There's lots of fear that Google will turn evil, and lots of folks (however biased) who would say they already have.
I tell Google what I want to know. That I want to know something about a certain topic is a datapoint. They take note of that, and who I am. They use click-through metrics to discover whether I found their results interesting. When I buy a product from an advertiser, they trace it back to my original search and use that to optimize the advertising to me to be the products I'm most interested in. Among other things they use that info to improve their search, to target their advertising, to gauge interest in topics and products across areas, regions, nations and the world.
This is no different than Walmart using their sales register metrics to discover that ice cream sells better after a hurricane.
In practice it's like having your own Concierge striving to anticipate your every desire. Don't ask what's available - say what you want. Anything can be had for ready cash, from a glass of Pinot Grigio to the robes of a senator with the senator still inside. (Apologies to RAH).
Does this aggressive quality of service erode my privacy? Well, yes it would, if I expected privacy on an Internet connection that maps to my home address and billing information. I haven't ever been that dumb. If I want to do something personal and private on the Internet I use my BSD box and my neighbor's open wireless access point to VPN to the offshore RDP account I paid for with eGold that I bought online using a disposable credit card I paid cash for, like anybody else who wants real privacy. You don't get privacy - you take it.
XBox Profits up, sales down. If they get a long tail they may break even - and for Microsoft that's doing well. What does it say for a company that they hope to get out what they put in over a decade, inflation notwithstanding? And what about the failure rate? Is putting out products that fail half the time harming their brand? That's a reasonable expectation. If Boeing planes failed to stay airborne, or Toyota cars failed to operate safely less than 99.99% of the time, we'd have a serious issue with that. Somehow though Microsoft is getting away with quality control that nets 0.5 9's. How is that even possible?
Actually, we don't know what sort of voodoo Steve Jobs used on the media heads to get them to put their content on iTunes under fairly reasonable terms. They were still working PlaysForNow at the time. It must have been some persuasive stuff. I'm thinking Quicktime video of studio executives snorting heroin off the nether parts of recently dead OD'd boyband members, but that may be my imagination getting the best of me. Maybe it was simple consensual human/gerbil interaction. I'm sure it wasn't reason though - they don't understand the term.
This part of the Book of Jobs will probably never be written. Some things are best left a mystery.
That'll build market share and ad revenues. Until you stop paying people to pretend to use your search engine to find stuff to buy. Then they abandon you and you'll find you've flushed a bunch of cash for absolutely nothing.
You're right! People have to be more forward thinking than that. It could take a few more decades at several billion dollars a year for Microsoft's Internet strategy to break even, but then these naysayers will be really sorry. Stick with the company that's not afraid to put in the investment and stick with a long term strategy.
Why is noone asking to examine the C programming language, to adapt it to modern programming processes and methods? Why is noone speaking out to defend C as a useful language, to update the aged methods of Kernighan and Ritchie?
Microsoft? Credible? On what planet? They didn't even write the DOS that was their reason for being. IBM helped them out with W32, and suffered from it when they killed OS/2. Dave Cutler took the features they needed from VMS to create NT - and today he's filing the serial numbers off of EC2 to complete their cloud offerring. They have been a sham this whole time and no change from that paradigm is anticipated.
This reads to me a little bit more like "They promised they'd take care of me for destroying the credibility of ISO. It's been years, and still no check!"
You couldn't watch this train wreck without knowing that Alex Brown had sold his soul. He jammed this thing through, destroying the value of ISO in the process. Now he's learning that if you dance with the devil, you pay his fee. We told him that then to no effect, and telling him again will do no better.
From here no slack for him. He did what he did and not only he but we have to live with it. He'll be a pariah forever.
On the upside, managers everywhere have the ISO standard to cite for their completely proprietary installations of Office and Exchange. Who knows, but maybe soon they'll have a similar standard to cite for IE: "It's standard, in the hypothetical sense that they sell a lot of it and so the standards bodies should write their standards to it."
