No. A company is considered the author for copyright purposes for a "work for hire", but an inventor is always a person. Show me a patent where the inventor is a company. For example, find a patent where IBM ("International Business Machines") is the inventor. If you search current applications for that, you'll find over 4000 with IBM as the assignee, and none with IBM as the inventor. If you search all issued patents (since 1976), you'll find over 33000, but not one where they are listed as the inventor.
A company can be an assignee, but can not be the inventor. Only the inventor(s) can apply for a patent. See who can apply for a patent, straight from the horse's mouth.
No, if you had to hit each button (and each button would be a different domain, not necessarily a different DNS), it wouldn't infringe. If, instead, you entered a command-line to a script such as:
finddns tricorn org com net edu biz tv com.uk
which simply did a nslookup of tricorn.org, tricorn.com, tricorn.net tricorn.edu tricorn.bix tricorn.tv tricorn.com.uk, trapped the results, and then presented a list of which domains were present (or which were not), that would infringe on this patent.
The hypertext, searching multiple servers at the same time by spawning multiple processes, etc. are all dependent claims. You don't have to do those things to infringe on the patent, just the independent claims (the ones that don't start with something like "the device in claim 1 where...").
"Obvious", in legal terms, is a very non-obvious concept. I think it's been distorted beyond reason. Prior art is often ruled to not be EXACTLY the same thing, and thus is discounted (even though any reasonable person would say that the patent was an obvious extension of quite a bit of existing practice). Different techniques are often ruled to be the same. I've seen at least one case where the prior art was ruled to not apply to the patent, even though if it hadn't been prior art it would have infringed the patent. The recent (couple years) change to the way the "Doctrine of Equivalents" or whatever is interpreted helped some with some of those types of cases, but didn't eliminate it. Ultimately, it comes down to lawyers, and money, and even a bad patent can be used to extort money out of a company, since paying can be cheaper than winning in court.
Oh, and yes, I know that in my earlier response to the parent, I referenced Verizon instead of Verisign. Ooops.
I wish the original DVD format had included a P&S data stream, which would simply give the coordinates of a 4:3 box out of the original aspect-ratio source (could even have multiple ones - possibly a 16:9 box for the 2.35:1 films). Then ALL films could be P&S without taking up any significant space on the disk, and there'd never be a reason to release something only in P&S because of some belief that's that the market wants.
No, "clean room reverse engineering" is a way to avoid copyright infringement. It doesn't do a thing for patented techniques. Patents (are supposed to) disclose exactly how they work, and you have to come up with a different way of doing it if you don't want to license or infringe it.
Your description of a patent is what it is supposed to be, but with patents like this, they tend to be much broader than that. Some patents issued have been like "producing light with electricity" - I think we'd all agree that using a carbon filament with electric current running over it, in an evacuated globe, is (or was at the time) non-obvious, novel and useful; however, that shouldn't apply (and wasn't, as far as I know) to the basic idea of producing light using electricity.
This particular patent covers ANY technique for searching multiple domains for a domain name (or, more to the point, lack of such a domain name in multiple domains). It mentions several specific techniques (spawning multiple processes to speed up the search by searching multiple domains in parallel), it mentions formatting in "a hypertext markup language" (note, NOT just HTML), but those aren't the primary claims. That just keeps anyone else from patenting those particular techniques (which, if they did, would block Verizon from using those techniques unless they licensed it - but the other person would not be able to use it either unless they licensed this patent!). For example, you might be able to patent doing multiple DNS searches using a quantum computer, and then Verizon would have to license that patent from you if they wanted to use quantum computers (I hereby dedicate this totally non-obvious, novel and useful idea to the public domain).
There are plenty of software patents that are as broad as "a 1st person 3-d shooter game" (both single- and multi-player, networked on a LAN, networked on a WAN, using client-server and peer-to-peer methods - each one of those, or various combinations, might be separate patents). And even with the existence of such patents, someone would probably still be able to get a patent where all the users are on a wireless network. AND THE PATENT OFFICE STILL WOULDN"T GET THAT THOSE ARE ALL OBVIOUS WAYS TO DO THINGS.
Patents are supposed to require inventiveness, not "mere engineering". You shouldn't be able to patent WHAT, only HOW, but unfortunately many software patents are for WHAT...
Unless things have changed recently, only the actual inventor(s) can file a patent. They may be under contractual obligation to assign it to the company they work for, but the patent goes to the individual(s).
You do realize that in 10-15 years, the kids of today will be looking back, complaining about how all the crap on TV, or the Internet, or the holo-tube or whatever, is just terrible compared to the good stuff that was playing in the 2000's; and the stuff you like they won't have heard of; and in about 25-30 years, a few of the things you like will be known as "those classic old cartoons".
