I used to be an assistant prof. of computer science, back when Pascal was the introductory language. We had a program named Big Brother that would compute various measures of student assignments, e.g. the number of occurrences of various constructs, and come up with a hashcode. If two submissions had the same hashcode, that was a warning flag that someone copied from someone else. Big Brother was written to be immune to trivial transformations, e.g. renaming a function.
Big Brother wasn't the final judge of course, but it was very good at picking up cheaters. At least I think it was. I'm sure it missed anyone smart enough to think about how Big Brother must work. So I guess it picked up lazy and dumb cheaters, but not lazy and smart cheaters. Which isn't a bad compromise if you think about it.
This is quite different from the sort of tool described in the cited article. I think it is an arguably good use of technology on the teachers' "side".
I would love to hear the reactions of the Intel engineers to this lame new name by the engineers who created Merced. After three years, I still can't get used to any of the names that marketing has applied to my product.
And what exactly is the performance problem with C++?
My C++ is admittedly rusty, (I moved on to Java three years ago), but I don't recall any performance issues. I do remember being completely screwed by over-enthusiastic use of operator overloading, virtual base classes, fancy conversions, and especially templates, but not performance issues.
You say that it's the C++ glitz that makes the language slow. Exactly what features are you referring to, and what sort of performance penalty is incurred?
If I had to write C or C++ again, I would definitely want to use C++ (performance permitting), but I would be much more selective about the features I used.
And another thing: Maybe Win 95/98 (but not NT) auto-detects hardware better. But if for any reason you have to control IRQs yourself, god help you. I spent a weekend trying to tweak IRQ settings in the BIOS so that PnP's AI module would be fooled into shuffling IRQs in just the right way to get the assignment of IRQs I was trying to achieve. This is complicated, of course, by IRQ sharing, which means that even figuring out what IRQ configuration you want is a big deal all by itself.
Microsoft presents this illusion of effortless software. But if you stray from The One True Path, by even a micron, you're off the freshly paved, 8-lane highway, and on a dark, scary unpaved road, with lots of invisible monsters who would like nothing better than to eat you and your hard drive for dinner.
I'm reconfiguring some of my PCs to be dual-boot machines (Win98/Linux). On one box, both OSs installed easily, with little intervention required.
The other box is a 1-year old Dell that SHIPPED with Windows 98. I wiped the disk and tried reinstalling Windows 98, from the Dell-supplied CD. With 3 minutes left in the install (they have a little timer), I got an error message about a broken or missing vxd file. Reformatted the disk again -- same thing. I have a real problem. Dell support was useless. ("Try starting the installation from the floppy instead of booting from the Windows 98 CD". I tried, just to be safe, and of course that failed.) Searched the web and found a 19-step procedure for fixing this file. The procedure takes about an hour to go through. Made a slight error after that, and had to start from scratch (more reformatting). Got it right the next time. Installed a NIC which caused the machine to lock up solid when booting. Removing the NIC (in safe mode, from the control panel) and physically removing the card didn't fix the problem. Another reformat and reinstall (along with the 19-step 1-hour procedure to avoid the vxd problem). Finally get through it all, and the machine is flaky -- freezes within five minutes.
It's hard to imagine a simpler installation scenario -- standard Dell machine with no extra hardware; blank disk; standard Windows 98 installation. After banging my head against this wall for three days I still have a useless PC.
Meanwhile, I'm quite new to Linux, and got through the RH 6.0 install in about 15 minutes on another machine, on my first attempt.
Now I have to say that the Windows install had prettier graphics flashing while the installation "proceeded", but I don't see that as contributing to simplicity.
IANESR* but: From my worms-eye view, as an engineer in my second startup, being a member of a board of directors appears to be very far from a full-time job. It's amazing how uninvolved BOD members are, even when the company they direct is swirling down the drain. I think the reason is they are involved, not committed. (A chicken is involved in making the egg in your omelette; she's committed to your chicken salad.) * IANESR: I Am Not Eric S. Raymond
Endings: Stephenson and Gibson
on
The Diamond Age
·
· Score: 1
I agree -- Diamond Age and Snow Crash both had pretty disappointing endings. I think Stephenson's books and William Gibson's are similar in many ways. The endings are interesting to contrast. For both authors, at the end of their books, there is the prospect of a quantum leap in evolution, (not of the human kind). In Neuromancer and one other by Gibson (Mona Lisa?) the networks of machines get so big that some sort of gestalt effect gives rise to a conscious being, and we find out that beings of this sort are in contact with one another. This isn't presented as good or bad; it just happens.