It doesn't surprise me that a few HP geeks are doing arithmetic analysis of what's supposed to be art. No doubt this model will be used to bang out a bunch of crap optimized for it, which will then disprove the model. It would not surprise me if a niche industry in punitry alts didn't spring up so that people could blog every possible permutation of like/dislike early and so market the winning alts as "market drivers" that could be marketed to studios. Why not? We're already doing that here on slashdot and other tech sites.
These guys should get laid. Then maybe they'd put down the slide rule and enjoy the movie.
Heh. It's my lack of understanding. I just don't "get it". Yeah, that's my problem. If only I understood the ROI proposition properly presented then I would grasp the essential nature of Oracle's value-add.
Bite me. It's tables and joins, SQL and IOPs. Oracle has no magic bits.
For GPL works you take a distro, file off the serial numbers somebody else was using (usually this is just trademark logos and such), stamp your name on it, and it's DeadPixelOS. Of course, DeadPixelOS isn't going to get much of a following unless you're continuously developing some value-add as well as keeping up with the patch management of 1000+ packages. It's some work. For your first clone distro I'd start with a low-maintenance one like Pentoo.
The line is drawn in the license. For OpenSolaris, that would be here. If the license says you can do what you want with the code, then you can. If it says you can't, then you can't. And if it doesn't say, then it's all about how good your lawyer is and how much you want it.
Will. Oracle is not in the business of giving stuff away for free.
Have you heard? They license their database software not by the servers it runs on, nor by the processor, but by the core. How absurd is that? Does it cost them more to produce a database that works on more than 4 cores, or to support it? Believe it or not, they also charge extra for installed memory, as if that had anything to do with their production or support costs. Failover? Now you're into serious money. And don't you dare run it on stuff that's not on the secret list, or your support contract is invalid.
If Cisco's motto is "that feature is enabled through the purchase of an optional license", Oracle's is more so.
I guess Oracle doesn't get that we have options, and the pace of hardware technology will quickly erase any software advantage they think they have.
We all know that one major problem with the Microsoft platform is that it's homogeneous. No matter how many times we hear the "ground up" reengineering story, we get these exploits that work vulnerabilities in a common code base. All of the platforms use the same code. All code has bugs, and one bug might grant entry, while two more might grant privilege escalation, and so once an exploit is found all the machines with that code base are pwned. The solution to this problem is deviously simple: do everything differently on every machine. No, I'm not talking about ASLR here, though that's a start.
Stop. I know the first reaction to that is "that's crazy talk". This is pretty revolutionary thinking. It's not possible to design a unique operating system for every user. It is however possible to avoid the complementary vulnerability trampoline by varying the ways that components implement various technologies.
Every action that a machine can perform can be done in various ways - various algorithms can be used to achieve the same result, and some algorithms are more efficient than others. As a part of development many of these ways are explored and until now all but one was discarded. Simply by retaining the discarded algorithms, exploring the variations permissible within the defined interface, and retaining each functional implementation as a heuristic option allows the system designer to thwart the advantage of the large static target. The varying algorithms can be distributed randomly across the installed base as polymorphic patches. As long as the variant algorithms are strictly conformant to the well-defined interfaces, and the interfaces are well designed, it works. The downside to this is that some algorithms are, let's face it - sub-optimal. The diversity of algorithms is an advantage here as a feedback mechanism will reveal optimizations that yield net losses due to secondary effects. This will winnow the dozens of algorithms to a few. Even with only a few performant options per algorithm given the vast number of subsystems in a desktop or server operating system, we'll not run out of permutations before the end of time.
When each subsystem might be any one of several implementations that achieve the same object, the monolithic cathedral of code with a universal backdoor is prevented. Patches can randomly rotate the heuristic until the exploitability of individual platforms is not predictable. Performance of an individual system will vary to a degree, but not necessarily so in net - the distribution of performant vs sub-optimal algorithms can be intelligently distributed so that they average out and one system doesn't have all sub-standard algorithms. Positive feedbacks can indicate exploited components and replace them in an evolutionary fashion before they can be combined synergistically into a chain of exploits that go from basic entry to system privilege. The feedback can also gauge the quality status of the code, and with proper tracking lead back to the outstanding developer for recognition (or the leakmaster for reassignment).