I'm pre-Sesame Street, pre-Electric Company, pre-TMNT, pre-Transformers. The cartoons I remember with fondness are Johnny Quest, Space Ghost, Rocky & Bullwinkle, all the silly shorts (Bugs Bunny, Road Runner, etc.), Flintstones, Jetsons, Yogi Bear. You know, those "classic old cartoons".
Don't forget the speed advantages in a system (not CPU, which is only part of a system) of vast amounts of much faster memory, both volatile and persistent. Twenty year ago, the microcomputer world was in the process of moving from 64K of RAM and 160-320K of slow floppy disk storage, to 512KB-1MB of RAM and 5-10MB hard drives. It is amazing to me that now I can read up the contents of my entire first hard disk drive into a small portion of RAM in 1-2 seconds, the entire amount of RAM on that machine was less than 1% of what I have now, and I can emulate it at least 50 times faster than it actually ran, and that's on a 2-year-old laptop. In 20 years, will we have computers where 1000GB of RAM is normal, I can read persistent storage at a rate of 50-100GB/sec, and I can emulate a current CPU as if it ran at 50GHz? You'll be able to do real-time stereo-vision HD rendering of things that currenly take several minutes per frame. Let's not forget communications, either - 300-1200bps was the norm, 9600bps was about the best you could do over the phone lines.
Computer systems are a lot more complicated today than they were 20 years ago. However, a lot of that complexity is arguably necessary to support the increased capabilities (that were simply not possible with the hardware limitations of 20 years ago compared to today). On the other hand, a lot of the complexity, bloat and poor performance is attributable to poor development models. It may make economic sense for a development group to use tools and methodologies that produce bloated slow buggy complex programs, but at least some of that is due to the end consumer not knowing how much it is costing them, or at any rate not being able to do anything about it.
You're describing the Achievement Tests, not the SAT. The Math tests in particular seemed quite inflated in score - for whatever reason, probably because few people took it unless they expected to do well, an 800 in Math I was only something like a 90th percentile (and I think Math II even worse).
You're confusing the scaled score with something real. An 800 score doesn't indicate you got every question right, just as a 200 doesn't indicate you got every question wrong. Read the article for lots of details on how scoring is done (one interesting point he illustrates: not answering any questions at all would have gotten him a 430 or so, not a 400). Depending on which test (and what year, of course), they stated that there was no statistical difference between an 800 and (say) a 750 score. Now, if you took the test multiple times, you could probably show a difference between someone who consistently scored 790-800 and someone who consistently scored 720-760, but it wouldn't be much of a difference. Someone who consistently scored 800 would indicate they were actually off the scale. Note also that one of the things you're measuring at that level is accuracy, not necessarily how well you understand Math.
You're right that the SAT doesn't have very good discrimination at the high or low ends. It isn't really designed to do that. Nevertheless, someone who scores well (say, 1200 or better) is, in my experience, quite different from someone who does poorly (800 or less). There are exceptions, of course. Some people take tests poorly, and their score would not be indicative of their actual ability.
At the same time, a good score doesn't mean you'll be a success. The test doesn't measure motivation and attitude, and doesn't measure a lot of aptitudes that can lead to success. A low score doesn't mean someone is worthless. It's a correlation with certain types of intelligences, but it doesn't say anything about what it isn't measuring.
It is, however, measuring much more than ability to take tests.
if an American voluntarily and with the intention of relinquishing U.S. citizenship enters or serves in foreign armed forces engaged in hostilities against the United States or serves in the armed forces of any foreign country as a commissioned or non-commissioned officer.
(emphasis mine). That's an `or'. Plain and simple. In other words, serving in a foreign armed force which is engaged in hostilities is grounds for loss of nationality.
The only reasonable parsing of that phrase is:
... if ((an American (voluntarily and (with the intention of relinquishing U.S. citizenship)))(((enters or serves) in (foreign armed forces engaged in hostilities against the United States)) or (serves in (the armed forces of any foreign country) as a (comissioned or non-commissioned) officer)))."
Aren't stem cells the cells that create the fetus and then disappear in the first year of life?
Your understanding is completely wrong. There are many kinds of differentiated stem cells in the human body. One trick that researchers would love to figure out is how to take one of those and turn it into an undifferentiated stem cell, which would be similar to or the same as a fetal stem cell.
My understanding of the current ban on federal funding is that it only applies to fetal stem cells, with an exception for a set of already established cell lines. Has that changed while I haven't been looking? I hadn't heard that there were any bans on products, domestic or imported, just on using federal funds to do research.
