Most software patents I am aware of are relatively obvious, including RIM and NTP's patents in this case. Prior art is not the only issue, the entire definition of what should be patentable is problematic. Currently any combination of technologies not previously written about that is non-obvious (by who's definition?) is legit. Maybe I should write a patent generator that randomly combines existing technologies.
We need to ask ourselves what the purpose of a patent is, and then make the rules aligned with the goals.
I have not read Ray's book, but the recap I did read made me think it was not worth reading. I don't remember all the details other than I logged him away in my brain as a person that was not making valid claims. I could be wrong.
While I think you're probably right on that point, I also think that's a straw man. If we're capable of building a computer with 1,000 times the human-style intelligence of a biological person, then sure that would be of interest - even if it weren't actually infinitely smarter than we are. Singularity advocates contend that we only need to build a computer smarter than the best computer designers, and then step back to let the positive feedback loop take care of the rest. I truly believe this will happen in my lifetime, and probably before I turn 50.
I agree it is very interesting and exciting to achieve almost any level of success in this field, especially a 1,000-times-smarter-than-humans computer. But my gut tells me that goals/motivation are as important a topic in this brain as the pure processing power. How do you get it to do what you want? I think we will make steady progress, I just think there is far more work than many realize.
Earliest estimates for simulating 1 human brain in real time using a very basic neural network setup is somewhere around the year 2020. This is just based on number of neurons and number of connections. When you throw in the glial cells that were previously thought to be relatively unimportant, despite being a large percentage of the brain, then it's probably further still than 2020.
Additionally, to try to simulate ALL of the chemical and electrical interactions/processes that go on inside the brain (just to be sure there isn't something important in those details), then it gets pushed out even further.
So, 1 human brain is a long way off, and that doesn't even take into account making the thing do something useful, it's just the raw processing power. To achieve AI requires understanding of how the brain works, so push it of even farther (further?).
Finally, intelligence is purely a mapping between specific goals, and the actions required to achieve those goals (problem solving by abstracting prior experience and pattern matching to current conditions). There is a mis-conception that you could have "god-like" intelligence that is all knowing and can even give the "correct" answer always, etc. The answer to any problem is specific to the conditions that surround it. Although I don't have a mathematical proof that says it's impossible to have this perfect "god-like" intelligence, I strongly suspect that it is mathematically provable.
"Is it the speed? A decade ago, we were looking at servers which weren't half as fast as our low end PCs today. If it is speed, do we have some magical cutoff which just keeps moving forward?"
I'm not sure what you consider a server, but even the low to midrange servers I deal with (including the 10 year old ones) blow away any PC in transaction volume and data throughput. You have no idea how much hardware exists in these servers (mainframe, as400 and unix) aside from the CPU to do all of the tasks required to get good performance in a multi-user business application/database environment.
There are differences in the hardware of mainframes, unix/as400 servers, pc servers and pc's. Most differences surround reliability (redundancy etc.) and parallel processing (multiple CPU's, multiple specialized processors controlling IO, etc. etc.) Here are a couple examples:
IO controller cache with error correcting checksum in memory and redundant power supply to ensure zero loss of data short of taking a sledgehammer to the thing.
Mainframe CPU - parity checking with automatic transaction rollback on error detection at the hardware level (on the CPU). This is why banks use mainframes, so they know the transaction completed, or didn't, no in between.
Thumbnails were ruled to be ok in the case of a professional photographer vs some search engine.
This case may be different if the thumbnail itself is sold by the company. The other case ruled that the thumbnail was not a good enough representation to harm the photographer commercially.
As a technical person that has worked on just about half the environments that ever existed, and as a person that currently works in a heterogeneous environment (Sun, AS400, Windows), I can tell you the command line is just about the least effective way to work for MANY activities.
There are times when the command line is preferrable, but this is a very small percentage of the time if the UI is done properly.
With Toshiba, Sony and IBM all putting cell's in various electronic devices (TV's, computers, consoles, etc.) the long term costrs will come down significantly.
Even if expensive at launch, Sony is probably looking at the longer term picture.
Enterprise applications have centralized data and applications (database, stored procs, servlets, etc.). You can't easily escape this fact and thus centralized computing will continue to thrive for the foreseeable future.
The backlash against the mainframe had more to do with the niche of applications that did not require the data center. Word processing, spreadsheets, etc. where the data can be local for X amount of time before being shared with others in the enterprise.
Internet applications are the current direction for a few reasons:
1) Centralized apps are still a must for multi-user, shared transaction processing.
2) Maintaining client applications on local computing resources is a logistical problem with zero gain. It's a drain of time and money.
3) The browser is the standardized client that is ubiquitous, no need to worry about whether the client is loaded with 3270, 5250, telnet, etc.
Something to keep in mind on these security ratings is that there are certain features required such that, if missing, the OS cannot be certified at that level, regardless of how secure the OS actually is.
For example, I believe tagging objects as something along the lines of "Secret" and "Top Secret" (seriously) is a requirement for some level (C2?).
As a designer/developer/implementor/trainer that has worked on large enterprise systems most of my career, I am surprised by your statement that anexperienced data entry clerk should enter someone else's data.
The person that owns the data/transaction and is responsible for that data/transaction should be the person to enter it. That is the person closest to it and it reduces the amount of paper transactions before the data actually gets entered.
The problem, in this case is most likely (based on the available information) a complex set of systems that have some sort of interface. Even though the problem was caught quickly in one system (probably due to reasonable auditing procedures), the data had already been replicated to another system (either in summary form or detail) at which point those same audits were probably no longer applied because the designers did not consider the situation.
Sun's new processor is designed for many-connection business server applications. Web stuff.
The Cell is designed for image processing and other high-volume number crunching.
The design decisions both companies made were heavily influenced by their target markets for these specific processors, and those target markets are very different.
Itanium may perform better for some number crunching apps, but not enough to outweigh the costs, generally.
The cell processor, on the other hand offers such a giant increase in performance (for some applications) that you will see people investing time and money to take advantage of it. In addition, with Toshiba, Sony and IBM all with product plans and thus the related volume and eco-system surrounding development tools, etc., I think the cell is positioned far better than Itanium to succeed.
I disgree with your assesment, substantial proposal have been made numerous times, see the end of my post.
Also, with respect to the author of this proposal, amount of time spent in the public sphere with respect to any topic does not automatically make said persons proposal any better than anyone else's opinion. As a matter of fact, I have routinely seen "experts" on significant computer issues with less of an understanding than other people you have never heard of. Maybe this person has good ideas, and maybe this one is a good idea when viewed from specific perspectives (large corporations, USPTO, legal, who knows?), but viewed from the perspective of people/businesses that develop software for a living, it's a poor proposal.
Dropping non-obvious and retaining novelty as key criteria will mathematically increase the number of trivial patents that can be legally obtained (e.g. Amazon 1 click). Furthermore it can be easily argued that with more of these patents legally obtained, significant resources will be required to work around these trivial patents (e.g. Barnes and Noble and Amazon 1 click).
It hardly takes an expert to see the negatives of this particular proposal.
The best solution: Revert back to NO SOFTWARE PATENTS, we used to do it this way.
Both RIM and NTP have patents that are trivial at best. There is absolutely no comparison to the others you listed (I realize you did qualify your statement, and I get the point, but this crap bugs me enough I had to say something).
Slashdot Mirror
"...and case sensitivity makes code easier to read without being a burden on the coder."
I think you have that backwards. Case sensitive languages cause additional compile errors when the case does not match.
You can still use case to make a program easy to read whether it is sensitive to the case or not.
Most software patents I am aware of are relatively obvious, including RIM and NTP's patents in this case. Prior art is not the only issue, the entire definition of what should be patentable is problematic. Currently any combination of technologies not previously written about that is non-obvious (by who's definition?) is legit. Maybe I should write a patent generator that randomly combines existing technologies.
We need to ask ourselves what the purpose of a patent is, and then make the rules aligned with the goals.
...Ray Kurzweil makes a good case...
