... when someone offers me an IP-based service that stays up, even when the electricity goes down, and has the long-running reliability of my landline phone.
I have VOIP with Vonage, connected to my cable modem, and a cellphone, but those are for the backup and away-from-home lines. The call-911-and-it-won't-fail line is POTS, thank you very much.
Once a product is established as a category leader, its ergonomics become the default expectation. Even if those ergonomics aren't particularly good, you have to beat them by a lot to get people to switch, or even to consider you as an alternative.
The iPod became the category leader in portable digital music because its (ergonomics in your pocket + iTunes client + iTunes store) were better than anything else at the time. There are now a lot of people who think (iPod + iTunes) are the way things are supposed to work, and minor ergonomic refinements aren't going to motivate them to explore alternatives.
Need I point out the parallel with MS Windows? It became dominant because its (tolerable ergonomics + lower cost of entry + backward compatibility) meant that a lot of people saw Windows first, as the way WIMP interfaces are supposed to work. Macs have always had better ergonomics, but not enough better to get people to switch on that alone.
The question is, would Dell be able to manufacture something to Jobs' standards of design and taste? Judging from their previous efforts, not unless Apple did most of the design and vetted the parts suppliers - in which case, why involve Dell at all?
Those of you who are up on the current state of the art here, please help me out. I was under the impression that multiple threads and automatic storage management were still not on good terms with each other, and that this was a big unsolved problem.
The admin people who are responsible for your paperwork will not get in trouble if they delay or lose your documents - you are pretty much at their mercy, relying on their skill and desire to do a good job. Those of you at state-run institutions can stop laughing now.
I have found that if you need special service from them, like when you miss a deadline, it helps if you start your request with the phrase "I am an idiot."
Today, x86 chips are a thin layer of instruction translation on a thoroughly modern core.
That translator is the bucket I speak of.
It makes it harder to optimize code, because it puts yet another layer of obscurity between the external model of execution and reality. I bet this is a major contributor to why Intel's compilers produce the best x86 code - only they really understand this translation process. Raising the bar for compiler implementors is good for Intel, but not for you as a compiler user.
It is the moral equivalent of another set of stages in the instruction pipeline, but one with no benefit for execution since it's doing "useless" work.
It chews up silicon that could be used for bigger local caches in simpler architectures.
Making the x86 instruction set hard to implement fast is in Intel's interest, because it makes it harder for competitors to challenge them. As long as this strategy is successful, they'll keep doing it, and if they ever manage to run AMD out of business, their next move will be to ratchet CPU prices way up. On that day, we'll collectively wish that Apple, IBM, and Motorola had kept at least one more architecture somewhat viable on the desktop.
I had so hoped, though, that we were finally going to get beyond the x86 architecture - that their strategy of piling kluge on top of kluge on top of kluge in the name of backwards-compatibility was finally going to come crashing down.
That the chip guys could start spending resources on actual innovation in hardware design, without having to keep one foot in the bucket of x86 binary compatibility.
That PowerPC, or the Cell, or anything with less than thirty years of binary baggage, might get out ahead and stay there long enough to put x86 to rest.
The tarball for the above weighs in at 200 KB, and it's a parser, plus a complete language implementation, plus non-trivial example code for things like FTP and HTTP clients. That's what I think of when I think of "lightweight".
Apple tends to set a price point and add more stuff to that point rather than keep the stuff the same and lower the price.
Witness the white iBook - the bottom-of-the-line notebook has always been $1000 - $1200 or so. Instead of changing that price much, Apple has upped the standard equipment over time: CD -> CD Burner/DVD, 128 -> 256 MB RAM, wireless card optional -> included.
If you must have the absolute lowest price up-front, you regrettably don't want Apple. If you're willing to spend a little more up front to get something you'll keep for a long time, Apple starts looking better - I just installed 10.4 on my 500 MHz G3 iBook (purchased early 2002) and expect to get a few more years out of it.
Has any progress been made in the last few years on improving the state of government computer security?
NASA used to be the part of the US Government with the cyber "Kick Me" sign on its butt - lots of machines, administered mostly by scientists who cared more about getting their work done than security. The prevalence of desktop Macs was probably the only thing that kept the place from being totally 0wned.