Why should it be the DSL provider leasing the line to my house? It seems to me that I, the customer, am the one leasing the line - I just want the damned phone company to let ME run over a DSL connection that they forward the the ISP of my choice - the phone company should charge ME the costs for running the equipment up to the ISP, the ISP charges ME additional amounts for their equipment and bandwidth to the Internet, and we're all happy, at least as long as the phone company charges me the same rate as they would if I was connecting to THEM as an ISP.
I just checked with my ISP to see if they could do DSL yet; they said they COULD, but it would be about 20% higher than what SBC would charge me, because SBC bills them at a higher rate than they'd charge me. This should be illegal. If SBC can afford to not only do the DSL transport, but connect me to the Internet, provide e-mail, DNS, Web services, and do customer support on all of that, for LESS than they'd charge to just do the transport to another ISP, then either they're using their monopoly to subsidize their ISP part, or they're overcharging my ISP.
if an ISP blocks inbound 25, they immediately prevent any of their systems from being used as open relays. The open relays on their network will not move around.
Blocking outbound will also prevent any open relays on their network from going off their network; it is then an internal affair which they can clean up, without getting onto blackhole lists or bothering anyone else.
This is an attempt to move IP itself to become hierarchical, to depend upon a single DNS system, and to have "special" hosts.
DNS is inherently hierarchical. e-mail is close to (and for all practical purposes) totally dependent on DNS anyway. MX records already indicate "special" hosts. This is just a way for a peer to only accept e-mail from a host that will accept a response for the mail being sent. It is still peer-to-peer. Unless you're going to advocate only using filtering methods, there has to be SOME way of distinguishing between "legitimate" mail servers and throw-away dialup accounts.
Note that what I'm suggesting is more to move mail clients to use the MSA port (587) instead of the SMTP port, and get everyone to send e-mail by routing it through the "proper" mail server, so that accountability and authentication can be maintained.
I don't accept a solution of only accepting mail from people I have authorized. I don't have a master list of everyone I'd accept mail from. Doing it by having a list of certifying authorities that I'll accept regardless is even more restrictive than what I'm suggesting.
Note that blocking port 25 is only a temporary measure, and that only because the current behavior of all the mail servers using port 25 can't be changed quickly. First everyone sending mail directly, or through the "wrong" mail server, have to be persuaded to start doing it the "right" way. My suggestion is that blocking port 25, along with universally supporting port 587, will get that to happen; once it does, servers that block inbound mail that isn't from the "right" server can do so without risking bouncing a significant amount of legitimate mail. Once enough servers are rejecting based on MX records, the block on port 25 can be removed. Then anyone who complains about receiving SPAM from your network, that didn't go through your mail server, can be simply told to fix their mail server.
My ISP already does a reverse lookup on the sending mail server, and on the return and envelope address, and bounces it if any of them don't resolve. I get maybe 2 or 3 SPAMs a week. Those few that get through would have been rejected if the return or envelope address didn't match the server. Unfortunately, currently, quite a few of the legitimate messages also wouldn't.
Check the sendmail AUTH info, it discusses sendmail connecting as a client. There's also some info on STARTTLS that describes how to authenticate using TLS, but that would require more than just a password configuration on the other end (you'd need to get the remote sendmail to recognize your certificate as being allowed; I don't think there's any way to specify it on a per-user basis).
simply means that there is now another port (say, 5305) to block
No, SMTP still runs on port 25, that is too difficult to change in one step; the port you'd actually use (say, 587) doesn't need to be blocked because you require authentication on that.
You can't require authentication on the MTA port because none of the servers out there will be able to authenticate. You can require authentication to relay, but that doesn't block mail directed into your domain. Blocking OUTbound port 25 allows an ISP to control SPAM coming from their ports, without blocking people from using any MSA they want. Not that I like the idea of blocking at all, but the only way to authenticate an SMTP server as being a legitimate MTA is by checking that the return address matches the server it is coming from, and that simply can't be done until pretty much everyone sends mail by routing it through the correct mail server. One way of getting that to happen is to encourage ISPs to block port 25 (along with making sure to enable an authenticating MSA on port 587). Eventually, the block on port 25 can be dropped, as anyone trying to send mail by connecting directly to an SMTP server will be rejected since the MX records don't match.
For a case like this, you'd need to have a relay program that forwarded to, and authenticated with, the correct mail server for the queued messages; you'd need to configure it with authentication info for each address, or else store authentication info (e.g. certificates for use with STARTTLS, or a password for use with AUTH) with each message. Presumably you'd batch things so all messages from the same address would only need one authentication round. The sendmail msp feature is pretty close to this, e.g. it can use the MSA port, by using
FEATURE(`msp', `', `MSA')
but I don't think it can dynamically change which host it relays to.
