon and rightly so.
(6) Uniformity of desktops Sorry, this is FUD. Windows is consistent within a single release at best. Supporting Windows sucks. Unless you do it day in and day out, or have a single homogenous installed base version, you are still searching for the correct dialogy box to configure various features. Sometimes location of control applets changes across service packs.
(7) Conclusion Linux is not hard to use, it is different. Once running and configured, it is 99.99% identical to Mac, Windows, or what ever. You click on an icon to bring up your office program, web brouser, or email program and you work. Done.
Slashdot Mirror
"Computer Science is no more about computers than astronomy is about telescopes."
-Edsger Wybe Dijkstra
I wonder how many times will people will bring this up? A cute quote does not an argument make.
In reference to this quote, which I disagree with, its like astronomers choosing to study a hypothetical universe because the real universe, with "constant" light speed is imperfect and not condusive to cleaner formula.
"The reasons universities don't teach people to most effectively code for a given platform is because it will be next to worthless 10 years down the road. You say they need to know "how computers work", but computers never work the same way one year to the next. The hypothetical high level stuff doesn't change."
I never mentioned "platform" and explicitly did not. Believe it or not, computers have not changed very much in the last 20 years. They've gotten bigger and faster, but they basically work the same way. There have been some cool "new" things, like SMP, NUMA, instruction pipelining, etc but computers, themselves, have remained fairly simple devices upon which more complex paradigms are implemented.
The fact that technologies and theories change over the years is nothing new to science, string theory?
Ignoring that "real" computers have real limitations, and that these limitations require advanced algorithms to over come, is not the answer. In reference to Djikstra, it is like an astronomer ignoring the "real" universe because fixed light speed makes his formula more complex. Real life is a bitch.
"I have sitting on my shelf at home Michael Abrash's infamous "Black Book of Graphics Programming". Is any of the code in there relevant to graphics\animation coding today? Not at all. However, the high level concepts it explains are still useful."
I'm not familiar with that book, but if it is any good and it describes how this is done, it would be a great foundation for technology to come.
The whole "computer engineering" vs "computer science" is a red herring. The difference between engineering and science is that engineering is the application of theory where as the science is the research of the theory. What "hypothetical computer" proponents fail to understand is that there is "real science" in understanding how "real" computers work. To address Dijkstra, it is like astronomers bitching about the speed of light not being infinite, and preferring to work in a hypothetical universe because it is more perfect. The limitations of computers and designing algorithms that can accomidate real limitations is the difference between playing with toys and real science.
Dijkstra is not the last word on the subject. A snappy quote does not an argument make.
This is not a closed debate at all.
"The job of maping C++ code to machine code is trivial."
I didn't mean this in the sense that the task was trivial, but that it has a trivial on development.
"Are you interviewing them for a job coding on a processor that only has shift and add operations?"
Oh the irony. Say, you are working on a display driver, which do you do? divide by 2 or shift once? Why?
I'll stop being hard on you, but you REALLY need to get a clue about what you don't know. Believe me, this is not intended as an insult, you are a student, you couldn't possibly know.
.net have better debugging tools, but you don't even know what debugging tools are currently available. The little you think you know is be better than the rest that you don't, and that is not an informed way of approaching problems.
>>"I can't say I will learn everything I need to know, but I think it will be pretty close"
You best lose that attitude right now. It is wrong on almost every level. Pretty much every word, as ordered in that sentense, is wrong. You don't really learn until you leave school. School gives you foundation so that you are not utterly lost in the real world.
The smartest thing you wrote was this:
>> "Maybe C and assembly have this, I don't know."
You really don't, and that's sad. Whats even more sad is you don't realize that you do not know. For instance, you wrote: "We made one tool that let you see how many times every line of code was executed," that's called a profiler and they've been around for years (decades?), and, yes, they have them for C and assembly.
Listen, I don't want to play bash the nubee all day, while it is sort of fun, it isn't very nice, your arguments are making my point. You claim that java and
Sorry, I have to laugh at this one (at your expense)
>> "I don't think the universities are the problem, at least not in my experience."
Do you *have* any experience?
>> "The final thing I'd like to point out is that one huge issue with software today is it is bug ridden. How easy something is a test makes a big difference in my opinion. Assembly and C will pretty much always be harder to test than languages like Java and C#."
What makes you think that Assembly and C are harder to test? Have you ever used a debugger? I step through code, I can even see the assembler being executed. There are FAR better tools for debugging "real" compiled code than there are interpreted code. If the program is hung, I can stop a program excatly where it is and see my stack and my actual code path. I can even send it a SIG_INT and break into the debugger. With a core dump I can do a post-mortum on a crash.