No more, at least at Goddard where I work. Machines are much better administered, and we live behind a very tight firewall. Network security routinely wardrives the grounds, and people with guns (!) will show up if you have an unauthorized wireless access point.
People who have the $$$$ will get stem cells created for them and kept on the shelf in labs. After all, that's what's great about stem cells - they'll reproduce indefinitely, until you put them in an environment where they get the right cues to specialize.
If stem cells turn out to be the universal replacement part bank, their creation and maintenance will probably start out as an expensive boutique service, mushroom into big business, and end up as a government-subsidised service in developed countries. Imagine egg donation being promoted as a public service, like blood donation is today.
At least, not at Goddard where I work. NASA used to be an easy target for crackers, but we've tightened up a lot since those days. Network security around here wardrives the grounds, and people with guns (!) will show up if they detect an unauthorized access point.
A good summary of the two main branches of lisp machine history is here. I personally first saw both a PERQ and a CADR at IJCAI 81.
MS really has only two revenue streams
on
Gates on Google
·
· Score: 1
Or three, depending on how you count:
Windows OS
Office
Macintosh business unit (which is really mostly the Office port)
Nothing else at MS makes money. The OS and Office are such money fountains that nothing else has to.
Integration was not innovation for MS
on
Gates on Google
·
· Score: 2, Insightful
MS's only big software innovation has been integration. They realised that people don't want programs. They want a computer.
Wasn't this "innovation" copied from the Macintosh?
(who copied it from Xerox, who copied it from Doug Englebart...)
To my knowledge, MS has only tried major innovation once. The result was Microsoft BOB.
However, where MS is _really_ innovative is in marketing. They have found ways to promote and market software that no-one else has ever thought of. Now, those ways may not be 'nice' but they are certainly innovative.
No argument there. Of course, many of those marketing innovations were eventually found illegal...
Guy Steele is relatively unknown outside the Common Lisp and Scheme communities, and he's overshadowed in the Java pantheon by Gosling, but he's had more influence on more programming languages used by more people than anyone else I can think of:
Primary author of Common Lisp the Language, the community-generated pre-spec for Common Lisp
The other half of Steele and Sussman, co-inventors of Scheme
Co-author of The C Programming Language by Harbison and Steele, which codified many of the techniques that made portable C code possible
As co-author of The Java Programming Language Specification, he reportedly brokered many design compromises between Bill Joy and James Gosling
Given his track record, I wouldn't bet against him if he says he's going to create a worthy successor to Fortran.
I think it would be a pretty horrible world if users just put all their stuff in the same folder - with no attempts to categorize things.
Don't know if this is true with OS X, but other Un*xes I've used have file system performance go to hell when individual folders have lots of files in them. Spotlight-dependent people may just dump everything in ~/Documents (default destination for lots of applications), then wonder why their system slows to a crawl when they open a document or create a new one.
I assume Apple has thought of this already and hacked the hell out of the file system for scalable performance on large directories... right?
Over here. The article linked to at the bottom is more detailed and convincing than Raymond's blog entry - do read it.
At the very least, Kuhn has been wrong for the last hundred years or so. His major accomplishment was to give a name ("paradigm shift") to a phenomenon that had been going on for awhile in the art world - you get attention for your work by making it radically different from everything else. After the one-two punch of relativity and quantum mechanics, lots of people dreamed of making a "paradigm-shattering" discovery in the sciences, and a bunch of them did, in biology, paleontology, geology, etc.
Sure, scientists are human, and there are fads and fashions in research and publishing. But, swallowing Kuhn whole is well down the path towards deconstruction in the sciences, which I think should be vigorously opposed.
One way to level the playing field would be to make the "May contain..." option above illegal. The reason for marking should be to let people know which patents are getting used - it's not useful to tell the consumer or potential reverse-engineer that "one of these 10,000 patents may be inside this product".
If correct, explicit marking becomes a requirement, big libraries of software patents become less useful:
If you license them you have to watch your licensees
If you don't license them, and don't use them yourself for commercial advantage, they're only useful in lawsuits to threaten competitors - a pretty weak reason for a state-granted monopoly
... is fine by me. I especially like the requirement that marking would put on the patent owners to "police their licensees". According to the FA, if B licenses a patent from A, A must watch B and insure that when B sells something using A's patent, B marks it with the appropriate patent numbers. If A lets B get away without marking, A loses its rights to enforce the licensed patents.
Currently when the big guys cross-license their patent libraries, usually no money changes hands and they don't change how they do business - it's a CYA formality used to make life difficult for businesses without big libraries. If marking were required, licensees would either have to:
read the libraries and pick the patents that they actually used
mark every product with "May contain technology based on... 10,000 patent numbers"
risk invalidating patents licensed to them, and getting sued into oblivion by the owners
Keeping a licensed patent in force would cost something if marking were required. Keeping a lot of patents in force would cost a lot. What's not to like?
I have a blind friend who is a Windows victim primarily because:
He needs to exchange MS documents with other people
Windows has screen readers like Jaws which make it possible for him to use web browsers and other standard WIMP software
I'm constantly going over to his place and:
Installing or configuring software that doesn't work with his screen reader due to non-conformance with UI guide lines
Fixing stuff that breaks because of malware or the general fragility of Windows (and this is XP, SP2, folks, complete with automatic updates)
With VoiceOver as part of the OS, a Mac mini with VoiceOver will now be cheaper than a PC with a third-party screen reader.
Also, it will likely work a lot better for him, because Mac applications tend to conform to Mac UI guidelines and will thus mesh better with VoiceOver.
I'm going to encourage him to Switch the next time his Windows box hoses itself.
In the CEO biz, your total compensation is the way you get compared to other CEOs. It appears to be a kind of penis-measuring exercise (female CEOs aside) - after all, does a $20M CEO really work twice as hard as a $10M CEO? The usual justification for big CEO pay is "everyone else does it".
I hope the $1/year salary is their way of saying "we may be a public company, but we aren't going to play those games - we run Google because we want to solve hard problems and make money at it, not so we can wave our paychecks at Yahoo's management and laugh about how small they are."
Mac OS's Software Update does not automatically install patches. The most you can do is to have it download stuff in the background and make it available - you still have to tell it to patch, and you have to give it an administrator password for anything dangerous.
Apple does both security patches and point releases between major yearly (or so) updates. It's rare for either kind of patch to break existing applications - the recent spate of point releases that broke stuff was news because it's rare.
I think it's legitimate to beat Microsoft up on security and patching strategies when other commercial entities exist that do them better on all counts.
Slashdot Mirror
... when someone offers me an IP-based service that stays up, even when the electricity goes down, and has the long-running reliability of my landline phone.
I have VOIP with Vonage, connected to my cable modem, and a cellphone, but those are for the backup and away-from-home lines. The call-911-and-it-won't-fail line is POTS, thank you very much.
Once a product is established as a category leader, its ergonomics become the default expectation. Even if those ergonomics aren't particularly good, you have to beat them by a lot to get people to switch, or even to consider you as an alternative.
The iPod became the category leader in portable digital music because its (ergonomics in your pocket + iTunes client + iTunes store) were better than anything else at the time. There are now a lot of people who think (iPod + iTunes) are the way things are supposed to work, and minor ergonomic refinements aren't going to motivate them to explore alternatives.
Need I point out the parallel with MS Windows? It became dominant because its (tolerable ergonomics + lower cost of entry + backward compatibility) meant that a lot of people saw Windows first, as the way WIMP interfaces are supposed to work. Macs have always had better ergonomics, but not enough better to get people to switch on that alone.
The question is, would Dell be able to manufacture something to Jobs' standards of design and taste? Judging from their previous efforts, not unless Apple did most of the design and vetted the parts suppliers - in which case, why involve Dell at all?
Those of you who are up on the current state of the art here, please help me out. I was under the impression that multiple threads and automatic storage management were still not on good terms with each other, and that this was a big unsolved problem.