Why not use port 587, as specified in RFC2476; sendmail supports that by default. Any e-mail client that supports authentication should be able to specify the port as well.
Although I do shudder at the idea of an ISP filtering based on port numbers, I believe that filtering OUTbound connections to port 25 would be just barely acceptable, and only because it is so universally used with a protocol so open to abuse. If you really truly want to run a sendmail that doesn't just relay to the ISP, then discuss it with the ISP. You can still receive on port 25 (unless the ISP also wants to restrict servers, but that's a different issue), and that's what most people would want to run their own sendmail for anyway.
Mail clients using the MSA port, combined with MTA's rejecting messages that aren't coming from the server identified by an MX record of the return address, combined with client programs that connected to the correct outbound mail server based on which return e-mail address you're using, would increase accountability. A legitimate server will have limitations on SPAM. A domain that adds bogus MX records could have return addresses black-listed (no more spoofing innocent domains). This would eventually make it safe to eliminate the port 25 filter. Ideally, even internal submissions would require the use of the MSA port, so that becomes the default for all mail clients, and you don't have to worry about switching configurations depending on what network you're connecting through.
The article doesn't say that the pixels can hold the image, only that it uses less power to keep the pixel at the same level. It sounds like you'd still need backing store, and doesn't necessarily mean you can access an individual pixel to change it.
Plasma panels were initially invented to avoid backing store, when memory was very expensive. They replaced storage display tubes for the PLATO project in the early '70s. In that mode, they're monochrome (for PLATO terminals, orange/black); the whole point was that you could turn an individual pixel on or off and it would stay that way.
Did you get explicit permission from the owners of Slashdot before posting (or even reading the article and responses)? I hope it wasn't by e-mail, unless you also received explicit permission to access their e-mail server as well.
How is a well-known protocol, with explicit provision for asking for and providing authorization, any different? Is not a machine exchange of "may I access this network"... "yes"... explicit permission, just as sending a HELO command to an SMTP server, and getting back a "200 Hello, pleased to meet you" response, indicates you can attempt to send mail?
Actually, I think dynamic limits could be great if they were done properly. By that I mean you shouldn't hit a brick wall, your usage should be metered down gradually over time as you use more average bandwidth, and metered back up as you stop using it. Could even be tied in with peak use times. Note that I'm talking about fairly short time periods - you should be able to get a burst for a short while (say, a minute or 5 minutes or whatever, depending on what rate you're talking about), then gradually reduce it as long as you're peaking, and gradually raise it when you're not using much. If you're completely idle for a while, you should have a clean slate.
I've done simulations of various algorithms for doing similar things for metering CPU usage on a timesharing system, and it can work. Note that another possibility is to enable the QoS bits in IP to allow you to lower priority in exchange for higher bandwidth; there's always been a problem with determing how to bill for QoS, perhaps just automatically setting rate limits based on it would work better.
If that's compatible, then S100 is compatible with VMEbus is compatible wiht NuBus is compatible with ISA/EISA/PCI. Just because IBM never licensed their MCA patents to anyone to make an ISA/MCA bridge doesn't mean it couldn't have been done. IBM was dumb in the way they handled MCA, but that doesn't make PCI "compatible" with ISA.
I know the 6502 had a limited set of registers. That was my point. It simplifies the architecture immensely, which means you can do load/operate/store very quickly. If you're going to have a limited number of registers, might as well go all the way. Besides, the 6502 effectively had 256 registers in low memory; in a modern version of the 6502, those locations would certainly be implemented on-chip.
The only reason IDE is cheap is because everyone uses it, because it is cheap. Thus, we're stuck with it. FireWire is one possible way out of it.
As for IRQs, I wasn't suggesting that computers shouldn't have interrupts, but that the PC architecture for handling them is lousy. PCI indeed handles things much better - but the PC is still saddled with the old architecture. In the meantime, the Mac had NuBus ages ago, and never had any such problems.
As for serial ports, again I wasn't saying that serial ports are bad, I'm saying that PC serial ports are lousy. Taking the Mac again as an example, they had great serial ports, could do RS-422 as well as pseudo-RS-232, could handle asynch and synchrounous, had a larger internal buffer than standard PC serial chips. Serial connections still have their uses, but for most home use, they are utterly useless now.
the shift away from ISA wouldn't successfully take place until many years later when they introduced the PCI bus, which maintained backwards compatibility with ISA. PCI was successful because it allowed a gradual transition away from ISA
PCI is not compatible in any sense with ISA. I think you're thinking of EISA, which was the industry response to IBM's attempt to corner the market by patenting various aspects of MCA so that no one else could make compatible devices or systems.