In Snow Crash and Diamond Age, the alternative (the Drummers in Diamond Age; the antenna-heads in Snow Crash), are portrayed negatively and defeated at the end.
The problem I have with both authors is the length of the elaborate "action sequences" needed to push these plots to their fantastic conclusions. I find the descriptions of technology and sociology, and the atmospherics much more interesting. Diamond Age was really spectacular in this way.
Saving 1 msec. is not what algorithm analysis is about. It's about taking 1 msec. to do something vs. 10 seconds or more. In 1 msec. you can go through the loop of a binary search, oh I don't know, 100 times, which means that you can search a list of 2^100 items. If you did a linear search, it would take you, on the average, 1/2 * 2^100 * (let's say 1 nanosecond) which is 2^69 seconds, which is a LONG time. This is the difference between O(log n) and O(n) for large values of n.
If you know n is small, (e.g. the number of lines in/etc/hosts), it won't matter what you do. If n is big, (e.g. a 100 meg log file), your choice of algorithm matters. If you're counting nanoseconds in a tight loop, then you're presumably already using a fast algorithm, but the exact variant is important. For example, heapsort is guaranteed O(n log n), but quicksort is often preferred because, although O(n^2) worst case, it's O(n log n) on the average, and the constant is lower than for heapsort. A lot of Knuth is about deriving these constants. Knuth goes even farther than this, e.g. showing how to derive the constants for different ways of choosing the pivot in a quicksort.
It's exhausting, non-productive and insane to apply this sort of analysis to everything. But once you've focused on the performance-critical parts of your program, Knuth's detailed analysis is essential.
The Hotspot VM is pretty amazing, especially considering how new it is. The speedups are really impressive. On our benchmarks, (middleware, lots of multithreading and network access), we're seeing 3-5x improvements on various benchmarks. The garbage collector is amazing. We used to find that pooling beat the (admittedly crude) JDK 1.1.x garbage collector by 2x; now the situation is exactly the opposite. Code volume stays reasonable -- hotspot doesn't compile everything in sight; it really does seem to focus on the hotspots. The stability is not perfect, but is very good for such an early release.
Finally is an interesting idea that has been proposed a few times that I just wanted to throw out for discussion. As it stands, you only see the moderator controls when you have moderator points to use. What if the moderator controls were always visible, but when you submitted the form, they were only counted if you had moderator points. Oh, and you wouldn't know if you had points.
Here's why I don't like this idea. First, there is the burnout issue. When I'm a moderator, I will make an effort to read and moderate lower-rated posts. If I don't know when I'm "on", then I'll be less likely to read those lower-rated posts. I think what this will do is to cause many good comments to never get moderated up. Second, it's easy enough to see if you're a real moderator -- just vote and see if the rating changes. (Not foolproof, but it has some accuracy.)
I think it was a joke. Fermat, wrote in a notebook that he found a proof of his theorem, but that it was too small to fit in the margin. Sounds like the RFC author is referring to that.
What are these tests used for? If test results are used to determine teacher raises, or funding for schools, then you can bet curricula will focus on subjects that are tested, at the expense of subjects that are not tested.
The names were so dumb, I reread the posting carefully with my joke detectors on high alert. I finally concluded it was serious. Reminds me of the "tongue troops" in France and Quebec, protecting the purity of their mother tongue by discouraging certain words (e.g. "weekend") that invaded from another language.
The phrase "predefined headers" is what set off the alarm bells for me. It suggests an exercise in filling out the required tonnage of paper. I suggest filling in the predefined headers with random paragraphs selected from Slashdot discussions, and see if anyone in management notices.
I don't know about open-source development, but what I've found -- even on 1-man projects -- is that the process of writing out a functional spec of design is valuable because it forces me (or my team) to think about the design issues early. That's the real value.
> In your experience, how often are > design documents revisited after the > project?
Rarely. The main value is in the thinking you have to do in order to produce the document.
> Have design documents helped in > technology transfers (that > is, have they been more helpful than the > source code alone)?
You mean to other engineers? Definitely. Assuming the document is written well, and with care, and is read carefully.
> Are engineers good at writing design documents? The absolute best engineers I've worked with *all* produce excellent design documents.
> Have you been able to read design > documents written by other > engineers?
