More importantly: I also agree that NYTimes has no "right" to require the client-side rendering of their website to match NYT's "expectations."
If the NYT wants that, they should deliver their homepage in some kind of encrypted protected fixed PDF-like format that doesn't allow modification.
I would go further and say that if the NYT wants that, they should send out an armed manservant with a briefcase (containing the HTML) handcuffed to his arm every time someone requests a page. Because that is the only way they will be able to control the manner in which the public views the NYT's work. Once those little bits travel out over the network, the game is over. Joe End-User can filter or reformat or do whatever he pleases. As long as it's short of redistribution, there is no limit on what you can do to process your data. Using some cockamamie "secure" format is very much akin to copy protection in its other forms: an annoyance to all, but not an obstacle to the determined.
It's worse than you think
on
Built For Use
·
· Score: 1
The bandwidth thing is a lot worse than everyone thinks. We all know that most of the lusers out there are on a 56k or worse. We should all know that they are not generally getting 56kb/s of bandwidth, that the common numbers are in the 28-44 kb/s range. But we should also keep in mind that people multitask. Poor idiots on a dialup are not connected 24/7, so they are more likely to have that Kazaa/service pack/game demo/background page load running. To work really well for a dialup user, a site should run nicely on half of a 30 kb/s link. That means responding well, say within 10 secs. on 2KB/s, which means loading 20KB/page or less. Much less would be better of course.
Bullshit. That's the way the Web works. Everyone understands the back button. In fact the very first things a new Web user learns are hyperlinks and the back button.
The problem is people who complicate their sites with half baked solutions to this non-problem, such as spawning new windows or the about.com style frameset mixing content from two sites.
Exactly. It's not secure telnet, it's the secure r* suite: rsh (both interactive and single command), rcp, and throw in port fowarding for good measure.
That's an interesting notion. But if someone was so fired up to ship unlicensed games, why were all the Dreamcast game releases licensed?
Now I do seem to recall that the commercial Pelican MP3 player was on a CD-R, but still, where were the unlicensed games? It was well known since about April 2000 how to boot on an unmodified Dreamcast.
And for that matter, you can boot on an unmodified PS1 at this point, I believe. Or someone could have included a $2 mod cartridge with their unlicensed title. Hmm.
Just to pile on here, Notes... UCE is never ok! None of your reasons matter a bit to me. I want _zero_ UCE. The only advertising email I will abide by is that which I asked for, which is generally none.
For this reason I would never have considered your company for anything after receiving your UCE.
I'll say one headless machine is drawing in the neighborhood of 200 watts. That's.2 KW. Per day, that's 4.8 KW-hours. One KW-hour costs neighborhood of a nickel, so one machine costs about 25 cents to run each day. 10,000 machines cost $2500 in electricity a day altogether.
You made a number of mistakes. The main one is: there is no notion of "watts per second" unless you are talking about a rate of change of power. 4.2M watts is a rate; it is the same whether measured over an instant, a second or an hour. If you use power at that rate for an hour you use 4.2M watt-hours. Not 15.1 billion watt-hours. So that was a factor of 3600 on the high side. Then you were off by a factor of about 14 on the low side in pricing kilowatt-hours. And I would say you were about a factor of 2 on the high side in wattage of a server. Altogether you were high by a factor of just about 500x which is the difference in our results.
Gawd, analogies are fun... Let's talk about bridge building.
This site at has some bridge history that I read recently. The Firth of Forth page in particular talks about a bridge that was built according to the maxim, "Everyone knows how a bridge should look." The result was an overbuilt, overcostly and ugly structure. A layman's intuition about what is necessary may be quite wrong.
The infamous Tacoma Narrows suspension bridge, of course, was built in modern times (1940) and according to conservative and well known principles, which didn't stop it from self-destructing. This was after more than fifty years of experience in designing large suspension spans. The judgement of an expert engineer with ample experience and time can also be very wrong!
Let's also consider the Charles River bridge, which looks like nothing you ever saw before. Nonetheless it is now being completed at a cost of $86M. To be sure, the design must have been reviewed, but if you saw a child's drawing of the bridge, you'd think it was a fantasy that could not be built. Again, so much for "Everyone knows how a bridge should look".
I agree with your basic point, which, as I understand it, is that the design must be visible and inspected to have a hope of correctness. But as bridge building shows us, that's not always enough!
Thanks for the link. That was one of the only on-topic comments in this whole story.
Were you using one or two monitors? Also I was surprised to see that you didn't go ahead and try some gigabit ethernet action since you were saturating your 100Mb link.
Do you see yourselves staying exclusively with the personal workstation model? What about driving these beasts using X over the network? You could buy a smaller farm of monster workstations and let your artists share them. Even in your high performance environment, "think time" has still got to have those systems idle more than half the time, not counting batch jobs of course. Not to mention, the slower simulation and filtering tasks could in principle be written to distribute themselves over a number of CPUs.
SGI's Visual Area Networking is their effort to use this kind of thinking.
For a while, in the 70's and 80's, companies poured money into their software engineering process, but recently they realized that publishing reports on how they cut debugging time by 66% might tell their competitors how to do it too.
Gimme a break. You think all the people who know how to do good software engineering are keeping quiet so the recipe for the secret sauce doesn't get out? You need a better conspiracy theory; this one ain't cutting it. Do they pull the strings of trade journals, too, so no one writes about the secrets of software success?
My own personal theory is that the typical software development organization is not in a competitive market with respect to software quality. They build bad software and get away with it, because for a variety of reasons their customers haven't called them on it. Many are serving internal customers; others serve customers who just don't care to discriminate between software that works well and software that doesn't. I am thinking of PC packaged software here.
Organizations with long memories or discriminating customers still have their relatively more developed software engineering practices.
Gaming is the last true hurdle for Linux and it's being jumped quickly.
Forget Linux: Gaming is the last true hurdle for PC compatibles, no matter what the O/S. And it's not being jumped quickly. The PC game market is a terrible technical environment. It's a nightmare of multiple hardware and software configurations, being called upon to perform fast and in real time with zero user configuration. Brutal price competition means good tech support would be an impossibly large cost. Short attention spans on the part of press and public mean a race to the bottom as far as bug count.
Having said all this, the game production companies are obviously going to continue to focus most of their effort on the 95% market share O/S. Linux and Mac gaming will continue to be a red headed stepchild, months or years behind and with a fraction of the titles. MS will continue to be very aggressive in game API software development, and API emulators will continue to lag these gaming APIs more than most any other APIs.
by Martin Spamer on 05:27 AM -- Friday June 21 2002 (Score:2) (#3742297) (User #244245 Info) http://www.kitv.co.uk/ [ Neutral ]
What and When are you planning to to do about the frauds being perpetuated on Google Answers (http://answers.google.com) with questioners paying their own alias instead of the expert poster who deserves to be paid. They are destroying the cedibility of what has the potential to be an absolutely brilliant service.
I think you need to be a much more strigent about suspending and banning these abusers if you wish this service to take off.
Yeah, yeah, MS are evil. But there is far more packaged software development effort for Windows than any other platform. That's been the case for ten years. This effort has produced the most innovative and diverse set of independently developed packaged software products, not to mention MS's own estimable efforts.
It's a network effect, people! Everyone runs Windows, therefore everyone develops first/only for Windows, therefore everyone wants to run Windows. I would be pissed, too, if I had a Mac that had no ability to efficiently run all that software.
I realize I'm exaggerating a bit here. If you're a geek, then Linux is the place where you find the software you most desire. But for the mainstream that's not at all the case, not even close. It's normal and healthy to want to run all the software at CompUSA!
In the grand Slashdot grammar flame tradition: you meant "for C driver code, no less". Nonetheless, or nevertheless, means "however". "No less" is the strengthener you were looking for.
I say fire the new hire. If he doesn't love a text approach, he probably won't grow to love it. And if he never loves it he'll never get his productivity up to a very high level. There are plenty of workplaces where he can be comfortable in his IDE orientation.
Mice are the most pain-inducing device for me. The 3M mouse is the best mouse I've used. I recommend giving it a try if mice make you hurt. Works better for me than my old Logitech Trackman Marble FX.
If you distribute an application that is built from GPL'd source code, you must make that source code available to anyone who wants it.
Actually you have two choices:
Include the source with all distributions of the binary
Make the source available to all at a nominal fee
Either way the license must be the GNU GPL. But you can, for example, sell the product to someone under the GPL, give them the source, but not give the source to all and sundry. This is essentially a technicality since the recipient of the source is free to redistribute. But the GPL doesn't mandate making the software available to everyone, nor does it mandate FSF style nominal fee distribution.
It will not be sufficient... There really is a Brenda in Marketing who edits images, maybe also video and 3D. She uses the hardware acceleration in her video card to make her session run at the rate to which she has become accustomed. She doesn't use a simple framebuffer on her current PC. What makes you think that a framebuffer with a relatively low-bandwidth high-latency connection to the CPU will suffice for her?
I really like the terminal model but it is not appropriate for every person in the enterprise. You cannot give someone a much slower computer than what they are using and have them accept it. The Sun Ray system was not built to compete with a high end graphics performance PC/workstation. A Sun Ray is not going to do it for the graphic professionals, not now, not ever. They will leave the company en masse.
SGI does sell some network terminal graphics systems that run fast, but they ain't Sun Ray.
People who put their fingers on screens need to be forcefully reeducated. People who don't know how to clean a mouse need the same treatment, although optical mice many render this skill obsolete.
Bah! Spring for the 32MB and run either Linux/X/Opera 6 or Win98/Opera 6. A 486 is plenty enough CPU to run modern web software. Heck, if all you have is a 386/8MB, still go for Win3.1/Opera 3(iirc). Arachne's niche is truly obsolete hardware like less than 6MB RAM or less than i386.
IE runs fine performance-wise if it has a big fast machine all to itself. But when your CPU is heavily utilized it will bog, and it is just slow on an older system. Opera is still quick under low resource conditions. Opening new browser windows is also still quick even with low resources.
As far as stability goes, both Opera and IE are pretty darn stable in my experience, but IE still dies more often.
Opera will do form autocompletion, but only using the personal info fields you put into the preferences dialog. So for example your address and email work fine, if you entered them. Put in that first letter or digit and you get the popdown list; hit down arrow and tab off to the next field. Not as automatic as IE but works fine for most purposes.
Which reminds me, Opera's overall keyboard form handling is way superior to IE. Hit tab and you go to the first form field on screen; keep (shift-)tabbing and you navigate only among form fields. IE highlights all sorts of crap when you tab through a page; tab is mostly useless in IE.
Slashdot Mirror
That's http://www.inet-one.com/cypherpunks/dir.2002.06.24 -2002.06.30/msg00045.html to you.
More importantly: I also agree that NYTimes has no "right" to require the client-side rendering of their website to match NYT's "expectations."
If the NYT wants that, they should deliver their homepage in some kind of encrypted protected fixed PDF-like format that doesn't allow modification.
I would go further and say that if the NYT wants that, they should send out an armed manservant with a briefcase (containing the HTML) handcuffed to his arm every time someone requests a page. Because that is the only way they will be able to control the manner in which the public views the NYT's work. Once those little bits travel out over the network, the game is over. Joe End-User can filter or reformat or do whatever he pleases. As long as it's short of redistribution, there is no limit on what you can do to process your data. Using some cockamamie "secure" format is very much akin to copy protection in its other forms: an annoyance to all, but not an obstacle to the determined.
The bandwidth thing is a lot worse than everyone thinks. We all know that most of the lusers out there are on a 56k or worse. We should all know that they are not generally getting 56kb/s of bandwidth, that the common numbers are in the 28-44 kb/s range. But we should also keep in mind that people multitask. Poor idiots on a dialup are not connected 24/7, so they are more likely to have that Kazaa/service pack/game demo/background page load running. To work really well for a dialup user, a site should run nicely on half of a 30 kb/s link. That means responding well, say within 10 secs. on 2KB/s, which means loading 20KB/page or less. Much less would be better of course.
Bullshit. That's the way the Web works. Everyone understands the back button. In fact the very first things a new Web user learns are hyperlinks and the back button.
The problem is people who complicate their sites with half baked solutions to this non-problem, such as spawning new windows or the about.com style frameset mixing content from two sites.
Exactly. It's not secure telnet, it's the secure r* suite: rsh (both interactive and single command), rcp, and throw in port fowarding for good measure.
That's an interesting notion. But if someone was so fired up to ship unlicensed games, why were all the Dreamcast game releases licensed?
Now I do seem to recall that the commercial Pelican MP3 player was on a CD-R, but still, where were the unlicensed games? It was well known since about April 2000 how to boot on an unmodified Dreamcast.
And for that matter, you can boot on an unmodified PS1 at this point, I believe. Or someone could have included a $2 mod cartridge with their unlicensed title. Hmm.
Just to pile on here, Notes... UCE is never ok! None of your reasons matter a bit to me. I want _zero_ UCE. The only advertising email I will abide by is that which I asked for, which is generally none.
For this reason I would never have considered your company for anything after receiving your UCE.
I read Ultimate Guide to the VI and EX Text Editors and UNIX Power Tools, and I took a college course in intro to theoretical computer science that covered regular expressions, and I learned Perl regexes pretty well. This tutorial series at was pretty cool also.
I'll say one headless machine is drawing in the neighborhood of 200 watts. That's .2 KW. Per day, that's 4.8 KW-hours. One KW-hour costs neighborhood of a nickel, so one machine costs about 25 cents to run each day. 10,000 machines cost $2500 in electricity a day altogether.
You made a number of mistakes. The main one is: there is no notion of "watts per second" unless you are talking about a rate of change of power. 4.2M watts is a rate; it is the same whether measured over an instant, a second or an hour. If you use power at that rate for an hour you use 4.2M watt-hours. Not 15.1 billion watt-hours. So that was a factor of 3600 on the high side. Then you were off by a factor of about 14 on the low side in pricing kilowatt-hours. And I would say you were about a factor of 2 on the high side in wattage of a server. Altogether you were high by a factor of just about 500x which is the difference in our results.
funct.jpg mix.jpg mix2.jpg linux.jpg win2k.jpg winnt.jpg win95.jpg win98.jpg cisco2.jpg cisco.jpg aix.jpg openbsd.jpg obsdnew.jpg hpux11.jpg sol7.jpg sol8.jpg sol2ip.jpg bsdi.jpg irix.jpg macos1.jpg macos.jpg dnslibc.jpg dnssol.jpg comp.jpg random.jpg data.jpg mix.jpg linc.jpg ttime.jpg
Thank you vi!
Gawd, analogies are fun... Let's talk about bridge building.
This site at has some bridge history that I read recently. The Firth of Forth page in particular talks about a bridge that was built according to the maxim, "Everyone knows how a bridge should look." The result was an overbuilt, overcostly and ugly structure. A layman's intuition about what is necessary may be quite wrong.
The infamous Tacoma Narrows suspension bridge, of course, was built in modern times (1940) and according to conservative and well known principles, which didn't stop it from self-destructing. This was after more than fifty years of experience in designing large suspension spans. The judgement of an expert engineer with ample experience and time can also be very wrong!
Let's also consider the Charles River bridge, which looks like nothing you ever saw before. Nonetheless it is now being completed at a cost of $86M. To be sure, the design must have been reviewed, but if you saw a child's drawing of the bridge, you'd think it was a fantasy that could not be built. Again, so much for "Everyone knows how a bridge should look".
I agree with your basic point, which, as I understand it, is that the design must be visible and inspected to have a hope of correctness. But as bridge building shows us, that's not always enough!
Thanks for the link. That was one of the only on-topic comments in this whole story.
Were you using one or two monitors? Also I was surprised to see that you didn't go ahead and try some gigabit ethernet action since you were saturating your 100Mb link.
Do you see yourselves staying exclusively with the personal workstation model? What about driving these beasts using X over the network? You could buy a smaller farm of monster workstations and let your artists share them. Even in your high performance environment, "think time" has still got to have those systems idle more than half the time, not counting batch jobs of course. Not to mention, the slower simulation and filtering tasks could in principle be written to distribute themselves over a number of CPUs.
SGI's Visual Area Networking is their effort to use this kind of thinking.
For a while, in the 70's and 80's, companies poured money into their software engineering process, but recently they realized that publishing reports on how they cut debugging time by 66% might tell their competitors how to do it too.
Gimme a break. You think all the people who know how to do good software engineering are keeping quiet so the recipe for the secret sauce doesn't get out? You need a better conspiracy theory; this one ain't cutting it. Do they pull the strings of trade journals, too, so no one writes about the secrets of software success?
My own personal theory is that the typical software development organization is not in a competitive market with respect to software quality. They build bad software and get away with it, because for a variety of reasons their customers haven't called them on it. Many are serving internal customers; others serve customers who just don't care to discriminate between software that works well and software that doesn't. I am thinking of PC packaged software here.
Organizations with long memories or discriminating customers still have their relatively more developed software engineering practices.
Gaming is the last true hurdle for Linux and it's being jumped quickly.
Forget Linux: Gaming is the last true hurdle for PC compatibles, no matter what the O/S. And it's not being jumped quickly. The PC game market is a terrible technical environment. It's a nightmare of multiple hardware and software configurations, being called upon to perform fast and in real time with zero user configuration. Brutal price competition means good tech support would be an impossibly large cost. Short attention spans on the part of press and public mean a race to the bottom as far as bug count.
Having said all this, the game production companies are obviously going to continue to focus most of their effort on the 95% market share O/S. Linux and Mac gaming will continue to be a red headed stepchild, months or years behind and with a fraction of the titles. MS will continue to be very aggressive in game API software development, and API emulators will continue to lag these gaming APIs more than most any other APIs.
Original link had a gratuitous space courtesy of slashcode. Correct link: http://science.slashdot.org/comments.pl?sid=34509& cid=3742297
Body:
Google Answers (http://answers.google.com)
by Martin Spamer on 05:27 AM -- Friday June 21 2002 (Score:2) (#3742297)
(User #244245 Info) http://www.kitv.co.uk/ [ Neutral ]
What and When are you planning to to do about the frauds being perpetuated on Google Answers (http://answers.google.com) with questioners paying their own alias instead of the expert poster who deserves to be paid. They are destroying the cedibility of what has the potential to be an absolutely brilliant service.
I think you need to be a much more strigent about suspending and banning these abusers if you wish this service to take off.
Yeah, yeah, MS are evil. But there is far more packaged software development effort for Windows than any other platform. That's been the case for ten years. This effort has produced the most innovative and diverse set of independently developed packaged software products, not to mention MS's own estimable efforts.
It's a network effect, people! Everyone runs Windows, therefore everyone develops first/only for Windows, therefore everyone wants to run Windows. I would be pissed, too, if I had a Mac that had no ability to efficiently run all that software.
I realize I'm exaggerating a bit here. If you're a geek, then Linux is the place where you find the software you most desire. But for the mainstream that's not at all the case, not even close. It's normal and healthy to want to run all the software at CompUSA!
In the grand Slashdot grammar flame tradition: you meant "for C driver code, no less". Nonetheless, or nevertheless, means "however". "No less" is the strengthener you were looking for.
I say fire the new hire. If he doesn't love a text approach, he probably won't grow to love it. And if he never loves it he'll never get his productivity up to a very high level. There are plenty of workplaces where he can be comfortable in his IDE orientation.
Mice are the most pain-inducing device for me. The 3M mouse is the best mouse I've used. I recommend giving it a try if mice make you hurt. Works better for me than my old Logitech Trackman Marble FX.
Actually you have two choices:
Either way the license must be the GNU GPL. But you can, for example, sell the product to someone under the GPL, give them the source, but not give the source to all and sundry. This is essentially a technicality since the recipient of the source is free to redistribute. But the GPL doesn't mandate making the software available to everyone, nor does it mandate FSF style nominal fee distribution.
It will not be sufficient... There really is a Brenda in Marketing who edits images, maybe also video and 3D. She uses the hardware acceleration in her video card to make her session run at the rate to which she has become accustomed. She doesn't use a simple framebuffer on her current PC. What makes you think that a framebuffer with a relatively low-bandwidth high-latency connection to the CPU will suffice for her?
I really like the terminal model but it is not appropriate for every person in the enterprise. You cannot give someone a much slower computer than what they are using and have them accept it. The Sun Ray system was not built to compete with a high end graphics performance PC/workstation. A Sun Ray is not going to do it for the graphic professionals, not now, not ever. They will leave the company en masse.
SGI does sell some network terminal graphics systems that run fast, but they ain't Sun Ray.
similar corporate mindthink.
Mindthink, the worst sort of think!
Groupthink, perhaps.
People who put their fingers on screens need to be forcefully reeducated. People who don't know how to clean a mouse need the same treatment, although optical mice many render this skill obsolete.
Bah! Spring for the 32MB and run either Linux/X/Opera 6 or Win98/Opera 6. A 486 is plenty enough CPU to run modern web software. Heck, if all you have is a 386/8MB, still go for Win3.1/Opera 3(iirc). Arachne's niche is truly obsolete hardware like less than 6MB RAM or less than i386.
IE runs fine performance-wise if it has a big fast machine all to itself. But when your CPU is heavily utilized it will bog, and it is just slow on an older system. Opera is still quick under low resource conditions. Opening new browser windows is also still quick even with low resources.
As far as stability goes, both Opera and IE are pretty darn stable in my experience, but IE still dies more often.
Opera will do form autocompletion, but only using the personal info fields you put into the preferences dialog. So for example your address and email work fine, if you entered them. Put in that first letter or digit and you get the popdown list; hit down arrow and tab off to the next field. Not as automatic as IE but works fine for most purposes.
Which reminds me, Opera's overall keyboard form handling is way superior to IE. Hit tab and you go to the first form field on screen; keep (shift-)tabbing and you navigate only among form fields. IE highlights all sorts of crap when you tab through a page; tab is mostly useless in IE.