Oh, and no patenting this stuff you bastards! This comment is prior art (ok, I adapted the ideas from some 1980's AI research and Conway's Life - but you can't prove that. Regardless, you didn't invent this stuff and the patents are NOT YOURS).
This is not a greek tragedy - it's a cheesy cliffhanger serial that has the hero facing near certain death at every turn. It won't end until advertisers won't buy time on the reruns.
[clerk] Come ye all presents to hear the case of SCO v. Novell
[Judge] Says here in the complaint that Novell slandered SCO's ownership of this "unix" copyright thing. Plaintiff, what say you?
[SCO] They did say we don't own it.
[Judge] Defendant, what say you?
[Novell] Yes, your honor, we bought it. Says so right here on the receipt. Copyrights are transferred only by writing and it plainly says we bought it.
[SCO] But we bought it after!
[Judge] You got a receipt?
[SCO] <crickets>
[Judge] Judgement for the defendant. (gavel) Next!
Seven years for eight hours work
on
Novell Wins vs. SCO
·
· Score: 5, Insightful
But at least that part is over. There's still a little cleaning up to do but this one could be over and done with finally this summer. If you like Groklaw, head over and give PJ a pat on the back for her long perseverence.
That paper is from a vendor who's selling a proprietary interconnect and flash storage. Your mileage may vary, but the dolphin interconnects are nice so use them if you like them. If you prefer FC - use FC. OpenFiler supports FC target.
They say BBWC adds up to 70usec of write latency compared to raw disk. That's just not true. If that were the case, why would anybody use the write cache? Naturally I prefer the flash-backed write cache but that's pretty new stuff.
I don't see where it says 2.2ms. The document I downloaded says ~240usec or 0.24ms, not 2.2ms.
On the graph 0.24ms is for a 100KB message size. If you're working with a smaller message size like, say, 4KB, the graph says 0.064ms. 64usec is more than I would like it to be, but given the provenance of the information I'll have to run that test for myself. I imagine if I had to I'd fix that part, but it's not likely to come up.
Slashdot Mirror
What, are they chopped liver already? Have you no respect for the awesomeness that is PocketExcel?
There's lots of fear that Google will turn evil, and lots of folks (however biased) who would say they already have.
I tell Google what I want to know. That I want to know something about a certain topic is a datapoint. They take note of that, and who I am. They use click-through metrics to discover whether I found their results interesting. When I buy a product from an advertiser, they trace it back to my original search and use that to optimize the advertising to me to be the products I'm most interested in. Among other things they use that info to improve their search, to target their advertising, to gauge interest in topics and products across areas, regions, nations and the world.
This is no different than Walmart using their sales register metrics to discover that ice cream sells better after a hurricane.
In practice it's like having your own Concierge striving to anticipate your every desire. Don't ask what's available - say what you want. Anything can be had for ready cash, from a glass of Pinot Grigio to the robes of a senator with the senator still inside. (Apologies to RAH).
Does this aggressive quality of service erode my privacy? Well, yes it would, if I expected privacy on an Internet connection that maps to my home address and billing information. I haven't ever been that dumb. If I want to do something personal and private on the Internet I use my BSD box and my neighbor's open wireless access point to VPN to the offshore RDP account I paid for with eGold that I bought online using a disposable credit card I paid cash for, like anybody else who wants real privacy. You don't get privacy - you take it.
XBox Profits up, sales down. If they get a long tail they may break even - and for Microsoft that's doing well. What does it say for a company that they hope to get out what they put in over a decade, inflation notwithstanding? And what about the failure rate? Is putting out products that fail half the time harming their brand? That's a reasonable expectation. If Boeing planes failed to stay airborne, or Toyota cars failed to operate safely less than 99.99% of the time, we'd have a serious issue with that. Somehow though Microsoft is getting away with quality control that nets 0.5 9's. How is that even possible?
