Well, over the last 5-10 years, just from memory: the ParcTab (wireless Palm-like device, years before the Palm), Aspect Oriented Programming, Gyricon (rewritable electric paper), digital rights management, reconfigurable robotics, extensive work on nanotechnology, to name just a few. PARC may not seem as visible anymore because every marketing group claims "groundbreaking technology", but PARC still delivers, and in terms of quality, the place hasn't lost its touch. For an institution with only a few hundred people, it's still just amazing how much is happening there.
While Xerox may have had a tough time making money from the inventions coming out of PARC, the inventions themselves have been hugely successful commercially. Just PARC's contribution to Ethernet, PostScript, OOP, and the Macintosh alone would be enough to justify the entire existence of PARC .
But I also don't think the relationship between Xerox and PARC is much worse than that between other big companies and their research labs. AT&T and IBM research labs both have invented lots of things, and only a small fraction of their inventions have made it into products. Microsoft research is on its way of following the trend. PARC has also contributed tremendously to Xerox's core businesses. What distinguishes PARC is not the fraction of inventions that "got away", but the visible impact some of the inventions have had that did.
But to take a more general perspective, basic computer science research is, unfortunately, in trouble everywhere. In the past, much of it was basically government financed (that's what gave you the Internet and a lot of the other neat computer inventions), and there was some long-term predictability. I view much of the stellar commercial success of Internet and technology companies over the recent years as simply taking government-financed R&D and bringing it to the market. And academia seems to have gotten caught up in the commercial and entrepreneurial maelstroem as well.
The current economy is not such much a testimony to entrepreneurism and private enterprise, but rather to long-term government investments in research and technology. I see nothing wrong with that, but once the government stops financing the kind of research that leads to something like the Internet, the well will run dry in a few years, since private sources clearly aren't taking over this effort.
From the web page, it still seems to require a TRG, so it probably hasn't been ported to the high end Visors. Even if it does, you still have to live within the limitations of uClinux (no MMU and limited multitasking, according to the web page).
Since the iPaq (with Linux, of course) may not be much more expensive once they ship in larger quantities, why not just get that?
In light of hardware like this, with its memory architecture and other limitations, PalmOS is beginning to look like DOS. The iPaq looks like a better deal to me.
Oh yeah, all those jobs I see looking for 2 years work experience, they'll be happy to grab anyone who learned how to code or sysadm from their computer at home.
If you know your stuff, people don't care where you learned it.
Last time I checked, the waiting time for a green card has grown beyond the time you're allowed to stay with an
H1B.
Well, that's wrong on several accounts. First, even under pretty bad conditions, getting a green card takes less than six years. Second, after the initial stages of the green card application have been processed, most people can continue to work indefinitely until the INS gets around to processing their application. There are lots of people that are getting screwed by INS regulations, but on average, the process works in a slow but predictable manner.
Did you ever think about WHY IT workers are mobile? Because they're paid enough to move. H1B's can move, but they're not going to. Read up on H1B to see why.
I have changed jobs on an H1B. It's a quick and simple procedure. There was nothing my former employer could have done to stop me. The notion that H1B's somehow tie employees to employers is fiction.
I think the situation with H1B's isn't quite as bad as you make it out to be: on average, most H1B's don't actually have problems, although the process itself can be nerve wracking and extremely tedious at times.
But I agree that the US should probably abolish H1B's altogether and just go directly to a skill-based immigration (green card) system at whatever level the US feels comfortable with.
That appears to have been the original intent of the US immigration system, and having the current H1B-to-green card system only increases administrative loads and puts both employees and employers at unnecessary risk.
That is technically true. The reality is somewhat different. Here's the reality:
I have changed jobs on an H1B, and it was a quick and easy process.
A person with an H1B visa who is unemployed has 7 days to exit the country or be considered an illegal alien.
The H1B is a work visa. If you don't have an H1B, you can't work, but you can stay in the country on some other visa.
Also, don't forget that anyone applying for a green card must stay with their current employer (often for 3-4 years)
or the green card application process must start all over again.
The green card application used to be fairly
quick. It's only since 1998 that processing times have skyrocketed, due to INS internal administrative problems. This has caused a lot of other serious problems in addition to tying employees to a particular employer for a couple of years. The solution to that is to fix the INS.
And the new H1B bill actually addresses this issue, allowing most people to change jobs while their green card application is pending.
H1B workers pay all the taxes a US citizen pays,
but without being eligible for most of the government benefits and programs they are paying for. They are also not eligible for some tax deductions that citizens can take advantage of.
The problem with this statement is that employers are requiring two years or more experience in the given
technology. Saying that I took the time to study J2EE while I was unemployed is NOT going to get me a job.
No, merely saying it isn't going to get you a job. But being competent and informed during an interview is.
I've interviewed lots of people for technical positions. If you know your stuff, it makes no difference to me where you learned it. In fact, many of the people I have worked with come from non-computer backgrounds are are self-taught. To me, that actually demonstrates interest, breadth, and commitment that goes beyond the easy way of just picking a lucrative major and drifting through school.
If you don't know your stuff, the most impressive background on your resume won't help.
IT companies who import workers have to pay MUCH higher
wages in the US than they do abroad. The incentive to set up low wage programming shops has been
IMMENSE for many years. There are in fact such shops, most notably in India. The problem is that there are
tremendous communications difficulties that make development of a team impossible, and team building is
perhaps the most critical element determining the success or failure of a programming project.
You are completely right: communications and team building are probably the biggest obstacles for why jobs don't move overseas. But the reason why those teams have been built here is because it has been easy for US companies to bring people in to work on existing teams here and because there are some advantages to have the teams here (access to capital markets).
But those are minor advantages. If you make it significantly harder than it is now to bring the workers in, companies will just move whole teams and several levels of management above them elsewhere. And given that many IT workers are already of foreign origin, it wouldn't be hard to find the people to build those teams to move overseas.
The real reason to object to the mass importation of H1B IT workers is that it is ruining the education system
in this country, distorting the employment marketplace, and destroying the attractiveness of technical
careers in the minds of the youth in this country.
If you look at the history of this, the US ruined its educational system long before H1Bs were of much significance. And your comment shows why: many people in the US view education as a means to a high paying career and as a market-driven commodity. (And what kind of IT workers do you expect to get anyway if people just go into the field for the money?)
The US educational system won't get fixed until Americans start valuing education as an end in itself and are willing to shift national priorities to pay for it.
If the ISP started blocking IP addresses, you might have a point. But it's easy enough for anybody to get their mail delivered somewhere else even if they are in the unusual situation that they only have a single ISP with local dial-up.
Reviews are clearly expressions of opinions, and saying, "In my opinion, it stinks." is truthful and does not constitute libel. MAPS is, in effect, saying: "Company X is sending SPAM, where we define SPAM as follows...". If they are truthful about it, they are protected from libel: you can't libel by making truthful statements. And, in fact, under US libel law, MAPS doesn't even have to be entirely truthful. And they certainly don't have to be fair or nice.
Black Ice's claims seem more insidious than that. They say that they are a "legitimate business" and that they are losing money because of MAPS. My first reaction is: well, tough. Businesses should not be legally protected from other people causing them to lose money through free, non-libelous speech. If we go down that road, you can't point out anymore that certain products are harmful or undesirable (but, then, we already have started to go down that road with "food libel laws").
Under libel laws, I think Black Ice would have a tough time making a claim. Most likely, they are claiming something like unfair trade or unfair competition.
Two focussed projects can be managed more effectively and can produce results faster than a single project that wants to be everything to everybody. So, when it happens for technical reasons, forking a project can be quite rational. Besides, the two projects can still share and exchange a lot of code--it just requires a little more hacking.