>> "The "appeal to an expert" fallacy?" Absolutely not, simply stating that this is not a new argument and it has continued as long as I've been doing this stuff professionally and the same arguments are the same arguments, they weren't right 20 years ago any more than they are today. >> "It also means that portability becomes ever harder, as well as adaptability to new hardware." Nonsense, your "VM" will be written in a low level language. >> "It's about algorithms. Computers just happen to be the most convienent means for trying them.." That is the most ignorant statement I have ever read, and it is likely you take offense to my previous statements because they are aimed directly towards your "education." Have you ever written graphic manipulation and animation code? Without full understanding of HOW your computer works, your system won't work acceptably. You need to be aware of virtual memory pages especially when you need to do things like matrix transforms of large blocks of memory. There are whole papers dedicated to the study of this topic. It can mean the difference between smooth graphics and horribly choppy ones. Obviously, there are classes of applications which do not require being tailored to real-world computers, and in these one should use a higher level language if convenient. That being said, one MUST have the knowledge of how computers work to be able to best judge which is correct. >> "With the trend towards VM's and virtualization, that "hypothetical" computer comes ever closer." We will *never* have a hypothetical computer any more than we will have a paperless office or a 4 day work week. As long as there is benefit to coding to the computer, people will take this advantage against their competition. You tell the game industry to use Java!
"Experience doth not beget competence."
One may observe this is is absolutely true, your post makes this perfectly clear. (OK, enough insulting)
Nice strawman argument about the assembler, it has, of course, no impact on this discussion.
"More like, 'software sucks today for the same reason it always has -- fossized thinkers can't change to make things easier for those who necessarily follow them.' Ego, no more."
My favorite Einstein quote is "It should be as simple as possible, but no simpler." There is a point at which over simplifying something inherently complex is counter productive.
High level languages have always had a place between fundimentally more complex concepts and mere user interfacing. What I have a problem with is the ignorance of "real" computers in supposed "computer scientists," who, in the course of their careers, will make poorly informed decisions and cause software to suck even more.
We can't forget that software which we write must eventually run on a real computer.
The first mistake: Confusing "compile" performance with execution performance. The job of maping C/C++ code to machine code is trivial.
.NET will make software worse.
I've been programming professionally for over 20 years, and for those 20 years, the argument is that computers are now fast enough to allow high level languages and we don't need those dirty nasty assemblers and low level languages.
What was true 20 years ago is still true today, well written code in a low level language tailored to how the computer actually works will always be faster than a higher level environment.
The problem with computer science today is that the professors are "preaching" a hypothetical computer with no limitations. Suggesting that "real" limitations of computers are somehow unimportant.
If computer science isn't about computers, what is it about? I haate that students coming out of universities, when asked about registers and how would they write a multiply routine if they only had shifts and adds, ask "why do I need to know this?"
Software sucks today because software engineers don't understand computers, and that's why languages and environments like Java and
Who says some developers can't get together and just start "www.reallyfreedb.org?" We can download July version of the dataase, host the software, and continue on as if nothing happened. Anyone interested?
Your post implies the reason for my caveat.
The positive aspect of this is that clueless users may be protected from identity theft.
The downside, and hopefully not the "real reason," is the data collection you speak of. However, learning 192.168.254.0 as a subnet is no great illumination.
It is only if the router has a default password and the officer knows the make and model, or tries several and gets lucky, can they find the native IP address behind the router.
http://www.egov.nh.gov/governor/goveforms/comments .asp
Tell the governor you are boycotting New Hampshire Tourism and products because these abusive and bullying tactics being used by the police are unacceptable in a state that use's the slogan "Live Free or Die."
I called the Nashua Police and asked if the article was true. They said it was. I said that is outrageous that they would do that and that the wiretapping act has a very strong wording that implies that it is for the "interception" of communications.
I asked if they had been contacted by the ACLU yet, at which point the officer hung up on me.
They are a corrupt bunch of bastards. A guy installs a security camera on his property because of crime. A camera readily perchased from Walmart. Because he excersizes his rights to require a warrent and not just let them in, and has evidence that they were abusive to him, they loosely interpret a statute intended to protect the privacy of electronic communications to include home security.
Bastards!!! Fucking bastards. What happened to the constitution? Jesus fucking christ, will someone step up and stop this shit?
The Nazi's would have loved police like this.
Sorry, I see NOTHING wrong with this as long as it is just for public safety.
I won't post the link, because SCO shouldn't boast "users" because of this, however.....
.Net and Java development platforms, having developed on those platforms. You must be able to demonstrate proficiency with these platform languages and may be asked to submit an example of Java application you or your company has developed."
Some nice items:
"You must be a qualified developer with the
So, they can say you are not qualified because they have given no criteria about what is or what is not qualified.
You need this much machine:
"a. Memory 768MB, HIGHLY recommend 1GB plus
b. Windows XP Pro or Windows 2003 server (Windows XP Home will not work)
c. Processor speed - faster the better, at minimum should be Pentium (P4 class) 1.8 - 2.0GHz plus.
d. If firewall software is installed, it must be configurable.
e. Need to be able to disable anti virus software"
This would require having Windows XP Pro or 2K3 Server, no thanks.
It would cost more than $1000 to get to vegas, stay at a hotel, make sure you have the software & hardware needed.
So, $1000? Not worth it.
I am depressed that the current state of science, politics, and media allows, actually promotes, a distrust in science and by being "fair and balanced," gives equal weight to opposing arguments regardless of merrit.
With science, it takes a lot of knowledge and study to fully understand complex topics. Well crafted arguments that depend on people ignorant of the body of facts can sound just as, or more, convincing as those dependent on the actual facts.
Finally, the bullies have found a way to beat the nerds, make truth and facts irrelevent and smart people can no longer win the debate. We are rules by idiots, and they have changed the rules so idiots win.
Short and simple, interpreted/virtual machine code has its place, but it does not make porting easier. Simple programs can be written in *any* language and be easily portable, obviously, some more than others, but all pretty easy.
Once you get in to complexity: memory constraints, performance constraints, large processing, and so on, interpreted and vitual machine systems directly impact the viability of the code. A virtual machine or interpreted system can use more than an order of magnitude more resources than a well coded C/C++ program. If your program is a couple megabytes, well, 20 or so megabytes is not too bad for Java. If your program takes a gig of RAM, then 10 Gig may make your program unworkable.
Then there's porting, don't get me started. Most of my time coding in "portable" java is figuring out the differences in the different virtual machines.
Like I said, an easy program is easy, almost any language and environment will do. More complex programs require more direct control over the actual processing.
As a open source developer, I declare that I will refund 100% of the purchase price if you find a bug in my code.
However, use of this software is provided strictly on an "as is" basis. The user assumes all risk and responsibility for determining the fitness of this software for their application.
If the servers are gone, shouldn't I get a no conection error and not a polite "site temporarily down. please check back later" message?
I mean, come on, 404 not "check back later."
If they are "required" to sign a pledge, it can be argued that they did no willingly sign the pledge, thus did not, in fact, pledge anything.
The educational system in this country is fucked up.
The problem with MySQL is that it isn't a good enough SQL dataase to actually run standard SQL benchmarks. I think, maybe, with 5.1 it may actually be able to run something like TPC, but I'm not sure. Arbitrary benchmarks are meaningless, how many inserts or queries a db can do is meaningless unless something is known about the data and the condidtions. If either the data or conditions of the test are bogus, then the benchmark is bogus. MySQL is very VERY good at crafting bogus benchmarks that make it look good.
Sorry for the flaim bait, but I can't read something like this without correcting certain errors!
MySQL is a very poor choice for high volume web sites that have a fair number of updates/inserts/deletes as, unless you are using InnoDB, you will be locking the tables and killing performance. If you are going to use InnoDB, you may as well use PostgreSQL for better performance.
I'm sure we've all seen grocklaw crash once or twice because of MySQL.
This is the sort of post that drives me nuts. Because I have a different opinion, you make assumptions about where those opinions comes from.
(1) I am very proficient in Windows.
That's your first mistake. One can be proficient in Windows and feel like they are in a prison cell. Windows and Linux are different and things that Linux does Windows does not, and vice vera, but Linux does more of what Windows does.
(2) Choosing a platform
I have chosen Linux because Windows is too unstable. This does not disqualify me from unbiased judgement, I have made my unbiased judgement and have chose which I wish to use.
(3) Qualifications
I have been a Windows developer since version 1.03 back in the 80s. I have personally, that's my home address, beta tested Windows 3.0, 3.1, 3.1WFW, NT, MSC and MSVC etc. I have published Windoes programming articles.
(4) UNIX
I have used UNIX since 1985 using some of the first Sun machines.
(5) Setup is not usability
Your example about the cannon printer, so what, for every example you can site about Linux, I can probably site a similar experience on the various Windows version. SETUP is NOT USABILITY. If you bought that cannon printer and it didn't have a driver for your Windows ME laptop, you'd be pissed off at cannon and rightly so.
(6) Uniformity of desktops
Sorry, this is FUD. Windows is consistent within a single release at best. Supporting Windows sucks. Unless you do it day in and day out, or have a single homogenous installed base version, you are still searching for the correct dialogy box to configure various features. Sometimes location of control applets changes across service packs.
(7) Conclusion
Linux is not hard to use, it is different. Once running and configured, it is 99.99% identical to Mac, Windows, or what ever. You click on an icon to bring up your office program, web brouser, or email program and you work. Done.