Actually, we don't know what sort of voodoo Steve Jobs used on the media heads to get them to put their content on iTunes under fairly reasonable terms. They were still working PlaysForNow at the time. It must have been some persuasive stuff. I'm thinking Quicktime video of studio executives snorting heroin off the nether parts of recently dead OD'd boyband members, but that may be my imagination getting the best of me. Maybe it was simple consensual human/gerbil interaction. I'm sure it wasn't reason though - they don't understand the term.
This part of the Book of Jobs will probably never be written. Some things are best left a mystery.
That'll build market share and ad revenues. Until you stop paying people to pretend to use your search engine to find stuff to buy. Then they abandon you and you'll find you've flushed a bunch of cash for absolutely nothing.
You're right! People have to be more forward thinking than that. It could take a few more decades at several billion dollars a year for Microsoft's Internet strategy to break even, but then these naysayers will be really sorry. Stick with the company that's not afraid to put in the investment and stick with a long term strategy.
Bing hasn't 'failed'. Not taking the top spot is not 'failure'.
$6B and running to buy 12% market share that will disappear once they stop dumping money in. That's not failure? Then what is?
The iPhone had 18% share of smartphone sales in the fourth quarter. Time to wake up.
Why is noone asking to examine the C programming language, to adapt it to modern programming processes and methods? Why is noone speaking out to defend C as a useful language, to update the aged methods of Kernighan and Ritchie?
Oh, wait. Never mind. That was a stupid question.
Microsoft? Credible? On what planet? They didn't even write the DOS that was their reason for being. IBM helped them out with W32, and suffered from it when they killed OS/2. Dave Cutler took the features they needed from VMS to create NT - and today he's filing the serial numbers off of EC2 to complete their cloud offerring. They have been a sham this whole time and no change from that paradigm is anticipated.
That was a beautiful synopsis. I think I like you.
This reads to me a little bit more like "They promised they'd take care of me for destroying the credibility of ISO. It's been years, and still no check!"
You couldn't watch this train wreck without knowing that Alex Brown had sold his soul. He jammed this thing through, destroying the value of ISO in the process. Now he's learning that if you dance with the devil, you pay his fee. We told him that then to no effect, and telling him again will do no better.
From here no slack for him. He did what he did and not only he but we have to live with it. He'll be a pariah forever.
On the upside, managers everywhere have the ISO standard to cite for their completely proprietary installations of Office and Exchange. Who knows, but maybe soon they'll have a similar standard to cite for IE: "It's standard, in the hypothetical sense that they sell a lot of it and so the standards bodies should write their standards to it."
It doesn't surprise me that a few HP geeks are doing arithmetic analysis of what's supposed to be art. No doubt this model will be used to bang out a bunch of crap optimized for it, which will then disprove the model. It would not surprise me if a niche industry in punitry alts didn't spring up so that people could blog every possible permutation of like/dislike early and so market the winning alts as "market drivers" that could be marketed to studios. Why not? We're already doing that here on slashdot and other tech sites.
These guys should get laid. Then maybe they'd put down the slide rule and enjoy the movie.
How many button skins does it take to make a nice pair of gloves?
Heh. It's my lack of understanding. I just don't "get it". Yeah, that's my problem. If only I understood the ROI proposition properly presented then I would grasp the essential nature of Oracle's value-add.
Bite me. It's tables and joins, SQL and IOPs. Oracle has no magic bits.
People who can be swayed by one persuasive salesman await the next to find their door.
It comes in. It goes out.
Talk about fleeing the cat to find the tiger. Wee todd you might be making this post, and id lacks something.