Another thing he got wrong was that the 8086 chip was not backwards compatible with the 8080 line. It was similar in architecture (limited non-orthogonal register set, awkward instruction set), and there were 8080 -> 8086 cross-assemblers (sometimes producing more than one 8086 instruction for each 8080 instruction), but it wasn't backwards compatible in the same way as 8086 -> 80186/286/386/486/Pentium were.
Wow, a whole 16 general-purpose registers, my heart flutters. Bah, might as well use a 64-bit extension to the 6502.
Slashdot Mirror
No. A company is considered the author for copyright purposes for a "work for hire", but an inventor is always a person. Show me a patent where the inventor is a company. For example, find a patent where IBM ("International Business Machines") is the inventor. If you search current applications for that, you'll find over 4000 with IBM as the assignee, and none with IBM as the inventor. If you search all issued patents (since 1976), you'll find over 33000, but not one where they are listed as the inventor.
A company can be an assignee, but can not be the inventor. Only the inventor(s) can apply for a patent. See who can apply for a patent, straight from the horse's mouth.
No, if you had to hit each button (and each button would be a different domain, not necessarily a different DNS), it wouldn't infringe. If, instead, you entered a command-line to a script such as:
which simply did a nslookup of tricorn.org, tricorn.com, tricorn.net tricorn.edu tricorn.bix tricorn.tv tricorn.com.uk, trapped the results, and then presented a list of which domains were present (or which were not), that would infringe on this patent.The hypertext, searching multiple servers at the same time by spawning multiple processes, etc. are all dependent claims. You don't have to do those things to infringe on the patent, just the independent claims (the ones that don't start with something like "the device in claim 1 where ...").
"Obvious", in legal terms, is a very non-obvious concept. I think it's been distorted beyond reason. Prior art is often ruled to not be EXACTLY the same thing, and thus is discounted (even though any reasonable person would say that the patent was an obvious extension of quite a bit of existing practice). Different techniques are often ruled to be the same. I've seen at least one case where the prior art was ruled to not apply to the patent, even though if it hadn't been prior art it would have infringed the patent. The recent (couple years) change to the way the "Doctrine of Equivalents" or whatever is interpreted helped some with some of those types of cases, but didn't eliminate it. Ultimately, it comes down to lawyers, and money, and even a bad patent can be used to extort money out of a company, since paying can be cheaper than winning in court.
Oh, and yes, I know that in my earlier response to the parent, I referenced Verizon instead of Verisign. Ooops.
I wish the original DVD format had included a P&S data stream, which would simply give the coordinates of a 4:3 box out of the original aspect-ratio source (could even have multiple ones - possibly a 16:9 box for the 2.35:1 films). Then ALL films could be P&S without taking up any significant space on the disk, and there'd never be a reason to release something only in P&S because of some belief that's that the market wants.
Actually, Dr. Strangelove is "multi-aspect-ratio" (it says so right there on the DVD box).
No, "clean room reverse engineering" is a way to avoid copyright infringement. It doesn't do a thing for patented techniques. Patents (are supposed to) disclose exactly how they work, and you have to come up with a different way of doing it if you don't want to license or infringe it.
Your description of a patent is what it is supposed to be, but with patents like this, they tend to be much broader than that. Some patents issued have been like "producing light with electricity" - I think we'd all agree that using a carbon filament with electric current running over it, in an evacuated globe, is (or was at the time) non-obvious, novel and useful; however, that shouldn't apply (and wasn't, as far as I know) to the basic idea of producing light using electricity.
This particular patent covers ANY technique for searching multiple domains for a domain name (or, more to the point, lack of such a domain name in multiple domains). It mentions several specific techniques (spawning multiple processes to speed up the search by searching multiple domains in parallel), it mentions formatting in "a hypertext markup language" (note, NOT just HTML), but those aren't the primary claims. That just keeps anyone else from patenting those particular techniques (which, if they did, would block Verizon from using those techniques unless they licensed it - but the other person would not be able to use it either unless they licensed this patent!). For example, you might be able to patent doing multiple DNS searches using a quantum computer, and then Verizon would have to license that patent from you if they wanted to use quantum computers (I hereby dedicate this totally non-obvious, novel and useful idea to the public domain).
There are plenty of software patents that are as broad as "a 1st person 3-d shooter game" (both single- and multi-player, networked on a LAN, networked on a WAN, using client-server and peer-to-peer methods - each one of those, or various combinations, might be separate patents). And even with the existence of such patents, someone would probably still be able to get a patent where all the users are on a wireless network. AND THE PATENT OFFICE STILL WOULDN"T GET THAT THOSE ARE ALL OBVIOUS WAYS TO DO THINGS.
Patents are supposed to require inventiveness, not "mere engineering". You shouldn't be able to patent WHAT, only HOW, but unfortunately many software patents are for WHAT...
Unless things have changed recently, only the actual inventor(s) can file a patent. They may be under contractual obligation to assign it to the company they work for, but the patent goes to the individual(s).
You do realize that in 10-15 years, the kids of today will be looking back, complaining about how all the crap on TV, or the Internet, or the holo-tube or whatever, is just terrible compared to the good stuff that was playing in the 2000's; and the stuff you like they won't have heard of; and in about 25-30 years, a few of the things you like will be known as "those classic old cartoons".
I'm pre-Sesame Street, pre-Electric Company, pre-TMNT, pre-Transformers. The cartoons I remember with fondness are Johnny Quest, Space Ghost, Rocky & Bullwinkle, all the silly shorts (Bugs Bunny, Road Runner, etc.), Flintstones, Jetsons, Yogi Bear. You know, those "classic old cartoons".
Don't forget the speed advantages in a system (not CPU, which is only part of a system) of vast amounts of much faster memory, both volatile and persistent. Twenty year ago, the microcomputer world was in the process of moving from 64K of RAM and 160-320K of slow floppy disk storage, to 512KB-1MB of RAM and 5-10MB hard drives. It is amazing to me that now I can read up the contents of my entire first hard disk drive into a small portion of RAM in 1-2 seconds, the entire amount of RAM on that machine was less than 1% of what I have now, and I can emulate it at least 50 times faster than it actually ran, and that's on a 2-year-old laptop. In 20 years, will we have computers where 1000GB of RAM is normal, I can read persistent storage at a rate of 50-100GB/sec, and I can emulate a current CPU as if it ran at 50GHz? You'll be able to do real-time stereo-vision HD rendering of things that currenly take several minutes per frame. Let's not forget communications, either - 300-1200bps was the norm, 9600bps was about the best you could do over the phone lines.
Computer systems are a lot more complicated today than they were 20 years ago. However, a lot of that complexity is arguably necessary to support the increased capabilities (that were simply not possible with the hardware limitations of 20 years ago compared to today). On the other hand, a lot of the complexity, bloat and poor performance is attributable to poor development models. It may make economic sense for a development group to use tools and methodologies that produce bloated slow buggy complex programs, but at least some of that is due to the end consumer not knowing how much it is costing them, or at any rate not being able to do anything about it.
You're describing the Achievement Tests, not the SAT. The Math tests in particular seemed quite inflated in score - for whatever reason, probably because few people took it unless they expected to do well, an 800 in Math I was only something like a 90th percentile (and I think Math II even worse).
You're confusing the scaled score with something real. An 800 score doesn't indicate you got every question right, just as a 200 doesn't indicate you got every question wrong. Read the article for lots of details on how scoring is done (one interesting point he illustrates: not answering any questions at all would have gotten him a 430 or so, not a 400). Depending on which test (and what year, of course), they stated that there was no statistical difference between an 800 and (say) a 750 score. Now, if you took the test multiple times, you could probably show a difference between someone who consistently scored 790-800 and someone who consistently scored 720-760, but it wouldn't be much of a difference. Someone who consistently scored 800 would indicate they were actually off the scale. Note also that one of the things you're measuring at that level is accuracy, not necessarily how well you understand Math.
You're right that the SAT doesn't have very good discrimination at the high or low ends. It isn't really designed to do that. Nevertheless, someone who scores well (say, 1200 or better) is, in my experience, quite different from someone who does poorly (800 or less). There are exceptions, of course. Some people take tests poorly, and their score would not be indicative of their actual ability.
At the same time, a good score doesn't mean you'll be a success. The test doesn't measure motivation and attitude, and doesn't measure a lot of aptitudes that can lead to success. A low score doesn't mean someone is worthless. It's a correlation with certain types of intelligences, but it doesn't say anything about what it isn't measuring.
It is, however, measuring much more than ability to take tests.
The only reasonable parsing of that phrase is:
Your understanding is completely wrong. There are many kinds of differentiated stem cells in the human body. One trick that researchers would love to figure out is how to take one of those and turn it into an undifferentiated stem cell, which would be similar to or the same as a fetal stem cell.
My understanding of the current ban on federal funding is that it only applies to fetal stem cells, with an exception for a set of already established cell lines. Has that changed while I haven't been looking? I hadn't heard that there were any bans on products, domestic or imported, just on using federal funds to do research.
