I refuse to let 95 or 98 in my office. Not even a CD-ROM. NT 3.51 is a bit more stable than 4.0 in my experience, but doesn't handle big hard drives, so I moved my data crunching system up to 4.0 and it has basically been fine.
What really sucks is having to reboot whenever you install a new app with one of Micro~1's crappy shared DLLs. The latest offender was Foxpro 6, which I only need for some light file management, and which had me reboot three times (twice for the mandatory install of evilbad IE which I never use). Rebooting for network changes is bad enough, but for mere mortal applications??
Oh, it was fine, Brian (but you know they will always try to trap you that way:)
I liked the emphasis on the "whole product" theme from Geoffrey Moore's book. Too many people read Crossing the Chasm and miss that, yet it's really the most important thing among many interesting things he says.
"Open Source" as a whole is starting to cross the chasm, and you can see how that affects the view of those who judge it from the outside. This is why KDE/GNOME, Samba, 24x7 support, boxes with red hats, VMware and so on are attracting attention more than equally interesting developments like ipchains, Beowulf and the GIMP. The former have more direct relevance to the departmental desktop managers that the slick trade magazines talk to. And that is one of the early steps in crossing the chasm to the mass market.
Now, it may be that Wintel PCs and to a lesser degree Macs continue to dominate the market, but just as there was always a large slice of personal vehicle drivers who chose pickup trucks and vans, it may not be long before the SUV of operating systems comes along to occupy the same kind of middle ground (with less pollution and more driving stability, to be sure!).
The branding issue is of less importance, though it was quite prominent in the published part of the interview. Branding is a given, not an option, in my view, but it got the (admittedly limited) attention of the marketeering world in the last couple of years.
Strong branding doesn't save your butt if the product isn't good or isn't differentiated. Remember the Peanut (PCjr)? Hmm. How about New Coke? That was a good one. There was certainly no problem with Prodigy's brand identity, or dBase, or OS/2 (sorry).
Again, this is not to dismiss the importance of branding, but just to suggest it's a necessary but not sufficient condition for success. One of the interesting things is how the Open Source/free software world has evolved a sort of organic approach to branding. I mean, that Penguin. Or, that slash and that dot.
And don't forget that despite a brilliant branding strategy,
Um, how can I say "get a clue" nicely. We have plenty of choices here in Portland, Oregon. Bill Gates is not interested in wiping out Linux and couldn't do so even if he tried. Paul Allen, not Bill Gates, is making the big investments in cable systems these days.
I offer you these Free Clues in spite of the fact that
You know, I have been very good in restraining myself, but this notion that RMS is some kind of "communist" is the worst kind of infantile name-calling. It shows very clearly that those who toss out such insults so casually have no concept of what RMS is trying to do and even less about what "socialist" or "communist" means.
RMS is a classic "liberal" in the original 18th century Enlightenment sense of that word.
You can agree or disagree with his project (and I find myself on one side or the other depending on the situation), but the name-calling only obscures the fundamental importance of the issues he has been raising.
After slogging through all this, I'm annoyed -- no, I'm irate -- that so much of the discussion has been sidetracked by a gale-force Slashdot Paranoia storm.
Sure, Microsoft is blah blah blah [insert favorite pejoratives here]. As a matter of fact, they are not the devil incarnate and some of their stuff does work pretty well. You can thank them for the fact that the Win32::ODBC module is the backbone of many a fine Web site (or didn't you notice that ODBC, a Microsoft project, is a protocol that despite some faults Doesn't Suck).
I'm also appalled that so many of the commenters here didn't even bother to check basic facts about Perl and its history in the DOS/Win realm. I've been using it since I was running everything under DOS 5.0 in 1992 (still have the old.exe of Perl 4.0.1.6, ported by Diomidis Spinellis and dated 2/23/1992 -- it would not be exaggerating to say that it changed my life almost more than any other single program).
I've also used the various ActiveState Win32 Perl releases for years with excellent results. For example, I use it to munch 2-gigabyte raw dumps from databases and produce crosstabs and statistics two orders of magnitude faster than the databases can do so themselves. One of my main reasons in moving to Linux, actually, was to double that performance, but it's certainly not bad under NT.
I'm glad tchrist showed up to correct some of the more egregious misunderstandings about Perl. The evolution of Perl between 1994 and 1997 from basically a fairly limited domain of text processing to a platform for all kinds of useful things, through the addition of object oriented structures, the revision of the library approach into the full-blown module environment we now have, and so forth, ought to be a classic study in how computing tools can evolve in a positive sense rather than a bloatware sense.
If Microsoft wants to hook into that in a bigger way, that's a very good thing. Perl makes my life in the NT context bearable. However, I don't want to see it fork from the standard distribution in any significant way. Larry understates the importance of reunifying the code bases between the standard development tree and the ActiveState version with the One Perl initiative. And he also plays down the fact that it was not a guaranteed win. But nonetheless it did work out.
So now consider the real story: Microsoft has conceded where the locus of authority is for Perl. They are promising in public to play nice. Their tendency not to do so is well understood and their activities will be thoroughly scrutinized.
In other words, this is precisely the opposite of what happened with Java. And it is no surprise, given that Perl is open source and Java is not.
And that is the most important lesson to be drawn from this.
I don't mind Microsoft throwing in some things to optimize local processing, as long as they are themselves open source and don't fork my code. Because the bottom line is that if I can't run the basic stuff across platforms, their optimizations will disappear from my code first. That is market discipline Microsoft has never had to face, and they may have to grow up (maybe just a bit, but grow up nevertheless) as a result.
Perl is a remarkable accomplishment. Learning more about its history, philosophy and development is a rewarding experience. And, golly guess what, it's all out there readily available to be discovered.
Enron's net operations are headquartered here in Portland, Oregon, where I live, and I buy my electricity from them (through subsidiary company Portland General Electric which they swallowed in 1997), which has more than a little to do with the location of their NOC here.
What we have learned about Enron is that they are an aggressive and less than trustworthy organization. They are dramatically expanding their business lines beyond the core of natural gas trading and natural gas transmission into electric markets, municipal water systems and now bandwidth. I would not be surprised to see them take a leap at telecom next, because their effort to take over the American electric industry stalled after they bought PGE.
Enron trumpets its reputation as an ethical multinational business. They belong to various councils and associations promoting good business practice. However, the record is actually quite different.
Last month, their new natural gas power plant at Dabhol, India (Maharashtra State, southwest of Bombay) opened up. This is not only the largest single power plant in the country, it has also been the center of a 7-year political struggle involving a close (some say corrupt) relationship with leading politicians and political parties, and repression against the thousands of local residents whose homes and livelihoods have been disrupted by the construction of the plant and its land and water needs.
India is not, as some people think, a moribund "third world" country. It has titanically huge problems but it is also the second largest nation in the world (and will possibly surpass China within a few decades), and the largest democracy on the planet. It has the ill-famed "streets of Bombay and Calcutta," and it also has more PhDs than any country on earth. (And lately, it has atomic bombs.)
What Enron did in India is hardly a matter of just pushing a few backward natives out of their huts. The Dabhol plant has been a major factor in national politics, the subject of parliamentary and Supreme Court inquiries, the focus of demonstrations where thousands were arrested on Enron's behest, and where Enron even paid for off-duty police to harass and assault opponents of the plant.
This is hardly grapevine stuff. It is documented in great detail in a full-scale report by Human Rights Watch issued earlier this year.
When speaking of Enron, remember that they have one of the most aggressive self-promotion efforts of any company involved in utility operations, whether it be gas, electric, water or bits.
I use a venerable DOS shareware program called qe.exe (QuickEdit 2.06). It's dated 5/19/1988 and takes all of 48K. It always runs, never complains, is fast and pretty easy to use. There are a couple things I wish it did, but since I can run it on any M$ box from DOS 2.11 to NT 4.0, it's just fine.
There is a current version called The SemWare Editor which has a few more bells and whistles.
Um, why does everything have to be denominated in dollars? Or do you just assume that enthusiasm has to be paid for any more on the Internet?
In a world where Marimba has a market cap of maybe $90 million at the moment, I find it laughable that CT is constantly being accused of highlighting companies, web sites, products or whatever, in exchange for filthy lucre.
But hey, when Lucas gives/. its well deserved check for promoting the Phantom Menace, then let me know.