Yes. I think the best approach is to do brainstorming and then record the important points. Everyone gets to do some writing, and everyone reads what everyone else writes. The brainstorming and documentation can be achieved simultaneously in email, (if the engineers are comfortable with this); or you can brainstorm in front of a whiteboard and write stuff up later.
> Have old design documents been kept up to > date > with the changes in the implementation?
Usually not. But on the other hand, if the original design is good, the distance between it and the implementation will be small.
> Has the quality of your products been better > because of design documentation?
Yes. If I skip design and start hacking, (always tempting), I find that I suffer for discovering important design issues too late.
All that said, the documentation effort has to come from the engineers. If imposed from above it won't work.
A friend of mine just started working for the DOJ in the area of crypto laws. Here's what I sent him (along with the article):
I'm really concerned about these terrorists. Do you realize that, at THIS VERY MOMENT, someone, somewhere is selling US-made guns and bullets to a terrorist, and that we can't do anything about it?
We should ban the export of guns and bullets. Really bad things might happen if these US-made guns and bullets fell into the hands of the wrong people. Now before you start thinking that this is a radical idea, I just want to point out that I'm not calling for an outright ban on the manufacture of guns and bullets in this country, just their export. After all, no one else in the world knows how to make US-made guns and bullets, no American citizen would *ever* sell US-made guns and bullets to terrorists surreptitiously, and no graduate students from fundamentalist Islamic nations who return home after school ever learn the secrets of making US-made guns and bullets. So an export ban on US-made guns and bullets is really a foolproof measure.
Now we wouldn't need to be so extreme about this if the firearms industry would cooperate. All they need to do is to modify their products ever so slightly, to incorporate radio transmitters which let the government trace any guns or bullets that they suspect are being used for evil purposes.
Could you please talk to Janet about this?
One last thing: GET THE HELL OUT OF THERE WHILE YOU STILL HAVE A SINGLE FUNCTIONING NEURON IN YOUR BRAIN!
The article points out that one reason for low coverage is the lag (search engines are months out of date), combined with the incredibly rapid increase in the number of pages, (100% growth in about a year). So even if *everything* six months old were indexed, coverage will still be only 50%. Anyone who does any searching quickly realizes this, so the study isn't breaking ground here, although maybe it quantifies the problem.
Beyond this, I don't see how the study's result could be meaningful.
1) How did they come up with their estimate of 800 million web pages? If that number is bogus, so is the %age. They can measure the pages they found, but how do they measure the pages they couldn't find? Different techniques of estimation might provide great variance in the number of web pages.
2) Counting pages (and computing coverage) is especially problematic given the increasing amount of content generated dynamically.
As long as we're pulling out historical analogies, I think Metcalfe sounds like Abby Hoffman: "Never trust anyone over 30". His comment about Linux's impending senility is especially ironic given the vast influence of Unix on the Windows APIs (cloning much of the standard C libraries; the very use of C itself). Also, I find it hilarious that the bottom of Metcalfe's column points out that he invented Ethernet in 1973. I guess it will soon be time to dump Ethernet too. It is getting on in years, and you know we are running out of IP addresses. And what's with these numbered ports? Ports should be objects, located by a naming service implemented in microcode, the BIOS, or at worst, the OS kernel.
I'm a mediaone subscriber. Overall, it's an excellent service. Servers are (or used to be) prohibited in the service contract, but there was no enforcement. A number of subscribers run ftp, http and mail servers. Proxy servers are commonly used to support LANs. Mediaone grudgingly accepts this, but have promised to end your service if you don't run a secure server -- there were a number of breaches that ended up making mediaone look bad.
I've been using the following system for the last three jobs: 1.Give me cool stuff to work on. 2.Recognize and appreciate the work I do for you. 3.Stay the hell out of my way.
This is my list too. I voluntarily left a startup (of which I was a cofounder) shortly before the IPO because the new regime wouldn't come up with even one of these.
What I found astounding was the degree to which other geeks would put up with boring work, and lack of recognition, (not to mention outright hostility and a pay cut).
I think too many geeks & nerds take whatever their stupid/evil management dish out. I find this incomprehensible given the opporunitites available today, but some people just hate change.
Loyalty to a company is dumb. Be loyal to your ideas and your values, and look for a company whose interests coincide with yours.
I got the MS Phone for a birthday present. Who knew a phone could be so complicated:
1. I had to fiddle with BIOS and IRQ assignments to get everything into an IRQ sharing arrangement that actually worked. PnP decided to have the phone and video share an IRQ, which caused lockups whenever the screen saver exited.
2. I have a UPS plugged into my one serial port. MS advises the use of a device that plugs into a USB port and creates more serial ports. However, the MS phone documentation advises against using this for the MS phone itself. I just unplugged the UPS from the serial port.
3. When I finally got everything set up and working, (the voice recognition was excellent), the phone had a stroke. I guess it's really sensitive to static because one day it started screeching like a banshee everytime I picked it up, (yes I tried scanning for a new frequency to the base station).
All in all, a totally frustrating, completely unsatisfying, typically Microsoft experience.
Gates said there was clearly a market for free software but this was mainly confined to relatively simple applications such as word processing and spreadsheets.
I don't think he's referring to open source apps; I think he means that Linux is OK for these simple apps, but that for real heavy-duty, mission-critical applications you really want to go with W2K. Hotmail, for example.
Slashdot Mirror
No, and I'd rather not name names. But Big Brother is a pretty obvious name once you've written the program.
Big Brother wasn't the final judge of course, but it was very good at picking up cheaters. At least I think it was. I'm sure it missed anyone smart enough to think about how Big Brother must work. So I guess it picked up lazy and dumb cheaters, but not lazy and smart cheaters. Which isn't a bad compromise if you think about it.
This is quite different from the sort of tool described in the cited article. I think it is an arguably good use of technology on the teachers' "side".
I would love to hear the reactions of the Intel engineers to this lame new name by the engineers who created Merced. After three years, I still can't get used to any of the names that marketing has applied to my product.
My C++ is admittedly rusty, (I moved on to Java three years ago), but I don't recall any performance issues. I do remember being completely screwed by over-enthusiastic use of operator overloading, virtual base classes, fancy conversions, and especially templates, but not performance issues.
You say that it's the C++ glitz that makes the language slow. Exactly what features are you referring to, and what sort of performance penalty is incurred?
If I had to write C or C++ again, I would definitely want to use C++ (performance permitting), but I would be much more selective about the features I used.
Microsoft presents this illusion of effortless software. But if you stray from The One True Path, by even a micron, you're off the freshly paved, 8-lane highway, and on a dark, scary unpaved road, with lots of invisible monsters who would like nothing better than to eat you and your hard drive for dinner.
The other box is a 1-year old Dell that SHIPPED with Windows 98. I wiped the disk and tried reinstalling Windows 98, from the Dell-supplied CD. With 3 minutes left in the install (they have a little timer), I got an error message about a broken or missing vxd file. Reformatted the disk again -- same thing. I have a real problem. Dell support was useless. ("Try starting the installation from the floppy instead of booting from the Windows 98 CD". I tried, just to be safe, and of course that failed.) Searched the web and found a 19-step procedure for fixing this file. The procedure takes about an hour to go through. Made a slight error after that, and had to start from scratch (more reformatting). Got it right the next time. Installed a NIC which caused the machine to lock up solid when booting. Removing the NIC (in safe mode, from the control panel) and physically removing the card didn't fix the problem. Another reformat and reinstall (along with the 19-step 1-hour procedure to avoid the vxd problem). Finally get through it all, and the machine is flaky -- freezes within five minutes.
It's hard to imagine a simpler installation scenario -- standard Dell machine with no extra hardware; blank disk; standard Windows 98 installation. After banging my head against this wall for three days I still have a useless PC.
Meanwhile, I'm quite new to Linux, and got through the RH 6.0 install in about 15 minutes on another machine, on my first attempt.
Now I have to say that the Windows install had prettier graphics flashing while the installation "proceeded", but I don't see that as contributing to simplicity.
IANESR* but: From my worms-eye view, as an engineer in my second startup, being a member of a board of directors appears to be very far from a full-time job. It's amazing how uninvolved BOD members are, even when the company they direct is swirling down the drain. I think the reason is they are involved, not committed. (A chicken is involved in making the egg in your omelette; she's committed to your chicken salad.) * IANESR: I Am Not Eric S. Raymond
In Snow Crash and Diamond Age, the alternative (the Drummers in Diamond Age; the antenna-heads in Snow Crash), are portrayed negatively and defeated at the end.
The problem I have with both authors is the length of the elaborate "action sequences" needed to push these plots to their fantastic conclusions. I find the descriptions of technology and sociology, and the atmospherics much more interesting. Diamond Age was really spectacular in this way.
If you know n is small, (e.g. the number of lines in /etc/hosts), it won't matter what you do. If n is big, (e.g. a 100 meg log file), your choice of algorithm matters. If you're counting nanoseconds in a tight loop, then you're presumably already using a fast algorithm, but the exact variant is important. For example, heapsort is guaranteed O(n log n), but quicksort is often preferred because, although O(n^2) worst case, it's O(n log n) on the average, and the constant is lower than for heapsort. A lot of Knuth is about deriving these constants. Knuth goes even farther than this, e.g. showing how to derive the constants for different ways of choosing the pivot in a quicksort.
It's exhausting, non-productive and insane to apply this sort of analysis to everything. But once you've focused on the performance-critical parts of your program, Knuth's detailed analysis is essential.
The Hotspot VM is pretty amazing, especially considering how new it is. The speedups are really impressive. On our benchmarks, (middleware, lots of multithreading and network access), we're seeing 3-5x improvements on various benchmarks. The garbage collector is amazing. We used to find that pooling beat the (admittedly crude) JDK 1.1.x garbage collector by 2x; now the situation is exactly the opposite. Code volume stays reasonable -- hotspot doesn't compile everything in sight; it really does seem to focus on the hotspots. The stability is not perfect, but is very good for such an early release.
Here's why I don't like this idea. First, there is the burnout issue. When I'm a moderator, I will make an effort to read and moderate lower-rated posts. If I don't know when I'm "on", then I'll be less likely to read those lower-rated posts. I think what this will do is to cause many good comments to never get moderated up. Second, it's easy enough to see if you're a real moderator -- just vote and see if the rating changes. (Not foolproof, but it has some accuracy.)
I think it was a joke. Fermat, wrote in a notebook that he found a proof of his theorem, but that it was too small to fit in the margin. Sounds like the RFC author is referring to that.
What are these tests used for? If test results are used to determine teacher raises, or funding for schools, then you can bet curricula will focus on subjects that are tested, at the expense of subjects that are not tested.
The names were so dumb, I reread the posting carefully with my joke detectors on high alert. I finally concluded it was serious. Reminds me of the "tongue troops" in France and Quebec, protecting the purity of their mother tongue by discouraging certain words (e.g. "weekend") that invaded from another language.
The phrase "predefined headers" is what set off the alarm bells for me. It suggests an exercise in filling out the required tonnage of paper. I suggest filling in the predefined headers with random paragraphs selected from Slashdot discussions, and see if anyone in management notices.
I don't know about open-source development, but what I've found -- even on 1-man projects -- is that the process of writing out a functional spec of design is valuable because it forces me (or my team) to think about the design issues early. That's the real value.
> In your experience, how often are > design documents revisited after the
> project?
Rarely. The main value is in the thinking you have to do in order to produce the document.
> Have design documents helped in > technology transfers (that
> is, have they been more helpful than the > source code alone)?
You mean to other engineers? Definitely. Assuming the document is written well, and with care, and is read carefully.
> Are engineers good at writing design documents?
The absolute best engineers I've worked with *all* produce excellent design documents.
> Have you been able to read design > documents written by other
> engineers?
Yes. I think the best approach is to do brainstorming and then record the important points. Everyone gets to do some writing, and everyone reads what everyone else writes. The brainstorming and documentation can be achieved simultaneously in email, (if the engineers are comfortable with this); or you can brainstorm in front of a whiteboard and write stuff up later.
> Have old design documents been kept up to > date > with the changes in the implementation?
Usually not. But on the other hand, if the original design is good, the distance between it and the implementation will be small.
> Has the quality of your products been better > because of design documentation?
Yes. If I skip design and start hacking, (always tempting), I find that I suffer for discovering important design issues too late.
All that said, the documentation effort has to come from the engineers. If imposed from above it won't work.
A friend of mine just started working for the DOJ in the area of crypto laws. Here's what I sent him (along with the article):
I'm really concerned about these terrorists. Do you realize that, at THIS VERY MOMENT, someone, somewhere is selling US-made guns and bullets to a terrorist, and that we can't do anything about it?
We should ban the export of guns and bullets. Really bad things might happen if these US-made guns and bullets fell into the hands of the wrong people. Now before you start thinking that this is a radical idea, I just want to point out that I'm not calling for an outright ban on the manufacture of guns and bullets in this country, just their export. After all, no one else in the world knows how to make US-made guns and bullets, no American citizen would *ever* sell US-made guns
and bullets to terrorists surreptitiously, and no graduate students from fundamentalist Islamic nations who return home after school ever learn the secrets of making US-made guns and bullets. So an export ban on US-made guns and bullets is really a foolproof measure.
Now we wouldn't need to be so extreme about this if the firearms industry would cooperate. All they need to do is to modify their products ever so slightly, to incorporate radio transmitters which let the government trace any guns or bullets that they suspect are being used for evil purposes.
Could you please talk to Janet about this?
One last thing: GET THE HELL OUT OF THERE WHILE YOU STILL HAVE A SINGLE FUNCTIONING NEURON IN YOUR BRAIN!
The article points out that one reason for low coverage is the lag (search engines are months out of date), combined with the incredibly rapid increase in the number of pages, (100% growth in about a year). So even if *everything* six months old were indexed, coverage will still be only 50%.
Anyone who does any searching quickly realizes this, so the study isn't breaking ground here, although maybe it quantifies the problem.
Beyond this, I don't see how the study's result could be meaningful.
1) How did they come up with their estimate of 800 million web pages? If that number is bogus, so is the %age. They can measure the pages they found, but how do they measure the pages they couldn't find? Different techniques of estimation might provide great variance in the number of web pages.
2) Counting pages (and computing coverage) is especially problematic given the increasing amount of content generated dynamically.
As long as we're pulling out historical analogies, I think Metcalfe sounds like Abby Hoffman: "Never trust anyone over 30". His comment about Linux's impending senility is especially ironic given the vast influence of Unix on the Windows APIs (cloning much of the standard C libraries; the very use of C itself). Also, I find it hilarious that the bottom of Metcalfe's column points out that he invented Ethernet in 1973. I guess it will soon be time to dump Ethernet too. It is getting on in years, and you know we are running out of IP addresses. And what's with these numbered ports? Ports should be objects, located by a naming service implemented in microcode, the BIOS, or at worst, the OS kernel.
Did you catch the slogan? "The Evolution of the Computer Ends With Z". Dopey.
No, but this month's Wired has an excerpt.
I'm a mediaone subscriber. Overall, it's an excellent service. Servers are (or used to be) prohibited in the service contract, but there was no enforcement. A number of subscribers run ftp, http and mail servers. Proxy servers are commonly used to support LANs. Mediaone grudgingly accepts this, but have promised to end your service if you don't run a secure server -- there were a number of breaches that ended up making mediaone look bad.
1.Give me cool stuff to work on.
2.Recognize and appreciate the work I do
for you.
3.Stay the hell out of my way.
This is my list too. I voluntarily left a startup (of which I was a cofounder) shortly before the IPO because the new regime wouldn't come up with even one of these.
What I found astounding was the degree to which other geeks would put up with boring work, and lack of recognition, (not to mention outright hostility and a pay cut).
I think too many geeks & nerds take whatever their stupid/evil management dish out. I find this incomprehensible given the opporunitites available today, but some people just hate change.
Loyalty to a company is dumb. Be loyal to your ideas and your values, and look for a company whose interests coincide with yours.
1. I had to fiddle with BIOS and IRQ assignments to get everything into an IRQ sharing arrangement that actually worked. PnP decided to have the phone and video share an IRQ, which caused lockups whenever the screen saver exited.
2. I have a UPS plugged into my one serial port. MS advises the use of a device that plugs into a USB port and creates more serial ports. However, the MS phone documentation advises against using this for the MS phone itself. I just unplugged the UPS from the serial port.
3. When I finally got everything set up and working, (the voice recognition was excellent), the phone had a stroke. I guess it's really sensitive to static because one day it started screeching like a banshee everytime I picked it up, (yes I tried scanning for a new frequency to the base station).
All in all, a totally frustrating, completely unsatisfying, typically Microsoft experience.
processing and spreadsheets.
I don't think he's referring to open source apps; I think he means that Linux is OK for these simple apps, but that for real heavy-duty, mission-critical applications you really want to go with W2K. Hotmail, for example.
I agree completely. Is Neal Stephenson the guy who had that great article in Wired on running cable all over the world? I really like his writing.
More Stephenson please, and less on how many hackers can dance on the head of a *PL.