I have not read Ray's book, but the recap I did read made me think it was not worth reading. I don't remember all the details other than I logged him away in my brain as a person that was not making valid claims. I could be wrong.
While I think you're probably right on that point, I also think that's a straw man. If we're capable of building a computer with 1,000 times the human-style intelligence of a biological person, then sure that would be of interest - even if it weren't actually infinitely smarter than we are. Singularity advocates contend that we only need to build a computer smarter than the best computer designers, and then step back to let the positive feedback loop take care of the rest. I truly believe this will happen in my lifetime, and probably before I turn 50.
I agree it is very interesting and exciting to achieve almost any level of success in this field, especially a 1,000-times-smarter-than-humans computer. But my gut tells me that goals/motivation are as important a topic in this brain as the pure processing power. How do you get it to do what you want? I think we will make steady progress, I just think there is far more work than many realize.
Earliest estimates for simulating 1 human brain in real time using a very basic neural network setup is somewhere around the year 2020. This is just based on number of neurons and number of connections. When you throw in the glial cells that were previously thought to be relatively unimportant, despite being a large percentage of the brain, then it's probably further still than 2020.
Additionally, to try to simulate ALL of the chemical and electrical interactions/processes that go on inside the brain (just to be sure there isn't something important in those details), then it gets pushed out even further.
So, 1 human brain is a long way off, and that doesn't even take into account making the thing do something useful, it's just the raw processing power. To achieve AI requires understanding of how the brain works, so push it of even farther (further?).
Finally, intelligence is purely a mapping between specific goals, and the actions required to achieve those goals (problem solving by abstracting prior experience and pattern matching to current conditions). There is a mis-conception that you could have "god-like" intelligence that is all knowing and can even give the "correct" answer always, etc. The answer to any problem is specific to the conditions that surround it. Although I don't have a mathematical proof that says it's impossible to have this perfect "god-like" intelligence, I strongly suspect that it is mathematically provable.
And how did the parent get a score of 2?
Unless, of course, you truly are "decompiling" in which case your credibility is gained back.
"Is it the speed? A decade ago, we were looking at servers which weren't half as fast as our low end PCs today. If it is speed, do we have some magical cutoff which just keeps moving forward?"
I'm not sure what you consider a server, but even the low to midrange servers I deal with (including the 10 year old ones) blow away any PC in transaction volume and data throughput. You have no idea how much hardware exists in these servers (mainframe, as400 and unix) aside from the CPU to do all of the tasks required to get good performance in a multi-user business application/database environment.
There are differences in the hardware of mainframes, unix/as400 servers, pc servers and pc's. Most differences surround reliability (redundancy etc.) and parallel processing (multiple CPU's, multiple specialized processors controlling IO, etc. etc.) Here are a couple examples:
IO controller cache with error correcting checksum in memory and redundant power supply to ensure zero loss of data short of taking a sledgehammer to the thing.
Mainframe CPU - parity checking with automatic transaction rollback on error detection at the hardware level (on the CPU). This is why banks use mainframes, so they know the transaction completed, or didn't, no in between.
"it probably would have just bankrupted IBM"
Nope.
This was during a time when IBM was still rolling in the money from mainframe and midrange sales, the PC was a small part of their business.
Thumbnails were ruled to be ok in the case of a professional photographer vs some search engine.
This case may be different if the thumbnail itself is sold by the company. The other case ruled that the thumbnail was not a good enough representation to harm the photographer commercially.
I am constantly stunned by statements like that.
As a technical person that has worked on just about half the environments that ever existed, and as a person that currently works in a heterogeneous environment (Sun, AS400, Windows), I can tell you the command line is just about the least effective way to work for MANY activities.
There are times when the command line is preferrable, but this is a very small percentage of the time if the UI is done properly.
Everything I find on the Internet sounds positive on cell mfg., including systems shipping by Mercury systems in 2q and IBM in 3q.
Do you have any evidence to backup this claim?
With Toshiba, Sony and IBM all putting cell's in various electronic devices (TV's, computers, consoles, etc.) the long term costrs will come down significantly.
Even if expensive at launch, Sony is probably looking at the longer term picture.