Another aspect of the market is that companies are in competition with each other. There is a limit to how much a company can pay for IT staff and still stay in business.
If you keep foreign IT workers out of the US to restrict the supply of IT workers in the US and raise salaries, US companies will be less competitive with foreign companies, not only because of the higher domestic salaries, but also because they just freed up lots of IT workers to work abroad.
Competition with other businesses and market elasticity is why companies cannot raise IT salaries arbitrarily, no matter how restricted the supply of IT workers may become.
MCSE, MCDBA, NCA, etc should = LARGE $$, but imported labor keeps that from happening.
MCSE, MCDBA, and NCA about commoditizing basic computer management skills. It's roughly the equivalent of learning some skills as part of working at a food or car maintenance franchise. That's hardly the sort of thing that commands high salaries. In fact, the thing that makes Microsoft software so popular with management is the promise that it makes IT functions supposedly easy enough so that anybody can be trained to perform them.
Of course, just like MacDonald's didn't eliminate the need for good, experienced chefs, Microsoft hasn't eliminated the need for good, experienced IT staff. At best, you can use an MCSE as a stepping stone to getting some real experience. If you don't quickly broaden your horizons beyond MCSE, you are likely going to see yourself out of a job pretty soon.
Three points that Matloff and other people who take his position make deserve some comment:
Matloff claims that companies are unwilling to pay the cost for retraining employees or new hires in the latest technologies. Actually, I think that's a misinterpretation of what is really going on. The cost of retraining isn't particularly high. But the fact that a new hire needs retraining is a potential red flag.
These days, you can learn the hottest technologies on your PC at home, at virtually no cost. A programmer who has been out of work and hasn't even bothered as much as to learn Perl or Java does not seem to be a very attractive candidate, simply because they don't seem very motivated or interested in the job.
In any case, the argument about retraining is hypothetical, since the situation that someone applies and asks for retraining is pretty rare. If anybody shows even that much initiative, in the current job climate, they will probably get hired.
The argument that people get hired on H1B for slave labor gets dragged out again and again. Well, it's bogus. H1B's have always allowed job changes. And for the last 6-7 years, under the new immigration laws, many employers have sponsored their H1B employees immediately for a green card. And with the pending immigration bill, H1B's are even more easily transferred, and foreign employees are even less tied to their companies. I also have never seen any wage differential between H1B workers at the companies where I have worked (although it doubtlessly occurs somewhere).
The argument for keeping foreign IT workers out of the US is that that would allow US workers to take those jobs, or at least increase demand and raise wages. But that is illusory. If the foreign IT workers can't come here, they'll simply work for a subsidiary or contracting firm in Europe or Asia. The US doesn't gain any jobs from that, they merely lose the tax revenue. And what do you think competing Indian wages is going to do to the wages of US programmers?
The US neither has a monopoly nor a god-given right to IT jobs. The IT industry is flourishing here because the US has been fairly open to foreign IT workers and because IT workers are mobile. If the US were to close the door, the IT workers would simply go elsewhere.
So, don't spoil a good thing. Both the US and its citizens, and foreign workers coming to the US benefit greatly from the fairly liberal US immigration policies. The only people who have cause to complain are the countries that all those skilled, educated IT workers leave.
Matloff is having a great time generating lots of publicity with funny statistics and describing individual cases.
With all these supposedly out of work, over the hill, US-born computer programmers, where are the resumes of these people? People in SV are desparately looking for qualified programmers. Yet, they don't seem to get any applicants, and they are nowhere to be seen online either. Why doesn't Matloff take some pro-active measures and actually create a job web site for the people he claims to care so much about?
My challenge to Matloff is this: put up or shut up. If there are large crowds of qualified, unemployed programmers, do everybody a favor and publish their resumes so that employers can find them.
I believe the Coda file system implementation for Linux allows you to hook into open/close requests. This ought to allow the creation of user mode tools for transparent file-level encryption and may be a much better approach than the current NFS-based or kernel-based methods.
There is actually a Perl user mode file system that's based on those hooks (PerlFS?).
That's thre right way only if you subscribe to the kitchen sink approach to system design.
What Linux really needs is better mechanisms for stacking file systems and implementing file systems in user code. Then, different people can implement what they need on top of a standard Linux system.
The best way of dealing with big projects is to not let them become big in the first place.
Mozilla should not have been a single project of tightly integrated GUI code and other bits and pieces. It really ought to have been five or six independent open source projects with a few, simple, well-defined interfaces.
The same is true for StarOffice: word processing, spread sheet, presentation, and other bits and pieces should really be stand-alone parts.
In addition to breaking projects up into smaller pieces, they should also use languages and tools that keep them small. If it becomes a 400kloc project in C++, rather than suffering through that, pick some better language that turns it into a 40kloc project.
The problem is that the US legal system is, at its core, about harsh retribution not rehabilitation. And the kind of punishments the US legal system can dole out are very serious: imprisonment in an overcrowded, disease-infested prison system that does not meet international human rights standards, loss of property and income without any kind of social safety net, or simply outright execution. People are right to be suspicious of giving such a judicial system extensive investigative powers in addition to its already extensive powers of punishment.
If the US justice system ever became oriented more towards rehabilitation and helping the offender reintegrate into society, then, and only then, would extensive investigative powers be justified. (That is the route many European countries have followed in the past, and their lower crime rates seem to justify it.)
So, people probably would be happy to give the US government broad new investigative powers if the US government abolishes the death penalty, decriminalizes drug posession, and shifts emphasis from retribution to rehabilitation. Otherwise, giving it both extensiv power to punish and extensive power to investigate means going down the road towards a police state.
The mechanisms anybody chooses for name resolution on their computer is entirely up to them. ICANN happens to have something to say about the mechanism most people choose right now, but you can pick an entirely different name resolution mechanism if you like.
RealNames was, of course, an attempt to do this. But handing keywords from a messy standards body to a single corporation with a non-federated protocol is clearly not the answer.
What we need to do is build more robust clients for name resolution that can integrate information from a variety of sources. Clients that take into account individual user's preferences, as well as context (e.g., host names inside web pages should usually refer to the ICANN-administered world).
Keep in mind that JFS does not protect your data integrity (it only journals file system structures), and it also doesn't protect you against hardware failure, bugs, or misconfiguration.
So, for anything that is critical or costly, you need redundant hardware and replication anyway. Once you have that, you are already protected against downtime from running an fsck should the need ever arise. There is no need to slow down your (replicated) systems additionally with JFS for a very marginal benefit.
The Aibo seems like a pretty lightweight design to me, and it doesn't seem to support much programming either.
For someone with geeky leanings, building or buying a mobot platform would give you something that's just as good around the house, is a lot more robust, and allows lots of actually interesting things to be done with it.
With tools like SWIG and gcj, Java and C++ easily work together.
But you don't get something for nothing. If you put C/C++ into Java or C#, you lose the runtime safety and security guarantees. Or, if you try to build special versions of C/C++ to preserve safety, you lose C/C++'s efficiency and control over memory usage. Microsoft has the same problem as anybody else: it's a fundamental mismatch between the design of C++ and languages like Java.
However, a safe-but-slower version of C++ make sense for Microsoft because they have been using C++ as a high-level applications programming language for so long. But people don't get the "advantages of both" that way because that kind of applications code never took advantage of C++'s strengths in the first place.
.NET is basically a Java clone. Since Java's success demonstrates that people care about being able to run the same application on different platforms, obviously Microsoft's answer needs to do the same thing.
It makes perfect sense if Microsoft wants total market dominance: they have something that, on paper, looks like a good response to Java. At the same time, they get their proprietary software onto other platforms and drive competing application vendors out of business. And since they control.NET, they can always stop support for other platforms when it suits them.
Of course, the problem with this is that Microsoft will probably have trouble delivering a good version of.NET (it's even harder than a good JVM) and that many peopel feel that their applications and web software sucks. So, overall, it probably won't make much of a difference either way.
Slashdot Mirror
Well, over the last 5-10 years, just from memory: the ParcTab (wireless Palm-like device, years before the Palm), Aspect Oriented Programming, Gyricon (rewritable electric paper), digital rights management, reconfigurable robotics, extensive work on nanotechnology, to name just a few. PARC may not seem as visible anymore because every marketing group claims "groundbreaking technology", but PARC still delivers, and in terms of quality, the place hasn't lost its touch. For an institution with only a few hundred people, it's still just amazing how much is happening there.
But I also don't think the relationship between Xerox and PARC is much worse than that between other big companies and their research labs. AT&T and IBM research labs both have invented lots of things, and only a small fraction of their inventions have made it into products. Microsoft research is on its way of following the trend. PARC has also contributed tremendously to Xerox's core businesses. What distinguishes PARC is not the fraction of inventions that "got away", but the visible impact some of the inventions have had that did.
But to take a more general perspective, basic computer science research is, unfortunately, in trouble everywhere. In the past, much of it was basically government financed (that's what gave you the Internet and a lot of the other neat computer inventions), and there was some long-term predictability. I view much of the stellar commercial success of Internet and technology companies over the recent years as simply taking government-financed R&D and bringing it to the market. And academia seems to have gotten caught up in the commercial and entrepreneurial maelstroem as well.
The current economy is not such much a testimony to entrepreneurism and private enterprise, but rather to long-term government investments in research and technology. I see nothing wrong with that, but once the government stops financing the kind of research that leads to something like the Internet, the well will run dry in a few years, since private sources clearly aren't taking over this effort.
Since the iPaq (with Linux, of course) may not be much more expensive once they ship in larger quantities, why not just get that?
In light of hardware like this, with its memory architecture and other limitations, PalmOS is beginning to look like DOS. The iPaq looks like a better deal to me.
If you know your stuff, people don't care where you learned it.
Last time I checked, the waiting time for a green card has grown beyond the time you're allowed to stay with an H1B.
Well, that's wrong on several accounts. First, even under pretty bad conditions, getting a green card takes less than six years. Second, after the initial stages of the green card application have been processed, most people can continue to work indefinitely until the INS gets around to processing their application. There are lots of people that are getting screwed by INS regulations, but on average, the process works in a slow but predictable manner.
Did you ever think about WHY IT workers are mobile? Because they're paid enough to move. H1B's can move, but they're not going to. Read up on H1B to see why.
I have changed jobs on an H1B. It's a quick and simple procedure. There was nothing my former employer could have done to stop me. The notion that H1B's somehow tie employees to employers is fiction.
But I agree that the US should probably abolish H1B's altogether and just go directly to a skill-based immigration (green card) system at whatever level the US feels comfortable with.
That appears to have been the original intent of the US immigration system, and having the current H1B-to-green card system only increases administrative loads and puts both employees and employers at unnecessary risk.
I have changed jobs on an H1B, and it was a quick and easy process.
The H1B is a work visa. If you don't have an H1B, you can't work, but you can stay in the country on some other visa.
The green card application used to be fairly quick. It's only since 1998 that processing times have skyrocketed, due to INS internal administrative problems. This has caused a lot of other serious problems in addition to tying employees to a particular employer for a couple of years. The solution to that is to fix the INS.
And the new H1B bill actually addresses this issue, allowing most people to change jobs while their green card application is pending.
H1B workers pay all the taxes a US citizen pays, but without being eligible for most of the government benefits and programs they are paying for. They are also not eligible for some tax deductions that citizens can take advantage of.
No, merely saying it isn't going to get you a job. But being competent and informed during an interview is.
I've interviewed lots of people for technical positions. If you know your stuff, it makes no difference to me where you learned it. In fact, many of the people I have worked with come from non-computer backgrounds are are self-taught. To me, that actually demonstrates interest, breadth, and commitment that goes beyond the easy way of just picking a lucrative major and drifting through school. If you don't know your stuff, the most impressive background on your resume won't help.
IT companies who import workers have to pay MUCH higher wages in the US than they do abroad. The incentive to set up low wage programming shops has been IMMENSE for many years. There are in fact such shops, most notably in India. The problem is that there are tremendous communications difficulties that make development of a team impossible, and team building is perhaps the most critical element determining the success or failure of a programming project.
You are completely right: communications and team building are probably the biggest obstacles for why jobs don't move overseas. But the reason why those teams have been built here is because it has been easy for US companies to bring people in to work on existing teams here and because there are some advantages to have the teams here (access to capital markets).
But those are minor advantages. If you make it significantly harder than it is now to bring the workers in, companies will just move whole teams and several levels of management above them elsewhere. And given that many IT workers are already of foreign origin, it wouldn't be hard to find the people to build those teams to move overseas.
The real reason to object to the mass importation of H1B IT workers is that it is ruining the education system in this country, distorting the employment marketplace, and destroying the attractiveness of technical careers in the minds of the youth in this country.
If you look at the history of this, the US ruined its educational system long before H1Bs were of much significance. And your comment shows why: many people in the US view education as a means to a high paying career and as a market-driven commodity. (And what kind of IT workers do you expect to get anyway if people just go into the field for the money?)
The US educational system won't get fixed until Americans start valuing education as an end in itself and are willing to shift national priorities to pay for it.
If the ISP started blocking IP addresses, you might have a point. But it's easy enough for anybody to get their mail delivered somewhere else even if they are in the unusual situation that they only have a single ISP with local dial-up.
Black Ice's claims seem more insidious than that. They say that they are a "legitimate business" and that they are losing money because of MAPS. My first reaction is: well, tough. Businesses should not be legally protected from other people causing them to lose money through free, non-libelous speech. If we go down that road, you can't point out anymore that certain products are harmful or undesirable (but, then, we already have started to go down that road with "food libel laws").
Under libel laws, I think Black Ice would have a tough time making a claim. Most likely, they are claiming something like unfair trade or unfair competition.
Two focussed projects can be managed more effectively and can produce results faster than a single project that wants to be everything to everybody. So, when it happens for technical reasons, forking a project can be quite rational. Besides, the two projects can still share and exchange a lot of code--it just requires a little more hacking.
If you keep foreign IT workers out of the US to restrict the supply of IT workers in the US and raise salaries, US companies will be less competitive with foreign companies, not only because of the higher domestic salaries, but also because they just freed up lots of IT workers to work abroad.
Competition with other businesses and market elasticity is why companies cannot raise IT salaries arbitrarily, no matter how restricted the supply of IT workers may become.
MCSE, MCDBA, and NCA about commoditizing basic computer management skills. It's roughly the equivalent of learning some skills as part of working at a food or car maintenance franchise. That's hardly the sort of thing that commands high salaries. In fact, the thing that makes Microsoft software so popular with management is the promise that it makes IT functions supposedly easy enough so that anybody can be trained to perform them.
Of course, just like MacDonald's didn't eliminate the need for good, experienced chefs, Microsoft hasn't eliminated the need for good, experienced IT staff. At best, you can use an MCSE as a stepping stone to getting some real experience. If you don't quickly broaden your horizons beyond MCSE, you are likely going to see yourself out of a job pretty soon.
These days, you can learn the hottest technologies on your PC at home, at virtually no cost. A programmer who has been out of work and hasn't even bothered as much as to learn Perl or Java does not seem to be a very attractive candidate, simply because they don't seem very motivated or interested in the job.
In any case, the argument about retraining is hypothetical, since the situation that someone applies and asks for retraining is pretty rare. If anybody shows even that much initiative, in the current job climate, they will probably get hired.
The US neither has a monopoly nor a god-given right to IT jobs. The IT industry is flourishing here because the US has been fairly open to foreign IT workers and because IT workers are mobile. If the US were to close the door, the IT workers would simply go elsewhere.
So, don't spoil a good thing. Both the US and its citizens, and foreign workers coming to the US benefit greatly from the fairly liberal US immigration policies. The only people who have cause to complain are the countries that all those skilled, educated IT workers leave.
With all these supposedly out of work, over the hill, US-born computer programmers, where are the resumes of these people? People in SV are desparately looking for qualified programmers. Yet, they don't seem to get any applicants, and they are nowhere to be seen online either. Why doesn't Matloff take some pro-active measures and actually create a job web site for the people he claims to care so much about?
My challenge to Matloff is this: put up or shut up. If there are large crowds of qualified, unemployed programmers, do everybody a favor and publish their resumes so that employers can find them.
There is actually a Perl user mode file system that's based on those hooks (PerlFS?).
What Linux really needs is better mechanisms for stacking file systems and implementing file systems in user code. Then, different people can implement what they need on top of a standard Linux system.
Mozilla should not have been a single project of tightly integrated GUI code and other bits and pieces. It really ought to have been five or six independent open source projects with a few, simple, well-defined interfaces.
The same is true for StarOffice: word processing, spread sheet, presentation, and other bits and pieces should really be stand-alone parts.
In addition to breaking projects up into smaller pieces, they should also use languages and tools that keep them small. If it becomes a 400kloc project in C++, rather than suffering through that, pick some better language that turns it into a 40kloc project.
If the US justice system ever became oriented more towards rehabilitation and helping the offender reintegrate into society, then, and only then, would extensive investigative powers be justified. (That is the route many European countries have followed in the past, and their lower crime rates seem to justify it.)
So, people probably would be happy to give the US government broad new investigative powers if the US government abolishes the death penalty, decriminalizes drug posession, and shifts emphasis from retribution to rehabilitation. Otherwise, giving it both extensiv power to punish and extensive power to investigate means going down the road towards a police state.
RealNames was, of course, an attempt to do this. But handing keywords from a messy standards body to a single corporation with a non-federated protocol is clearly not the answer.
What we need to do is build more robust clients for name resolution that can integrate information from a variety of sources. Clients that take into account individual user's preferences, as well as context (e.g., host names inside web pages should usually refer to the ICANN-administered world).
So, for anything that is critical or costly, you need redundant hardware and replication anyway. Once you have that, you are already protected against downtime from running an fsck should the need ever arise. There is no need to slow down your (replicated) systems additionally with JFS for a very marginal benefit.
For someone with geeky leanings, building or buying a mobot platform would give you something that's just as good around the house, is a lot more robust, and allows lots of actually interesting things to be done with it.
But you don't get something for nothing. If you put C/C++ into Java or C#, you lose the runtime safety and security guarantees. Or, if you try to build special versions of C/C++ to preserve safety, you lose C/C++'s efficiency and control over memory usage. Microsoft has the same problem as anybody else: it's a fundamental mismatch between the design of C++ and languages like Java.
However, a safe-but-slower version of C++ make sense for Microsoft because they have been using C++ as a high-level applications programming language for so long. But people don't get the "advantages of both" that way because that kind of applications code never took advantage of C++'s strengths in the first place.
It makes perfect sense if Microsoft wants total market dominance: they have something that, on paper, looks like a good response to Java. At the same time, they get their proprietary software onto other platforms and drive competing application vendors out of business. And since they control .NET, they can always stop support for other platforms when it suits them.
Of course, the problem with this is that Microsoft will probably have trouble delivering a good version of .NET (it's even harder than a good JVM) and that many peopel feel that their applications and web software sucks. So, overall, it probably won't make much of a difference either way.