The admin people who are responsible for your paperwork will not get in trouble if they delay or lose your documents - you are pretty much at their mercy, relying on their skill and desire to do a good job. Those of you at state-run institutions can stop laughing now.
I have found that if you need special service from them, like when you miss a deadline, it helps if you start your request with the phrase "I am an idiot."
That translator is the bucket I speak of.
It makes it harder to optimize code, because it puts yet another layer of obscurity between the external model of execution and reality. I bet this is a major contributor to why Intel's compilers produce the best x86 code - only they really understand this translation process. Raising the bar for compiler implementors is good for Intel, but not for you as a compiler user.
It is the moral equivalent of another set of stages in the instruction pipeline, but one with no benefit for execution since it's doing "useless" work.
It chews up silicon that could be used for bigger local caches in simpler architectures.
Making the x86 instruction set hard to implement fast is in Intel's interest, because it makes it harder for competitors to challenge them. As long as this strategy is successful, they'll keep doing it, and if they ever manage to run AMD out of business, their next move will be to ratchet CPU prices way up. On that day, we'll collectively wish that Apple, IBM, and Motorola had kept at least one more architecture somewhat viable on the desktop.
I had so hoped, though, that we were finally going to get beyond the x86 architecture - that their strategy of piling kluge on top of kluge on top of kluge in the name of backwards-compatibility was finally going to come crashing down.
That the chip guys could start spending resources on actual innovation in hardware design, without having to keep one foot in the bucket of x86 binary compatibility.
That PowerPC, or the Cell, or anything with less than thirty years of binary baggage, might get out ahead and stay there long enough to put x86 to rest.
Dammit!
The tarball for that weighs in at 300K. The language it parses is ludicrously complex.
http://people.delphiforums.com/gjc/siod.html/
The tarball for the above weighs in at 200 KB, and it's a parser, plus a complete language implementation, plus non-trivial example code for things like FTP and HTTP clients. That's what I think of when I think of "lightweight".
I believe that's the first time I've ever seen "XML" and "lightweight parser" in the same sentence.
And, relative to MS Word, it's probably accurate!
Apple tends to set a price point and add more stuff to that point rather than keep the stuff the same and lower the price.
Witness the white iBook - the bottom-of-the-line notebook has always been $1000 - $1200 or so. Instead of changing that price much, Apple has upped the standard equipment over time: CD -> CD Burner/DVD, 128 -> 256 MB RAM, wireless card optional -> included.
If you must have the absolute lowest price up-front, you regrettably don't want Apple. If you're willing to spend a little more up front to get something you'll keep for a long time, Apple starts looking better - I just installed 10.4 on my 500 MHz G3 iBook (purchased early 2002) and expect to get a few more years out of it.
NASA used to be the part of the US Government with the cyber "Kick Me" sign on its butt - lots of machines, administered mostly by scientists who cared more about getting their work done than security. The prevalence of desktop Macs was probably the only thing that kept the place from being totally 0wned.
No more, at least at Goddard where I work. Machines are much better administered, and we live behind a very tight firewall. Network security routinely wardrives the grounds, and people with guns (!) will show up if you have an unauthorized wireless access point.
People who have the $$$$ will get stem cells created for them and kept on the shelf in labs. After all, that's what's great about stem cells - they'll reproduce indefinitely, until you put them in an environment where they get the right cues to specialize.
If stem cells turn out to be the universal replacement part bank, their creation and maintenance will probably start out as an expensive boutique service, mushroom into big business, and end up as a government-subsidised service in developed countries. Imagine egg donation being promoted as a public service, like blood donation is today.
At least, not at Goddard where I work. NASA used to be an easy target for crackers, but we've tightened up a lot since those days. Network security around here wardrives the grounds, and people with guns (!) will show up if they detect an unauthorized access point.
A good summary of the two main branches of lisp machine history is here. I personally first saw both a PERQ and a CADR at IJCAI 81.
Windows OS
Office
Macintosh business unit (which is really mostly the Office port)
Nothing else at MS makes money. The OS and Office are such money fountains that nothing else has to.
Wasn't this "innovation" copied from the Macintosh?
(who copied it from Xerox, who copied it from Doug Englebart...)
To my knowledge, MS has only tried major innovation once. The result was Microsoft BOB.
No argument there. Of course, many of those marketing innovations were eventually found illegal...
Primary author of Common Lisp the Language, the community-generated pre-spec for Common Lisp
The other half of Steele and Sussman, co-inventors of Scheme
Co-author of The C Programming Language by Harbison and Steele, which codified many of the techniques that made portable C code possible
As co-author of The Java Programming Language Specification, he reportedly brokered many design compromises between Bill Joy and James Gosling
Given his track record, I wouldn't bet against him if he says he's going to create a worthy successor to Fortran.
Don't know if this is true with OS X, but other Un*xes I've used have file system performance go to hell when individual folders have lots of files in them. Spotlight-dependent people may just dump everything in ~/Documents (default destination for lots of applications), then wonder why their system slows to a crawl when they open a document or create a new one.
I assume Apple has thought of this already and hacked the hell out of the file system for scalable performance on large directories... right?
Over here. The article linked to at the bottom is more detailed and convincing than Raymond's blog entry - do read it.
At the very least, Kuhn has been wrong for the last hundred years or so. His major accomplishment was to give a name ("paradigm shift") to a phenomenon that had been going on for awhile in the art world - you get attention for your work by making it radically different from everything else. After the one-two punch of relativity and quantum mechanics, lots of people dreamed of making a "paradigm-shattering" discovery in the sciences, and a bunch of them did, in biology, paleontology, geology, etc.
Sure, scientists are human, and there are fads and fashions in research and publishing. But, swallowing Kuhn whole is well down the path towards deconstruction in the sciences, which I think should be vigorously opposed.
If correct, explicit marking becomes a requirement, big libraries of software patents become less useful:
If you license them you have to watch your licensees
If you don't license them, and don't use them yourself for commercial advantage, they're only useful in lawsuits to threaten competitors - a pretty weak reason for a state-granted monopoly
Currently when the big guys cross-license their patent libraries, usually no money changes hands and they don't change how they do business - it's a CYA formality used to make life difficult for businesses without big libraries. If marking were required, licensees would either have to:
read the libraries and pick the patents that they actually used
mark every product with "May contain technology based on ... 10,000 patent numbers"
risk invalidating patents licensed to them, and getting sued into oblivion by the owners
Keeping a licensed patent in force would cost something if marking were required. Keeping a lot of patents in force would cost a lot. What's not to like?
He needs to exchange MS documents with other people
Windows has screen readers like Jaws which make it possible for him to use web browsers and other standard WIMP software
I'm constantly going over to his place and:
Installing or configuring software that doesn't work with his screen reader due to non-conformance with UI guide lines
Fixing stuff that breaks because of malware or the general fragility of Windows (and this is XP, SP2, folks, complete with automatic updates)
With VoiceOver as part of the OS, a Mac mini with VoiceOver will now be cheaper than a PC with a third-party screen reader.
Also, it will likely work a lot better for him, because Mac applications tend to conform to Mac UI guidelines and will thus mesh better with VoiceOver.
I'm going to encourage him to Switch the next time his Windows box hoses itself.
In the CEO biz, your total compensation is the way you get compared to other CEOs. It appears to be a kind of penis-measuring exercise (female CEOs aside) - after all, does a $20M CEO really work twice as hard as a $10M CEO? The usual justification for big CEO pay is "everyone else does it".
I hope the $1/year salary is their way of saying "we may be a public company, but we aren't going to play those games - we run Google because we want to solve hard problems and make money at it, not so we can wave our paychecks at Yahoo's management and laugh about how small they are."
Does this mean we'll see a bunch of new unsecured wireless APs soon?
Mac OS's Software Update does not automatically install patches. The most you can do is to have it download stuff in the background and make it available - you still have to tell it to patch, and you have to give it an administrator password for anything dangerous.
Apple does both security patches and point releases between major yearly (or so) updates. It's rare for either kind of patch to break existing applications - the recent spate of point releases that broke stuff was news because it's rare.
I think it's legitimate to beat Microsoft up on security and patching strategies when other commercial entities exist that do them better on all counts.