Fortunately nobody really cares what you think is right or wrong.
He asked a question and those that care to help responded.
Very good!
Enterprise applications have centralized data and applications (database, stored procs, servlets, etc.). You can't easily escape this fact and thus centralized computing will continue to thrive for the foreseeable future.
The backlash against the mainframe had more to do with the niche of applications that did not require the data center. Word processing, spreadsheets, etc. where the data can be local for X amount of time before being shared with others in the enterprise.
Internet applications are the current direction for a few reasons:
1) Centralized apps are still a must for multi-user, shared transaction processing.
2) Maintaining client applications on local computing resources is a logistical problem with zero gain. It's a drain of time and money.
3) The browser is the standardized client that is ubiquitous, no need to worry about whether the client is loaded with 3270, 5250, telnet, etc.
Something to keep in mind on these security ratings is that there are certain features required such that, if missing, the OS cannot be certified at that level, regardless of how secure the OS actually is.
For example, I believe tagging objects as something along the lines of "Secret" and "Top Secret" (seriously) is a requirement for some level (C2?).
AIX ran on mainframe from 1990 to 1993.
Linux IS an OS and thus AIX does not run ON Linux.
These inaccuracies bring your post into question.
As a designer/developer/implementor/trainer that has worked on large enterprise systems most of my career, I am surprised by your statement that anexperienced data entry clerk should enter someone else's data.
The person that owns the data/transaction and is responsible for that data/transaction should be the person to enter it. That is the person closest to it and it reduces the amount of paper transactions before the data actually gets entered.
The problem, in this case is most likely (based on the available information) a complex set of systems that have some sort of interface. Even though the problem was caught quickly in one system (probably due to reasonable auditing procedures), the data had already been replicated to another system (either in summary form or detail) at which point those same audits were probably no longer applied because the designers did not consider the situation.
To the programmer, communicating with the SPU is abstracted to file i/o operations. Go check out IBM developerworks pages for lots of info.
Sun's new processor is designed for many-connection business server applications. Web stuff.
The Cell is designed for image processing and other high-volume number crunching.
The design decisions both companies made were heavily influenced by their target markets for these specific processors, and those target markets are very different.
These are apples and oranges.
Itanium may perform better for some number crunching apps, but not enough to outweigh the costs, generally.
The cell processor, on the other hand offers such a giant increase in performance (for some applications) that you will see people investing time and money to take advantage of it. In addition, with Toshiba, Sony and IBM all with product plans and thus the related volume and eco-system surrounding development tools, etc., I think the cell is positioned far better than Itanium to succeed.
"unless it runs 0x86 it's never going to show up in home or bussiness computers"
x86 certainly owns the desktop, including inside the enterprise, but Windows (and x86) has made very little progress in the data center.
The data center is owned by Mainframe, Power (Unix/Linux/AS400), Sparc(Solaris) and then Unix on x86.
I disgree with your assesment, substantial proposal have been made numerous times, see the end of my post.
Also, with respect to the author of this proposal, amount of time spent in the public sphere with respect to any topic does not automatically make said persons proposal any better than anyone else's opinion. As a matter of fact, I have routinely seen "experts" on significant computer issues with less of an understanding than other people you have never heard of. Maybe this person has good ideas, and maybe this one is a good idea when viewed from specific perspectives (large corporations, USPTO, legal, who knows?), but viewed from the perspective of people/businesses that develop software for a living, it's a poor proposal.
Dropping non-obvious and retaining novelty as key criteria will mathematically increase the number of trivial patents that can be legally obtained (e.g. Amazon 1 click). Furthermore it can be easily argued that with more of these patents legally obtained, significant resources will be required to work around these trivial patents (e.g. Barnes and Noble and Amazon 1 click).
It hardly takes an expert to see the negatives of this particular proposal.
The best solution: Revert back to NO SOFTWARE PATENTS, we used to do it this way.
No.
Both RIM and NTP have patents that are trivial at best. There is absolutely no comparison to the others you listed (I realize you did qualify your statement, and I get the point, but this crap bugs me enough I had to say something).