If there's one thing we learned in April, it's that benchmarks and Orange Book security levels are both virtually guaranteed to be misused in marketing. And I'm not just pointing the finger at Micros~1.
The people who design planes are not the best ones to ask about flying them.
It's all well and good for Linus and Alan Cox to help tune things for a benchmark. But actually, they are not the best people to do so. The best people to do so are those with domain experience -- that is, people who implement and manage large systems using Linux.
And again, let me warn everyone: benchmarks are a moral hazard.
A very clever idea from a very clever man. But I'm afraid my view of Lessig is that he's too clever by half, which is readily apparent after you start reading his proposed "solution" to the problems posed by the CDA.
He keeps trying to start up bandwagons that will name their tune for him, but they keep sputtering and stalling. Remember the "New Chicago School" notions that were supposed to supplant the "old school" law-and-economics approach of Posner and others at the University of Chicago? Well, Posner's crowd keeps flattening people everywhere with its mechanistic view of how the world works (which generally somehow, I really don't know how, always seem to end up supporting those who already have the greater economic power in any given situation -- but I digress).
And "New Chicago"? Wellll, New Chicago looks kind of old school after all, if you look at the table of contents of some of Lessig's recent confabs, such as this.
I'm sure Lessig is far smarter than I'll ever be. But he strikes me as far more a politician than an innovator or synthesizer in the field of law, and this "open law" thing looks like just another attempt to grab the limelight.
What is missing in all this -- and I'm afraid in ESR's rather sloppy summary as well -- is any sense of whether benchmarking is reliable in the least for analyzing application processing. Things are bad enough with processor benchmarking -- remember the tuning code in Quake?
In databases, things are much worse. The TPC has been wrestling with these issues for a decade now and still doesn't really have a good handle on it. It is too easy to put your thumb on the benchmarking scale without anyone noticing, and make the results go the way you want. This is true even if the vendors themselves do the tuning.
Even if you can equalize the platforms, it still gets down to issues like the mix of instructions in the test suite. The TPC has wrestled with that one for years, trying to avoid tipping the balance toward any given vendor.
Schematically, of course, Web servers are database servers (aside from the issue that they may request data from an actual database).
Frankly, I'm even more appalled at Mindcraft (and by extension, Microsoft) for pushing this "study" the way they did. It was at least borderline unethical, given their admittedly lame-if-you-are-being-kind-about-it effort to equalize tuning between the two systems.
The biggest drawback to NT 4 (SPx) is that the security has gone from terrible to merely bad.
It still amazes me that Microsoft, the ultimate "grab the good ideas from the competition and bundle them into the OS" outfit, has not seen fit to install a decent port monitor, much less something like tcp wrappers. This borders on criminal negligence, in my view.
All the same, I use NT for some fairly big database projects (some in the multi-gigabyte data range) and it is solid as long as it's not running memory leakers like browsers or (yikes) Excel.
I don't agree with the idea that W95 runs 16 bit apps better than NT. Quite the opposite, from my view, and I run a lot of 16-bit stuff on NT (Word 2.0, for example:) The problem I do have is with 32-bit apps that won't run on NT 3.51, so I have NT 4.0 on my big database server.
I used W95 when it was in beta for a week in May 1995, erased it and reinstalled NT 3.51. I have never regretted it. I avoid even helping my friends fix their W95/98 problems. What should take 10 minutes often becomes an open-ended 8 to 12 hour repair job. I have never had any real trouble with NT, it's just slower and a little less stable than I'd like, and the security is not nearly as good as it should be. My big Perl data jobs run twice as fast under Linux as NT on the same box (even without tuning).
First of all, a small correction: XENIX was an early Microsoft product, but their first was Tiny BASIC.
A more serious matter: NT is not based on Unix, its roots are more in VMS. Read Helen Custer's excellent Inside Windows NT which explains the history in interesting detail. I actually like NT the operating system; it's the Windows architecture and especially the W95-look-and-feel subsystems that run on top of it that I find excruciating. I still use NT 3.51 as my primary desktop because the interface doesn't get in my way as much. (Fear not, Linux fans, I already have Debian running and it will be where I "live" once my project migration is done).
Finally, a broader point, fair is in the business person's lexicon, or else they will be saying hello to Mr. Tax Inspector or Ms. Prosecutor on transgressions of business law. As Mr. Gates himself discovered, somewhat to his astonishment, the authorities do take these things seriously, at some eventual point.
Data warehouse/mart is precisely where I would not go with NT for both performance and reliability reasons. This is one sector where Microsoft's marketing and Linux are going to collide head-on. You can get stuff done in the Microsoft arena, but it's such a chore. Don't even get me started on SQL Server performance.
Lightweight data marts using even MySQL look feasible for many organizations looking into this, who will be put off by the big iron/big ticket entry level for traditional data warehouse development. DB2 on Linux makes me very happy, on the other hand. People talk about the flakiness of PC hardware, but that applies primarily to knockoff boxes. At the server end is a lot of hardware misappropriated to running flaky buggy Windows stuff, and destined for a stark choice in the near future: W2K or Linux. I already know where I am going.
1. The/. Effect is a capacity issue, not a demand one. 10,000 pageviews at zdnet is nothing in the scheme of things, but if they all happen at once that might be noticeable. Maybe.
2. Clickthroughs are the wobbly foundation of web site cash flow. The industry average is around $15/M and going down. Clickthroughs per pageview on many sites are below 1%. Do the math. 10,000 page views *might* be worth a few dollars.
3. You work in this business and you believe ZD would try to build traffic this way? Even if they could, never attribute to conspiracy what you can attribute to coincidence.
The big mistake in backup, especially with large systems, is to scale up the backup to meet the total size of the system being backed up. This is because people seem to believe that image or total backups are just keen.
What you end up doing is backing up 90% of the same data over and over again. All this does is cause hard drives to fail faster and tape heads to wear out sooner.
A better approach is to build a two-level backup scheme where "system" backups (of system stuff, non-changing libraries, data archives, etc.) are done only occasionally (weekly or monthly), and daily or more frequent backups are done of changin data only.
Right after this article came out in dead trees format I sent Jon a note (it still seems weird that he's a Compaq employee but he still has the dec address:)
In brief, it seems surprising that nobody in the trade press noticed until six months ago that Linux has more users now than Windows did in 1991 (yet Windows dominated the news pages of PC Week, InfoWorld and so on from 1990 onwards).
Two factors are of interest here. One is that Linux is not really a corporate desktop thing (yet), as Windows was at first (mostly because most home users couldn't afford the hardware in 1990-91; later with the fall in memory, disk and CPU prices, the release of W95 and the takeoff of the Web it zoomed madly in the home market). Linux has been quietly off in departmental corners serving print jobs and Web pages and email.
The second factor is that Linux has been thoroughly international from the get go. Thus its 10 or 20 or more million users have been spread far more widely than the concentrated corporate office visibility of Windows in the early years. In addition the localization that is key to expansion in non-US markets is very advanced under Linux. The market can grow UP, not OUT, from its current status. That is a crucial consideration for the future development trajectory of Linux.
My conclusion is that Linux is in a *stronger* position than many people realize to make further gains in departmental computing. There will be an inevitable backlash against Linux as potential users start to wrestle with its unfamiliar approach. But, mark my words, they will be assimilated. Why? Performance, price, features, reliability. Nuff said.
The new technology is very interesting indeed. New ideas come along in the speaker world maybe once a decade or less, and this is one.
But this is hardly ready for prime time.
I'm intrigued by the flat panel approach because it makes the idea of creating a "mini sound stage" more feasible.
At the moment I have two Crown CE1000s, two JBL 15s and a Cerwin Vega 18 sub. Compared to that, these flat panel things are toys -- can't handle real power and produce serious SPLs, don't have the frequency response.
In a few more years, this may develop into an interesting submarket though.
Slashdot Mirror
No, as you noted, clearly 356.25 was the design goal.
---------
I refuse to let 95 or 98 in my office. Not even a CD-ROM. NT 3.51 is a bit more stable than 4.0 in my experience, but doesn't handle big hard drives, so I moved my data crunching system up to 4.0 and it has basically been fine.
What really sucks is having to reboot whenever you install a new app with one of Micro~1's crappy shared DLLs. The latest offender was Foxpro 6, which I only need for some light file management, and which had me reboot three times (twice for the mandatory install of evilbad IE which I never use). Rebooting for network changes is bad enough, but for mere mortal applications??
This reminds me yet again why
Oh, it was fine, Brian (but you know they will always try to trap you that way :)
I liked the emphasis on the "whole product" theme from Geoffrey Moore's book. Too many people read Crossing the Chasm and miss that, yet it's really the most important thing among many interesting things he says.
"Open Source" as a whole is starting to cross the chasm, and you can see how that affects the view of those who judge it from the outside. This is why KDE/GNOME, Samba, 24x7 support, boxes with red hats, VMware and so on are attracting attention more than equally interesting developments like ipchains, Beowulf and the GIMP. The former have more direct relevance to the departmental desktop managers that the slick trade magazines talk to. And that is one of the early steps in crossing the chasm to the mass market.
Now, it may be that Wintel PCs and to a lesser degree Macs continue to dominate the market, but just as there was always a large slice of personal vehicle drivers who chose pickup trucks and vans, it may not be long before the SUV of operating systems comes along to occupy the same kind of middle ground (with less pollution and more driving stability, to be sure!).
The branding issue is of less importance, though it was quite prominent in the published part of the interview. Branding is a given, not an option, in my view, but it got the (admittedly limited) attention of the marketeering world in the last couple of years.
Strong branding doesn't save your butt if the product isn't good or isn't differentiated. Remember the Peanut (PCjr)? Hmm. How about New Coke? That was a good one. There was certainly no problem with Prodigy's brand identity, or dBase, or OS/2 (sorry).
Again, this is not to dismiss the importance of branding, but just to suggest it's a necessary but not sufficient condition for success. One of the interesting things is how the Open Source/free software world has evolved a sort of organic approach to branding. I mean, that Penguin. Or, that slash and that dot.
And don't forget that despite a brilliant branding strategy,
Um, how can I say "get a clue" nicely. We have plenty of choices here in Portland, Oregon. Bill Gates is not interested in wiping out Linux and couldn't do so even if he tried. Paul Allen, not Bill Gates, is making the big investments in cable systems these days.
I offer you these Free Clues in spite of the fact that
You know, I have been very good in restraining myself, but this notion that RMS is some kind of "communist" is the worst kind of infantile name-calling. It shows very clearly that those who toss out such insults so casually have no concept of what RMS is trying to do and even less about what "socialist" or "communist" means.
RMS is a classic "liberal" in the original 18th century Enlightenment sense of that word.
You can agree or disagree with his project (and I find myself on one side or the other depending on the situation), but the name-calling only obscures the fundamental importance of the issues he has been raising.
--------
After slogging through all this, I'm annoyed -- no, I'm irate -- that so much of the discussion has been sidetracked by a gale-force Slashdot Paranoia storm.
.exe of Perl 4.0.1.6, ported by Diomidis Spinellis and dated 2/23/1992 -- it would not be exaggerating to say that it changed my life almost more than any other single program).
Sure, Microsoft is blah blah blah [insert favorite pejoratives here]. As a matter of fact, they are not the devil incarnate and some of their stuff does work pretty well. You can thank them for the fact that the Win32::ODBC module is the backbone of many a fine Web site (or didn't you notice that ODBC, a Microsoft project, is a protocol that despite some faults Doesn't Suck).
I'm also appalled that so many of the commenters here didn't even bother to check basic facts about Perl and its history in the DOS/Win realm. I've been using it since I was running everything under DOS 5.0 in 1992 (still have the old
I've also used the various ActiveState Win32 Perl releases for years with excellent results. For example, I use it to munch 2-gigabyte raw dumps from databases and produce crosstabs and statistics two orders of magnitude faster than the databases can do so themselves. One of my main reasons in moving to Linux, actually, was to double that performance, but it's certainly not bad under NT.
I'm glad tchrist showed up to correct some of the more egregious misunderstandings about Perl. The evolution of Perl between 1994 and 1997 from basically a fairly limited domain of text processing to a platform for all kinds of useful things, through the addition of object oriented structures, the revision of the library approach into the full-blown module environment we now have, and so forth, ought to be a classic study in how computing tools can evolve in a positive sense rather than a bloatware sense.
If Microsoft wants to hook into that in a bigger way, that's a very good thing. Perl makes my life in the NT context bearable. However, I don't want to see it fork from the standard distribution in any significant way. Larry understates the importance of reunifying the code bases between the standard development tree and the ActiveState version with the One Perl initiative. And he also plays down the fact that it was not a guaranteed win. But nonetheless it did work out.
So now consider the real story: Microsoft has conceded where the locus of authority is for Perl. They are promising in public to play nice. Their tendency not to do so is well understood and their activities will be thoroughly scrutinized.
In other words, this is precisely the opposite of what happened with Java. And it is no surprise, given that Perl is open source and Java is not.
And that is the most important lesson to be drawn from this.
I don't mind Microsoft throwing in some things to optimize local processing, as long as they are themselves open source and don't fork my code. Because the bottom line is that if I can't run the basic stuff across platforms, their optimizations will disappear from my code first. That is market discipline Microsoft has never had to face, and they may have to grow up (maybe just a bit, but grow up nevertheless) as a result.
Perl is a remarkable accomplishment. Learning more about its history, philosophy and development is a rewarding experience. And, golly guess what, it's all out there readily available to be discovered.
-------
maybe he funds Perl, but
Hey, let's not give them any ideas!
-------
Enron's net operations are headquartered here in Portland, Oregon, where I live, and I buy my electricity from them (through subsidiary company Portland General Electric which they swallowed in 1997), which has more than a little to do with the location of their NOC here.
What we have learned about Enron is that they are an aggressive and less than trustworthy organization. They are dramatically expanding their business lines beyond the core of natural gas trading and natural gas transmission into electric markets, municipal water systems and now bandwidth. I would not be surprised to see them take a leap at telecom next, because their effort to take over the American electric industry stalled after they bought PGE.
Enron trumpets its reputation as an ethical multinational business. They belong to various councils and associations promoting good business practice. However, the record is actually quite different.
Last month, their new natural gas power plant at Dabhol, India (Maharashtra State, southwest of Bombay) opened up. This is not only the largest single power plant in the country, it has also been the center of a 7-year political struggle involving a close (some say corrupt) relationship with leading politicians and political parties, and repression against the thousands of local residents whose homes and livelihoods have been disrupted by the construction of the plant and its land and water needs.
India is not, as some people think, a moribund "third world" country. It has titanically huge problems but it is also the second largest nation in the world (and will possibly surpass China within a few decades), and the largest democracy on the planet. It has the ill-famed "streets of Bombay and Calcutta," and it also has more PhDs than any country on earth. (And lately, it has atomic bombs.)
What Enron did in India is hardly a matter of just pushing a few backward natives out of their huts. The Dabhol plant has been a major factor in national politics, the subject of parliamentary and Supreme Court inquiries, the focus of demonstrations where thousands were arrested on Enron's behest, and where Enron even paid for off-duty police to harass and assault opponents of the plant.
This is hardly grapevine stuff. It is documented in great detail in a full-scale report by Human Rights Watch issued earlier this year.
When speaking of Enron, remember that they have one of the most aggressive self-promotion efforts of any company involved in utility operations, whether it be gas, electric, water or bits.
------------
I use a venerable DOS shareware program called qe.exe (QuickEdit 2.06). It's dated 5/19/1988 and takes all of 48K. It always runs, never complains, is fast and pretty easy to use. There are a couple things I wish it did, but since I can run it on any M$ box from DOS 2.11 to NT 4.0, it's just fine.
There is a current version called The SemWare Editor which has a few more bells and whistles.
---------
Um, why does everything have to be denominated in dollars? Or do you just assume that enthusiasm has to be paid for any more on the Internet?
/. its well deserved check for promoting the Phantom Menace, then let me know.
In a world where Marimba has a market cap of maybe $90 million at the moment, I find it laughable that CT is constantly being accused of highlighting companies, web sites, products or whatever, in exchange for filthy lucre.
But hey, when Lucas gives
--------
For some reason (I might just be a dummy, duh) I didn't realize this old AltaVista interface was still available!
I think it first went up the first week of December 1995; I've been using it since a day or two after it went live.
google is good but it has been unavailable for considerable periods over the last week or so . . .
#-------
If there's one thing we learned in April, it's that benchmarks and Orange Book security levels are both virtually guaranteed to be misused in marketing. And I'm not just pointing the finger at Micros~1.
--------
The people who design planes are not the best ones to ask about flying them.
It's all well and good for Linus and Alan Cox to help tune things for a benchmark. But actually, they are not the best people to do so. The best people to do so are those with domain experience -- that is, people who implement and manage large systems using Linux.
And again, let me warn everyone: benchmarks are a moral hazard.
--------
A very clever idea from a very clever man. But I'm afraid my view of Lessig is that he's too clever by half, which is readily apparent after you start reading his proposed "solution" to the problems posed by the CDA.
He keeps trying to start up bandwagons that will name their tune for him, but they keep sputtering and stalling. Remember the "New Chicago School" notions that were supposed to supplant the "old school" law-and-economics approach of Posner and others at the University of Chicago? Well, Posner's crowd keeps flattening people everywhere with its mechanistic view of how the world works (which generally somehow, I really don't know how, always seem to end up supporting those who already have the greater economic power in any given situation -- but I digress).
And "New Chicago"? Wellll, New Chicago looks kind of old school after all, if you look at the table of contents of some of Lessig's recent confabs, such as this.
I'm sure Lessig is far smarter than I'll ever be. But he strikes me as far more a politician than an innovator or synthesizer in the field of law, and this "open law" thing looks like just another attempt to grab the limelight.
--------
What is missing in all this -- and I'm afraid in ESR's rather sloppy summary as well -- is any sense of whether benchmarking is reliable in the least for analyzing application processing. Things are bad enough with processor benchmarking -- remember the tuning code in Quake?
In databases, things are much worse. The TPC has been wrestling with these issues for a decade now and still doesn't really have a good handle on it. It is too easy to put your thumb on the benchmarking scale without anyone noticing, and make the results go the way you want. This is true even if the vendors themselves do the tuning.
Even if you can equalize the platforms, it still gets down to issues like the mix of instructions in the test suite. The TPC has wrestled with that one for years, trying to avoid tipping the balance toward any given vendor.
Schematically, of course, Web servers are database servers (aside from the issue that they may request data from an actual database).
Frankly, I'm even more appalled at Mindcraft (and by extension, Microsoft) for pushing this "study" the way they did. It was at least borderline unethical, given their admittedly lame-if-you-are-being-kind-about-it effort to equalize tuning between the two systems.
The biggest drawback to NT 4 (SPx) is that the security has gone from terrible to merely bad.
It still amazes me that Microsoft, the ultimate "grab the good ideas from the competition and bundle them into the OS" outfit, has not seen fit to install a decent port monitor, much less something like tcp wrappers. This borders on criminal negligence, in my view.
All the same, I use NT for some fairly big database projects (some in the multi-gigabyte data range) and it is solid as long as it's not running memory leakers like browsers or (yikes) Excel.
-------
I don't agree with the idea that W95 runs 16 bit apps better than NT. Quite the opposite, from my view, and I run a lot of 16-bit stuff on NT (Word 2.0, for example :) The problem I do have is with 32-bit apps that won't run on NT 3.51, so I have NT 4.0 on my big database server.
I used W95 when it was in beta for a week in May 1995, erased it and reinstalled NT 3.51. I have never regretted it. I avoid even helping my friends fix their W95/98 problems. What should take 10 minutes often becomes an open-ended 8 to 12 hour repair job. I have never had any real trouble with NT, it's just slower and a little less stable than I'd like, and the security is not nearly as good as it should be. My big Perl data jobs run twice as fast under Linux as NT on the same box (even without tuning).
--------
First of all, a small correction: XENIX was an early Microsoft product, but their first was Tiny BASIC.
A more serious matter: NT is not based on Unix, its roots are more in VMS. Read Helen Custer's excellent Inside Windows NT which explains the history in interesting detail. I actually like NT the operating system; it's the Windows architecture and especially the W95-look-and-feel subsystems that run on top of it that I find excruciating. I still use NT 3.51 as my primary desktop because the interface doesn't get in my way as much. (Fear not, Linux fans, I already have Debian running and it will be where I "live" once my project migration is done).