For GPL works you take a distro, file off the serial numbers somebody else was using (usually this is just trademark logos and such), stamp your name on it, and it's DeadPixelOS. Of course, DeadPixelOS isn't going to get much of a following unless you're continuously developing some value-add as well as keeping up with the patch management of 1000+ packages. It's some work. For your first clone distro I'd start with a low-maintenance one like Pentoo.
The line is drawn in the license. For OpenSolaris, that would be here. If the license says you can do what you want with the code, then you can. If it says you can't, then you can't. And if it doesn't say, then it's all about how good your lawyer is and how much you want it.
Will. Oracle is not in the business of giving stuff away for free.
Have you heard? They license their database software not by the servers it runs on, nor by the processor, but by the core. How absurd is that? Does it cost them more to produce a database that works on more than 4 cores, or to support it? Believe it or not, they also charge extra for installed memory, as if that had anything to do with their production or support costs. Failover? Now you're into serious money. And don't you dare run it on stuff that's not on the secret list, or your support contract is invalid.
If Cisco's motto is "that feature is enabled through the purchase of an optional license", Oracle's is more so.
I guess Oracle doesn't get that we have options, and the pace of hardware technology will quickly erase any software advantage they think they have.
We all know that one major problem with the Microsoft platform is that it's homogeneous. No matter how many times we hear the "ground up" reengineering story, we get these exploits that work vulnerabilities in a common code base. All of the platforms use the same code. All code has bugs, and one bug might grant entry, while two more might grant privilege escalation, and so once an exploit is found all the machines with that code base are pwned. The solution to this problem is deviously simple: do everything differently on every machine. No, I'm not talking about ASLR here, though that's a start.
Stop. I know the first reaction to that is "that's crazy talk". This is pretty revolutionary thinking. It's not possible to design a unique operating system for every user. It is however possible to avoid the complementary vulnerability trampoline by varying the ways that components implement various technologies.
Every action that a machine can perform can be done in various ways - various algorithms can be used to achieve the same result, and some algorithms are more efficient than others. As a part of development many of these ways are explored and until now all but one was discarded. Simply by retaining the discarded algorithms, exploring the variations permissible within the defined interface, and retaining each functional implementation as a heuristic option allows the system designer to thwart the advantage of the large static target. The varying algorithms can be distributed randomly across the installed base as polymorphic patches. As long as the variant algorithms are strictly conformant to the well-defined interfaces, and the interfaces are well designed, it works. The downside to this is that some algorithms are, let's face it - sub-optimal. The diversity of algorithms is an advantage here as a feedback mechanism will reveal optimizations that yield net losses due to secondary effects. This will winnow the dozens of algorithms to a few. Even with only a few performant options per algorithm given the vast number of subsystems in a desktop or server operating system, we'll not run out of permutations before the end of time.
When each subsystem might be any one of several implementations that achieve the same object, the monolithic cathedral of code with a universal backdoor is prevented. Patches can randomly rotate the heuristic until the exploitability of individual platforms is not predictable. Performance of an individual system will vary to a degree, but not necessarily so in net - the distribution of performant vs sub-optimal algorithms can be intelligently distributed so that they average out and one system doesn't have all sub-standard algorithms. Positive feedbacks can indicate exploited components and replace them in an evolutionary fashion before they can be combined synergistically into a chain of exploits that go from basic entry to system privilege. The feedback can also gauge the quality status of the code, and with proper tracking lead back to the outstanding developer for recognition (or the leakmaster for reassignment).
Oh, and no patenting this stuff you bastards! This comment is prior art (ok, I adapted the ideas from some 1980's AI research and Conway's Life - but you can't prove that. Regardless, you didn't invent this stuff and the patents are NOT YOURS).
This is not a greek tragedy - it's a cheesy cliffhanger serial that has the hero facing near certain death at every turn. It won't end until advertisers won't buy time on the reruns.
But at least that part is over. There's still a little cleaning up to do but this one could be over and done with finally this summer. If you like Groklaw, head over and give PJ a pat on the back for her long perseverence.
Congrats to Novell's legal team.
/SCO die,die,die!
A few things: