Some years ago, someone from the MIT Media lab was trying to recruit me. As we were walking across the MIT campus to the T station, it was sleeting, and he said "Out here there are fewer distractions".
I'm still in California.
I live in Silicon Valley, have since 1974, and live within walking distance of downtown Palo Alto. I went to Stanford, have been through some startups, and did reasonably well. So here's how I see it.
Stanford plays an interesting role. Stanford was started by a robber baron, and it still shows. Stanford isn't primarily a university. It's really a landowning company and investment bank that runs a school on the side for the tax break. This is clear if you look at Stanford's IRS filings. This works out quite well for all parties. Stanford's investment unit invests in private venture capital partnerships, which is an unusual investment for a university but works out well, because they have people who can evaluate which portfolios have enough potential winners to come out ahead.
The second item that made Silicon Valley is a little provision in the Californa Labor Code. This is the famous Section 2870:
(a) Any provision in an employment agreement which provides that an employee shall assign, or offer to assign, any of his or her rights in an invention to his or her employer shall not apply to an invention that the employee developed entirely on his or her own time without using the employer's equipment, supplies, facilities, or trade secret information except for those inventions that either:
(1) Relate at the time of conception or reduction to practice of the invention to the employer's business, or actual or demonstrably anticipated research or development of the employer; or
(2) Result from any work performed by the employee for the employer.
(b) To the extent a provision in an employment agreement purports to require an employee to assign an invention otherwise excluded from being required to be assigned under subdivision (a), the provision is against the public policy of this state and is unenforceable.
So what you do on your own time, unrelated to your employment, is yours. Period. And that's why employees can work on startups in their spare time.
Few other states have that, and it's never something that seems to come up when other places try to copy California, because employers hate it.
Then there's 3000 Sand Hill Road, the address known to everyone who's ever had anything significant to do with a startup. This is a quiet little place near the intersection of Sand Hill Road and Interstate 280. It looks like a nice little housing development composed of concentric rings. The outer ring has houses and condos, and is adjacent to a golf course. The middle ring has small offices. At the center is a restaurant, the Sundeck. It's all very peaceful, and there's no indication that you're in one of the world's great financial capitals. Except that there's a directory board. On that directory board are all the big names in venture capital. Even the VCs who've outgrown the place maintain offices there. It's unique in the world; all the big players are in one small place and talk to each other.
Silicon Valley as a center of innovation is kind of slow right now. The dot-com boom messed it up. Before the dot-com boom, Silicon Valley was about doing cool stuff. The dot-com boom was about retailing. And retailing people just aren't that innovative. The huge increase in land prices pushed manufacturing out of the Valley. Then engineering moved to follow the manufacturing. Now, Palo Alto is really a kind of retirement town; you see students and old people, but not that many twentysomethings.
In downtown Palo Alto, we lost Stacy's, one of the world's best technical bookstores, to get some store selling overpriced kitchen utensils. It's not clear if the Valley will come back, or remain the place you stay after you've made it.
The "you can talk to random people far away" thing isn't as exciting as it used to be. You have to be a really good electronic engineer to build anything half as good as off-the-shelf gear.
There are a small number of hams doing interesting stuff, like working on optimal modulation strategies for data over HF, but there aren't many. And the ones that do that typically are designing cell phones as their day job.
A few years back, someone re-edited "The Phantom Menace", and called it The Phantom Edit. Some people thought it was an improvement over the original. About 20 minutes of Jar-Jar was cut, which didn't hurt.
This works even better than the article says
on
Robo-Gecko Climbs Glass
·
· Score: 4, Informative
Cutkowsky has had this technology working for several years now. It's not just for glass; it works on many other building surfaces, too, like concrete walls. It doesn't require a smooth surface. They've had robots climbing up buildings at Stanford for a while now.
They have a new and powerful fabrication technique, too. They use a stereolithography machine to make their parts, but they use it in an unusual way. They use a machine that's intended to make multicolored objects from several different colored materials, and load it up with materials with different physical and electrical properties. So they can make a one-piece 3D part with soft parts and hard parts, or insulating parts and conductive parts. This is the beginning of a whole new kind of fabrication, which is what Cutkowsky is really into.
Fry's in Palo Alto just announced a price cut on the XBox 360. The base machine is still $299, but the higher-priced bundles just dropped over $100. This makes sense; some of the accessories were way overpriced, and there's a huge glut of unsold XBox 360 accessories left over from the holiday season.
eBay prices are now slighly below retail.
With that happening to the XBox 360, Sony is going to have real trouble at a higher price point.
On the developer front, the general reaction to the Cell processor is "groan". (Except for audio guys, who finally get their very own CPU.) The Xbox 360 is a 3-CPU shared memory multiprocessor driving a conventional graphics chip, something well-understood by developers. Porting from an x86 PC (or an original XBox, which is an x86 PC) to an XBox 360 is straightforward. The Cell is a new, wierd architecture, little limited-memory CPUs with bulk DMA access to main memory. (Architecture people will remember unsuccessful supercomputers of the past organized like this.) In fact, Sony already has had a huge architectural disaster. Originally, the Cell was supposed to do the rendering. That was a dud, and Sony had to put a conventional graphics chip on the back end, running up the cost.
It's certainly possible to develop good games for the thing, but the extra work required means the games willl be out later. It took about two years before the PS2 hardware was really being used effectively. The PS3 is completely different from the PS2 and will require new techniques. So Sony is launching late on a machine you can't just port to. Not good.
What's really going to happen is that the early PS3 games will be doing most of the game work in the main CPU and the graphics engine, mostly ignoring the Cell processors. If the game talks to the network, one of the Cell processors will be handling that. Audio work will be in a Cell processor. PS3 games will probably have really good sound, because there's plenty of extra Cell CPU capacity to devote to audio. As Lucasfilm people like to point out, good audio will compensate for lousy graphics, but the reverse isn't true.
It might make a nice render farm. But who supports Sparcs any more?
ResPower is successfully serving that market, for money, and their prices are good, probably better than Sun's. Supports 3DS, Lightwave, Maya, Blender, etc. 484 frames are rendering right now. "Over 12 million frames rendered".
ResPower is one of the very few "grid" companies actually selling a service for which people will pay.
Fixing the attack problem is going to require a combination of forensics work, investigation, and muscle. There are companies in that business, such as Kroll International and Securitas. These are the companies you call when there's a big problem. They have the resources to conduct an international investigation, from accountants to former British SAS people, and if it takes people with guns to solve the problem, they have them on the payroll. The bill might be in six or seven figures, but there are times when a company needs to spend that kind of money.
The basic requirement here is that DNS servers shouldn't be accepting queries from clients outside their local organizations. This is like the old "open relay" problem with SMTP. Obviously, such DNS servers have to be fixed. To force the issue, DNS servers queried by other DNS servers should find out if the querying server incorrectly accepts queries from the outside. If it does, that server is marked as a loser, and its queries get processed only after any other queries, and maybe with a deliberate delay. That should deal with the problem in the near term.
The stronger form of this protection is that many queries from loser servers are answered with an address that returns a page saying something like "Your DNS server at [xxx.xxx.xxx.xxx] has a problem and must be upgraded." The screaming users will get the problem fixed.
There doesn't seem to be anything on the site explaining what licensing terms apply to the content. If the Game Innovation Database uses the GFDL, like Wikipedia, then Wikipedia game content (of which there is rather a lot) can be moved to the Game Innovation Database. If CMU is taking an "all your base are belong to us" approach to content ownership, that can't happen.
Ironport went over to the dark side for a while, with their Bonded Spammer service and their "A-series" rackmount spam sending engines (only for "opt-in mailing", they claimed.) It got to be really embarassing; there were sites in both the Spamcop block list and the Bonded Sender approved list at the same time.
Ironport management finally decided they couldn't play both sides of the street, sold off Bonded Spammer to ReturnPath, and discontinued the "A-series". The A-series supposedly reaches end of life at the end of 2006, so there are probably still supported Ironport engines out there spamming away. After that, the community can consider whether Ironport is a white hat or not.
you'd just encode the MPEG2 file for the DVD demo and burn it to a data CD.
You can encode a CD blank in DVD format. This is nonstandard, but it works, sort of. Most software players on computers will play the thing. A few standalone DVD players will play it, but most will reject the disk, hang, or crash. It tends to work on low-end DVD players that use computer drives, because those drives can crank up a CD to 48x or so and get the data rate of a DVD.
Can you do the same thing to put HDTV on a DVD blank? That would be useful for short pieces, and very useful until low-cost writeable HD/Blu-Ray drives appear.
T1 is an asynchronous transmission system and was designed to handle differences in clocks.
If only. Old modem users will remember the regular, periodic appearance of junk characters when a T1 in the path was free-running, and on every bit slip, a phase error was inserted. It's still a problem. See the T1 Survival Guide.
That's what a core system just went for on eBay. So that's the real price, the amount a willing buyer will pay a willing seller. There's been slow price erosion on eBay, from the premium above retail at launch to slightly below it now. The "pent-up demand" is gone. When Microsoft relaunches the thing, they'll probably have to cut the price. (Or, more likely, throw in unsold game and accessory inventory.)
The PS3 is going to face some real price resistance. For most kids, it's only slightly better than the PS2, and for the parents, it's more than twice the price.
Dave Mills used to like to observe what happens after a leap second. Among other things, every generator on the power grid has to make sixty extra turns, which takes about four hours. Some computer clocks used to count the power line (this seems to be rare today) and you could watch, via NTP, the stress in the clock network as the power line clocks disagreed with the WWV clocks, and slowly came into synchronism.
Actually, synchronization is less important than it used to be, because more stuff is buffered. All three US television networks used to be locked together in frame sync to a master clock in New York, so that video sources could be switched without all the TV receivers rolling for a few frames. Now everything goes through frame buffers, so that's not an issue.
Similarly, US telephony used to be locked to a master clock in New Jersey, so that all the T1 lines ran in sync and bit for bit transfer worked. That's not as important as it used to be, with so many different transmission media, some synchronous and some packetized.
Actually, Intel has tried several times to change the I/O interface, but they usually interpose some proprietary programmable I/O controller between the peripheral and memory, which means that drivers need to be written for the controller/device combination. Then Microsoft gets upset because Intel is moving onto their turf. Then Intel backs down.
The real truth about microkernels is about like this:
Getting the architecture of a microkernel right is really hard. There are some very subtle issues in how interprocess communication interacts with scheduling. If these are botched, performance under load will be terrible. QNX got the performance part right. Mach got it wrong. Early Minix didn't address this issue. See this article in Wikipedia. Other big issues include the mechanism by which one process finds another, and how mutually mistrustful processes interact. If you botch the basic design decisions, your microkernel will suck. Guaranteed.
Most academic attempts at microkernels have been duds. One can argue over why, but it's the commercial ones, like QNX, VM, and KeyKos that work well, while the academic ones, like Mach, EROS, and the Hurd have been disappointing.
Security models really matter. And they're hard. Multics got this right. KeyKos got this right. QNX is no better than UNIX in this area.
Designers must work through "A can't do X, but A can trick B into doing X" issues.
Trying to turn a monolithic kernel into a microkernel doesn't work well. Mach, which started life as BSD UNIX, ran into this problem, which is why MacOS X isn't based on the microkernel version of Mach.
Drivers in user space have real advantages. Not only is the protection and restartability better, but because they have access to all the regular user program facilities, drivers for more modern devices are much easier. Things like Firewire and USB device discovery and hot-plugging reconfiguration are far easier at the user level, where you have threads, can block, and can call other programs. The old "top half and bottom half" driver approach doesn't generalize well to today's more dynamic configurations. Monolithic kernels have had to add kernel threads and dynamic loading of modules to handle all this, resulting in kernel bloat.
Of course, a big advantage of less-privileged drivers is blame management - you can tell whether the OS or the driver is at fault.
Startup requires more attention. A microkernel often doesn't contain the drivers needed to get itself started. So the startup and booting process is more complex. QNX has a boot loader which loads the kernel and any desired set of programs as part of the boot image. This gets the console driver and disk driver in at startup, without having to make them part of the kernel.
The performance penalty is real, but not that big There's a performance penalty associated with the extra interprocess communication. It's usually not that big, but there are areas where it's a problem. If it takes one interprocess call for each graphics operation, for example, performance will be terrible. NT 3.51 had a nice solution to this problem, designed by Dave Cutler. (NT 4 and later have a more monolithic kernel, but that had to do more with making NT bug-compatible with Windows 95 than with performance problems.)
I/O channels would help IBM mainframe channels, which have an MMU between the peripheral and main memory, are better suited to a microkernel architecture than the "bus" model the microcomputer world uses. In the mainframe world, the kernel can put program in direct communication with the hardware without giving it the ability to write all over memory. So there's little penalty for having drivers in user space. Which is why VM for IBM mainframes has been doing this for decades.
If you get it right, the kernel doesn't change much over time. This is the big win, and why microkernels become more stable over time. In the QNX world, USB and Firewire support were added with no changes to the kernel. (I wrote a FireWire camera driver for QNX, so I'm sure of this.)
The IBM VM kernel has changed little in decades.
So that's what you need to know about microkernels.
OK, if the telcos can't handle the bulk bandwidth from the central office to the server, I want to switch to a non-telco ISP for my DSL line. Covad, for example. Don't let the telecom monopolies prevent that. If they want to offer tiered services, they must be required to offer connectivity to DSL services other than their own.
Business Week is late on this one. We've already had the new Internet bubble, and are now starting the new Internet bubble collapse.
Tribe just had a layoff. The dating-service business has consolidated; most of the remaining sites are fronts for a few back-end companies.
The "blog" scene is cluttered to the point of near-uselessness. Most of the "Web 2.0" startups are me-too operations.
The real problem is that all these players are chasing the same pot of ad revenue. Most of them are bottom-feeders off Google AdWords or something similar. They're not selling anything themselves. There's no new product.
Lucasfilm is unionized. Parts of Pixar/Disney are unionized. Dreamworks is unionized. Many computer graphics artists and technicians are represented by The Animation Guild, local 839, IATSE. That doesn't seem to be hurting the creativity of those organizations.
IATSE is working on organizing game developers. There are some interesting things going on with EA in Redwood City, where both Dreamworks and EA have buildings in the same complex, with people doing roughly the same jobs. But the Dreamworks people have a union, reasonable hours, and overtime pay, while the EA people don't.
Most of this stuff is unreliable, has rendering artifacts, and has to use tiny type to fit in all the cutesy graphical elements. If you have to have cases based on which browser is rendering, you shouldn't be using that feature.
Absolute positioning in CSS was a horrible mistake. Don't use it. If tables can do the job, use tables. Try to write your own table engine in Javascript, and you'll botch it.
Overcomplicating a plain site is just stupid. If you want to make it look better, put more effort into the artwork, not the HTML/XML.
Slashdot Mirror
Some years ago, someone from the MIT Media lab was trying to recruit me. As we were walking across the MIT campus to the T station, it was sleeting, and he said "Out here there are fewer distractions". I'm still in California.
Stanford plays an interesting role. Stanford was started by a robber baron, and it still shows. Stanford isn't primarily a university. It's really a landowning company and investment bank that runs a school on the side for the tax break. This is clear if you look at Stanford's IRS filings. This works out quite well for all parties. Stanford's investment unit invests in private venture capital partnerships, which is an unusual investment for a university but works out well, because they have people who can evaluate which portfolios have enough potential winners to come out ahead.
The second item that made Silicon Valley is a little provision in the Californa Labor Code. This is the famous Section 2870:
So what you do on your own time, unrelated to your employment, is yours. Period. And that's why employees can work on startups in their spare time. Few other states have that, and it's never something that seems to come up when other places try to copy California, because employers hate it.
Then there's 3000 Sand Hill Road, the address known to everyone who's ever had anything significant to do with a startup. This is a quiet little place near the intersection of Sand Hill Road and Interstate 280. It looks like a nice little housing development composed of concentric rings. The outer ring has houses and condos, and is adjacent to a golf course. The middle ring has small offices. At the center is a restaurant, the Sundeck. It's all very peaceful, and there's no indication that you're in one of the world's great financial capitals. Except that there's a directory board. On that directory board are all the big names in venture capital. Even the VCs who've outgrown the place maintain offices there. It's unique in the world; all the big players are in one small place and talk to each other.
Silicon Valley as a center of innovation is kind of slow right now. The dot-com boom messed it up. Before the dot-com boom, Silicon Valley was about doing cool stuff. The dot-com boom was about retailing. And retailing people just aren't that innovative. The huge increase in land prices pushed manufacturing out of the Valley. Then engineering moved to follow the manufacturing. Now, Palo Alto is really a kind of retirement town; you see students and old people, but not that many twentysomethings. In downtown Palo Alto, we lost Stacy's, one of the world's best technical bookstores, to get some store selling overpriced kitchen utensils. It's not clear if the Valley will come back, or remain the place you stay after you've made it.
But it's been great fun being here.
There are a small number of hams doing interesting stuff, like working on optimal modulation strategies for data over HF, but there aren't many. And the ones that do that typically are designing cell phones as their day job.
A few years back, someone re-edited "The Phantom Menace", and called it The Phantom Edit. Some people thought it was an improvement over the original. About 20 minutes of Jar-Jar was cut, which didn't hurt.
Here's the web site for the project.
They have a new and powerful fabrication technique, too. They use a stereolithography machine to make their parts, but they use it in an unusual way. They use a machine that's intended to make multicolored objects from several different colored materials, and load it up with materials with different physical and electrical properties. So they can make a one-piece 3D part with soft parts and hard parts, or insulating parts and conductive parts. This is the beginning of a whole new kind of fabrication, which is what Cutkowsky is really into.
With that happening to the XBox 360, Sony is going to have real trouble at a higher price point.
On the developer front, the general reaction to the Cell processor is "groan". (Except for audio guys, who finally get their very own CPU.) The Xbox 360 is a 3-CPU shared memory multiprocessor driving a conventional graphics chip, something well-understood by developers. Porting from an x86 PC (or an original XBox, which is an x86 PC) to an XBox 360 is straightforward. The Cell is a new, wierd architecture, little limited-memory CPUs with bulk DMA access to main memory. (Architecture people will remember unsuccessful supercomputers of the past organized like this.) In fact, Sony already has had a huge architectural disaster. Originally, the Cell was supposed to do the rendering. That was a dud, and Sony had to put a conventional graphics chip on the back end, running up the cost.
It's certainly possible to develop good games for the thing, but the extra work required means the games willl be out later. It took about two years before the PS2 hardware was really being used effectively. The PS3 is completely different from the PS2 and will require new techniques. So Sony is launching late on a machine you can't just port to. Not good.
What's really going to happen is that the early PS3 games will be doing most of the game work in the main CPU and the graphics engine, mostly ignoring the Cell processors. If the game talks to the network, one of the Cell processors will be handling that. Audio work will be in a Cell processor. PS3 games will probably have really good sound, because there's plenty of extra Cell CPU capacity to devote to audio. As Lucasfilm people like to point out, good audio will compensate for lousy graphics, but the reverse isn't true.
The doctor will be in a call center in some low-wage country, of course.
ResPower is successfully serving that market, for money, and their prices are good, probably better than Sun's. Supports 3DS, Lightwave, Maya, Blender, etc. 484 frames are rendering right now. "Over 12 million frames rendered".
ResPower is one of the very few "grid" companies actually selling a service for which people will pay.
Fixing the attack problem is going to require a combination of forensics work, investigation, and muscle. There are companies in that business, such as Kroll International and Securitas. These are the companies you call when there's a big problem. They have the resources to conduct an international investigation, from accountants to former British SAS people, and if it takes people with guns to solve the problem, they have them on the payroll. The bill might be in six or seven figures, but there are times when a company needs to spend that kind of money.
The basic requirement here is that DNS servers shouldn't be accepting queries from clients outside their local organizations. This is like the old "open relay" problem with SMTP. Obviously, such DNS servers have to be fixed. To force the issue, DNS servers queried by other DNS servers should find out if the querying server incorrectly accepts queries from the outside. If it does, that server is marked as a loser, and its queries get processed only after any other queries, and maybe with a deliberate delay. That should deal with the problem in the near term.
The stronger form of this protection is that many queries from loser servers are answered with an address that returns a page saying something like "Your DNS server at [xxx.xxx.xxx.xxx] has a problem and must be upgraded." The screaming users will get the problem fixed.
There doesn't seem to be anything on the site explaining what licensing terms apply to the content. If the Game Innovation Database uses the GFDL, like Wikipedia, then Wikipedia game content (of which there is rather a lot) can be moved to the Game Innovation Database. If CMU is taking an "all your base are belong to us" approach to content ownership, that can't happen.
Ironport management finally decided they couldn't play both sides of the street, sold off Bonded Spammer to ReturnPath, and discontinued the "A-series". The A-series supposedly reaches end of life at the end of 2006, so there are probably still supported Ironport engines out there spamming away. After that, the community can consider whether Ironport is a white hat or not.
You can encode a CD blank in DVD format. This is nonstandard, but it works, sort of. Most software players on computers will play the thing. A few standalone DVD players will play it, but most will reject the disk, hang, or crash. It tends to work on low-end DVD players that use computer drives, because those drives can crank up a CD to 48x or so and get the data rate of a DVD.
Can you do the same thing to put HDTV on a DVD blank? That would be useful for short pieces, and very useful until low-cost writeable HD/Blu-Ray drives appear.
If only. Old modem users will remember the regular, periodic appearance of junk characters when a T1 in the path was free-running, and on every bit slip, a phase error was inserted. It's still a problem. See the T1 Survival Guide.
The PS3 is going to face some real price resistance. For most kids, it's only slightly better than the PS2, and for the parents, it's more than twice the price.
Actually, synchronization is less important than it used to be, because more stuff is buffered. All three US television networks used to be locked together in frame sync to a master clock in New York, so that video sources could be switched without all the TV receivers rolling for a few frames. Now everything goes through frame buffers, so that's not an issue.
Similarly, US telephony used to be locked to a master clock in New Jersey, so that all the T1 lines ran in sync and bit for bit transfer worked. That's not as important as it used to be, with so many different transmission media, some synchronous and some packetized.
Actually, Intel has tried several times to change the I/O interface, but they usually interpose some proprietary programmable I/O controller between the peripheral and memory, which means that drivers need to be written for the controller/device combination. Then Microsoft gets upset because Intel is moving onto their turf. Then Intel backs down.
This is getting bad.
So that's what you need to know about microkernels.
OK, if the telcos can't handle the bulk bandwidth from the central office to the server, I want to switch to a non-telco ISP for my DSL line. Covad, for example. Don't let the telecom monopolies prevent that. If they want to offer tiered services, they must be required to offer connectivity to DSL services other than their own.
The real problem is that all these players are chasing the same pot of ad revenue. Most of them are bottom-feeders off Google AdWords or something similar. They're not selling anything themselves. There's no new product.
1991, when they dropped out of the computer business.
2. When was the last Electronic Arts layoff?
Feb. 2006 , 5% of EA's Redwood City workforce was laid off. This followed the 2005 layoffs in LA.
That's exactly what Java applets do. If you want to run an application in the browser, that's the appropriate mechanism.
IATSE is working on organizing game developers. There are some interesting things going on with EA in Redwood City, where both Dreamworks and EA have buildings in the same complex, with people doing roughly the same jobs. But the Dreamworks people have a union, reasonable hours, and overtime pay, while the EA people don't.
Absolute positioning in CSS was a horrible mistake. Don't use it. If tables can do the job, use tables. Try to write your own table engine in Javascript, and you'll botch it.
Overcomplicating a plain site is just stupid. If you want to make it look better, put more effort into the artwork, not the HTML/XML.