Finally, a broader point, fair is in the business person's lexicon, or else they will be saying hello to Mr. Tax Inspector or Ms. Prosecutor on transgressions of business law. As Mr. Gates himself discovered, somewhat to his astonishment, the authorities do take these things seriously, at some eventual point.
-------
It's not NT, really. Excel is the all-time champion memory leaker. It's so bad I'd put it in a category all by itself, even above browsers.
-------
Data warehouse/mart is precisely where I would not go with NT for both performance and reliability reasons. This is one sector where Microsoft's marketing and Linux are going to collide head-on. You can get stuff done in the Microsoft arena, but it's such a chore. Don't even get me started on SQL Server performance.
Lightweight data marts using even MySQL look feasible for many organizations looking into this, who will be put off by the big iron/big ticket entry level for traditional data warehouse development. DB2 on Linux makes me very happy, on the other hand. People talk about the flakiness of PC hardware, but that applies primarily to knockoff boxes. At the server end is a lot of hardware misappropriated to running flaky buggy Windows stuff, and destined for a stark choice in the near future: W2K or Linux. I already know where I am going.
-------
1. The /. Effect is a capacity issue, not a demand one. 10,000 pageviews at zdnet is nothing in the scheme of things, but if they all happen at once that might be noticeable. Maybe.
2. Clickthroughs are the wobbly foundation of web site cash flow. The industry average is around $15/M and going down. Clickthroughs per pageview on many sites are below 1%. Do the math. 10,000 page views *might* be worth a few dollars.
3. You work in this business and you believe ZD would try to build traffic this way? Even if they could, never attribute to conspiracy what you can attribute to coincidence.
-------
It's too bad that both KDE and GNOME are going down the CORBA primrose path.
Take it from me, that way lies madness. It's only a matter of time before people understand that.
-------
The big mistake in backup, especially with large systems, is to scale up the backup to meet the total size of the system being backed up. This is because people seem to believe that image or total backups are just keen.
What you end up doing is backing up 90% of the same data over and over again. All this does is cause hard drives to fail faster and tape heads to wear out sooner.
A better approach is to build a two-level backup scheme where "system" backups (of system stuff, non-changing libraries, data archives, etc.) are done only occasionally (weekly or monthly), and daily or more frequent backups are done of changin data only.
-------
Right after this article came out in dead trees format I sent Jon a note (it still seems weird that he's a Compaq employee but he still has the dec address :)
In brief, it seems surprising that nobody in the trade press noticed until six months ago that Linux has more users now than Windows did in 1991 (yet Windows dominated the news pages of PC Week, InfoWorld and so on from 1990 onwards).
Two factors are of interest here. One is that Linux is not really a corporate desktop thing (yet), as Windows was at first (mostly because most home users couldn't afford the hardware in 1990-91; later with the fall in memory, disk and CPU prices, the release of W95 and the takeoff of the Web it zoomed madly in the home market). Linux has been quietly off in departmental corners serving print jobs and Web pages and email.
The second factor is that Linux has been thoroughly international from the get go. Thus its 10 or 20 or more million users have been spread far more widely than the concentrated corporate office visibility of Windows in the early years. In addition the localization that is key to expansion in non-US markets is very advanced under Linux. The market can grow UP, not OUT, from its current status. That is a crucial consideration for the future development trajectory of Linux.
My conclusion is that Linux is in a *stronger* position than many people realize to make further gains in departmental computing. There will be an inevitable backlash against Linux as potential users start to wrestle with its unfamiliar approach. But, mark my words, they will be assimilated. Why? Performance, price, features, reliability. Nuff said.
------
The new technology is very interesting indeed. New ideas come along in the speaker world maybe once a decade or less, and this is one.
But this is hardly ready for prime time.
I'm intrigued by the flat panel approach because it makes the idea of creating a "mini sound stage" more feasible.
At the moment I have two Crown CE1000s, two JBL 15s and a Cerwin Vega 18 sub. Compared to that, these flat panel things are toys -- can't handle real power and produce serious SPLs, don't have the frequency response.
In a few more years, this may develop into an interesting submarket though.
signed,
/The Beat That Goes BOOM/
-------