Why should it be the DSL provider leasing the line to my house? It seems to me that I, the customer, am the one leasing the line - I just want the damned phone company to let ME run over a DSL connection that they forward the the ISP of my choice - the phone company should charge ME the costs for running the equipment up to the ISP, the ISP charges ME additional amounts for their equipment and bandwidth to the Internet, and we're all happy, at least as long as the phone company charges me the same rate as they would if I was connecting to THEM as an ISP.
I just checked with my ISP to see if they could do DSL yet; they said they COULD, but it would be about 20% higher than what SBC would charge me, because SBC bills them at a higher rate than they'd charge me. This should be illegal. If SBC can afford to not only do the DSL transport, but connect me to the Internet, provide e-mail, DNS, Web services, and do customer support on all of that, for LESS than they'd charge to just do the transport to another ISP, then either they're using their monopoly to subsidize their ISP part, or they're overcharging my ISP.
Blocking outbound will also prevent any open relays on their network from going off their network; it is then an internal affair which they can clean up, without getting onto blackhole lists or bothering anyone else.
DNS is inherently hierarchical. e-mail is close to (and for all practical purposes) totally dependent on DNS anyway. MX records already indicate "special" hosts. This is just a way for a peer to only accept e-mail from a host that will accept a response for the mail being sent. It is still peer-to-peer. Unless you're going to advocate only using filtering methods, there has to be SOME way of distinguishing between "legitimate" mail servers and throw-away dialup accounts.
Note that what I'm suggesting is more to move mail clients to use the MSA port (587) instead of the SMTP port, and get everyone to send e-mail by routing it through the "proper" mail server, so that accountability and authentication can be maintained.
I don't accept a solution of only accepting mail from people I have authorized. I don't have a master list of everyone I'd accept mail from. Doing it by having a list of certifying authorities that I'll accept regardless is even more restrictive than what I'm suggesting.
Note that blocking port 25 is only a temporary measure, and that only because the current behavior of all the mail servers using port 25 can't be changed quickly. First everyone sending mail directly, or through the "wrong" mail server, have to be persuaded to start doing it the "right" way. My suggestion is that blocking port 25, along with universally supporting port 587, will get that to happen; once it does, servers that block inbound mail that isn't from the "right" server can do so without risking bouncing a significant amount of legitimate mail. Once enough servers are rejecting based on MX records, the block on port 25 can be removed. Then anyone who complains about receiving SPAM from your network, that didn't go through your mail server, can be simply told to fix their mail server.
My ISP already does a reverse lookup on the sending mail server, and on the return and envelope address, and bounces it if any of them don't resolve. I get maybe 2 or 3 SPAMs a week. Those few that get through would have been rejected if the return or envelope address didn't match the server. Unfortunately, currently, quite a few of the legitimate messages also wouldn't.
Check the sendmail AUTH info, it discusses sendmail connecting as a client. There's also some info on STARTTLS that describes how to authenticate using TLS, but that would require more than just a password configuration on the other end (you'd need to get the remote sendmail to recognize your certificate as being allowed; I don't think there's any way to specify it on a per-user basis).
No, SMTP still runs on port 25, that is too difficult to change in one step; the port you'd actually use (say, 587) doesn't need to be blocked because you require authentication on that.
You can't require authentication on the MTA port because none of the servers out there will be able to authenticate. You can require authentication to relay, but that doesn't block mail directed into your domain. Blocking OUTbound port 25 allows an ISP to control SPAM coming from their ports, without blocking people from using any MSA they want. Not that I like the idea of blocking at all, but the only way to authenticate an SMTP server as being a legitimate MTA is by checking that the return address matches the server it is coming from, and that simply can't be done until pretty much everyone sends mail by routing it through the correct mail server. One way of getting that to happen is to encourage ISPs to block port 25 (along with making sure to enable an authenticating MSA on port 587). Eventually, the block on port 25 can be dropped, as anyone trying to send mail by connecting directly to an SMTP server will be rejected since the MX records don't match.
For a case like this, you'd need to have a relay program that forwarded to, and authenticated with, the correct mail server for the queued messages; you'd need to configure it with authentication info for each address, or else store authentication info (e.g. certificates for use with STARTTLS, or a password for use with AUTH) with each message. Presumably you'd batch things so all messages from the same address would only need one authentication round. The sendmail msp feature is pretty close to this, e.g. it can use the MSA port, by using
but I don't think it can dynamically change which host it relays to.Why not use port 587, as specified in RFC2476; sendmail supports that by default. Any e-mail client that supports authentication should be able to specify the port as well.
Although I do shudder at the idea of an ISP filtering based on port numbers, I believe that filtering OUTbound connections to port 25 would be just barely acceptable, and only because it is so universally used with a protocol so open to abuse. If you really truly want to run a sendmail that doesn't just relay to the ISP, then discuss it with the ISP. You can still receive on port 25 (unless the ISP also wants to restrict servers, but that's a different issue), and that's what most people would want to run their own sendmail for anyway.
Mail clients using the MSA port, combined with MTA's rejecting messages that aren't coming from the server identified by an MX record of the return address, combined with client programs that connected to the correct outbound mail server based on which return e-mail address you're using, would increase accountability. A legitimate server will have limitations on SPAM. A domain that adds bogus MX records could have return addresses black-listed (no more spoofing innocent domains). This would eventually make it safe to eliminate the port 25 filter. Ideally, even internal submissions would require the use of the MSA port, so that becomes the default for all mail clients, and you don't have to worry about switching configurations depending on what network you're connecting through.
The article doesn't say that the pixels can hold the image, only that it uses less power to keep the pixel at the same level. It sounds like you'd still need backing store, and doesn't necessarily mean you can access an individual pixel to change it.
Plasma panels were initially invented to avoid backing store, when memory was very expensive. They replaced storage display tubes for the PLATO project in the early '70s. In that mode, they're monochrome (for PLATO terminals, orange/black); the whole point was that you could turn an individual pixel on or off and it would stay that way.
Did you get explicit permission from the owners of Slashdot before posting (or even reading the article and responses)? I hope it wasn't by e-mail, unless you also received explicit permission to access their e-mail server as well.
How is a well-known protocol, with explicit provision for asking for and providing authorization, any different? Is not a machine exchange of "may I access this network" ... "yes" ... explicit permission, just as sending a HELO command to an SMTP server, and getting back a "200 Hello, pleased to meet you" response, indicates you can attempt to send mail?
Actually, I think dynamic limits could be great if they were done properly. By that I mean you shouldn't hit a brick wall, your usage should be metered down gradually over time as you use more average bandwidth, and metered back up as you stop using it. Could even be tied in with peak use times. Note that I'm talking about fairly short time periods - you should be able to get a burst for a short while (say, a minute or 5 minutes or whatever, depending on what rate you're talking about), then gradually reduce it as long as you're peaking, and gradually raise it when you're not using much. If you're completely idle for a while, you should have a clean slate.
I've done simulations of various algorithms for doing similar things for metering CPU usage on a timesharing system, and it can work. Note that another possibility is to enable the QoS bits in IP to allow you to lower priority in exchange for higher bandwidth; there's always been a problem with determing how to bill for QoS, perhaps just automatically setting rate limits based on it would work better.
If that's compatible, then S100 is compatible with VMEbus is compatible wiht NuBus is compatible with ISA/EISA/PCI. Just because IBM never licensed their MCA patents to anyone to make an ISA/MCA bridge doesn't mean it couldn't have been done. IBM was dumb in the way they handled MCA, but that doesn't make PCI "compatible" with ISA.
I know the 6502 had a limited set of registers. That was my point. It simplifies the architecture immensely, which means you can do load/operate/store very quickly. If you're going to have a limited number of registers, might as well go all the way. Besides, the 6502 effectively had 256 registers in low memory; in a modern version of the 6502, those locations would certainly be implemented on-chip.
The only reason IDE is cheap is because everyone uses it, because it is cheap. Thus, we're stuck with it. FireWire is one possible way out of it.
As for IRQs, I wasn't suggesting that computers shouldn't have interrupts, but that the PC architecture for handling them is lousy. PCI indeed handles things much better - but the PC is still saddled with the old architecture. In the meantime, the Mac had NuBus ages ago, and never had any such problems.
As for serial ports, again I wasn't saying that serial ports are bad, I'm saying that PC serial ports are lousy. Taking the Mac again as an example, they had great serial ports, could do RS-422 as well as pseudo-RS-232, could handle asynch and synchrounous, had a larger internal buffer than standard PC serial chips. Serial connections still have their uses, but for most home use, they are utterly useless now.
PCI is not compatible in any sense with ISA. I think you're thinking of EISA, which was the industry response to IBM's attempt to corner the market by patenting various aspects of MCA so that no one else could make compatible devices or systems.
Another thing he got wrong was that the 8086 chip was not backwards compatible with the 8080 line. It was similar in architecture (limited non-orthogonal register set, awkward instruction set), and there were 8080 -> 8086 cross-assemblers (sometimes producing more than one 8086 instruction for each 8080 instruction), but it wasn't backwards compatible in the same way as 8086 -> 80186/286/386/486/Pentium were.
Wow, a whole 16 general-purpose registers, my heart flutters. Bah, might as well use a 64-bit extension to the 6502.