Even the LAWYER is lying to hide the facts. Quoting from the SJ Mercury:
The company's lawyer, Irwin Schwartz, said damage to its product is "at least at a minimum" now because relatively few people were believed to have downloaded the bypass software
That's obviously a pile of it in some attempt to recover from the majorly bad publicity they have been getting. Of course he also forgets to mention the real reason the loss is so low in the first place.... the product was a piece of crap to start with.
Your honor, they attached this, "word document" or whatever it is, but it's a gibberish inside. I tried to patch it up with vi, but it was totally hopeless.
I'd like to see a web site that lists all these supposed sysadmin jobs.
Seriously, I find it hard to believe all this that recruiters are saying. I want so see some kind of evidence that these jobs really exist as opposed to recruiters merely collecting resumes.
Your problem is that you are a competent manager. Now if you were an INcompetent manager like most of them in business, then you, too, will get to see the same shortage that all the others see. So because you are competent and know what you are doing, you are missing out on all the opportunities to whine like your INcompetent counterparts in corporate American.
Businesses need to pay these programmers and order of magnitude more than the average. If they are that crucial to the corporate success then give them an offer they cannot refuse: $250,000/yr plus $250,000 worth of stock options a year. THAT programmer will f------ a-- STAY there!
If a software developer DOES know how to delete a node from a doubly linked list AND learned that from on the job experience, then I do NOT want to hire them because they will waste to much tim re-inventing things that have already been done and nicely packaged up in many languages already.
Such details are generally NOT of concern for programmers in higher level languages like Perl or Java or even C++. Standard resources and classes provide that for you. If I were hiring to code in those languages, those details would be unimportant.
OTOH, it might still be good to ask that question anyway and see if they answer in a way I just did in the above 2 paragraphs. If they did, I'd probably hire them.
Were you hiring for a C or assembly programming position?
Head hunters and recruiters (99.9% of them) could never have asked the doubly-linked-list question. So that's how their resume slipped through.
Now can you tell me how to search for the node in a binary tree which is has the next higher key than the given key when the given key is not actually present in the tree?
Re:B.S. - want a job? Re:The only "shortage" ...
on
The IT Labor Shortage
·
· Score: 1
You want efficient and Java at the same time? No wonder you aren't getting any resumes.
Re:The only "shortage" is of **CHEAP** tech worker
on
The IT Labor Shortage
·
· Score: 1
Web designers are all over the place. Why did you need to hire someone from Belarus? What kind of special skills did you need, that he managed to acquire in that former Soviet state? Or what out of the way town did you locate in just so you could go hiring people from outside the country?
The TOP TEN (serious) reasons why a shortage claim
on
The IT Labor Shortage
·
· Score: 2
The TOP TEN reasons why business says there is an IT labor shortage are:
10 - Smaller business is still too strapped for cash, and offers puny resources, such maybe one or two servers to run a network from, poor hand-me-down workstations to develop software on, and on, and on.
9 - Larger corporations might be more attractive to some because of the great resources such a business would have, but these same corporations are a big turn off to most of the technical people that could otherwise serve them well. A large portion (most) of technical people would rather come to work totally casual every day, as well as other things like telecommuting from home, or working whatever hours they like (software developers should be about to do this quite flexibly while network administrators might have more difficulty).
8 - It's quite hard to find someone who is good, even though there are plenty of people out there. Recruiters even have a hard to finding them. One of the reasons for this is that there is no one good central place to look. And this situation has even gotten worse with there being hundreds of JOBS web sites out there now. And none of those sites lists all available jobs, so anyone looking for a job, as well as anyone looking to hire, and to scrounge around lots of them to find good jobs or people.
7 - Larger corporations tend to favor commercialized technology and are afraid of the emerging open source and free software because they don't know who they can sue if it doesn't work. Similarly they are afraid they won't be able to hire someone that knows open source and free software when the one who did set it up leaves for more money somewhere else (because they somehow get the impression the corporate vendor of the commercial stuff can come bail them out if they suddenly find themselves way understaffed with commercial software, which is totally untrue).
6 - The making of internet millionaires (and a few billionaires) has diverted the interests of many to seek out more money, and especially perks like stock options, employee ownership, and even partnership. Some are even starting their own business. And much of this is because the traditional businesses are failing to offer substantial incentives like stock options, which they could (but tend to reserve for the top corporate officers).
5 - Jobs that businesses do have open are often so complex, with so many things to be dealt with, that the requirements for the job end up listing dozens of different skills and experiences that are needed. In many cases HR doesn't realize that perhaps only half a dozen people in the world can qualify. Business doesn't want to hire people who need to learn something to do the job, while at the same time, people do tend to want a job where they are challenged to learn something new, and increase their skills (which in the end should help business, but managers haven't figured that out, yet).
4 - The rapid pace of computer technology, made even more rapid by the emergence of the internet, has put tremendous time pressures on business to implement and deploy everything from software to networks to web sites, quickly. While business has in the past been willing to hire people who've learned the theory in college, and train them on their ongoing technology, business today has found it has to hurry and pull their pants up and bring in people who already know how to do things that the business has no idea about. They need these people to be able to start even before they've finished the paperwork in the HR office.
3 - Unemployment statistics are effectively spin-doctored. At the beginning of the Clinton administration, USA employment statistics rules were changed. They now do not count as unemployed the software engineer who is currently working flipping burgers at the local greasy burger place. As a result, these underemployed people are just not counted and the job market looks different than it really is.
2 - In its zeal to hire only experienced people, and put back on the street those who are not experienced, they have caused fewer and fewer people than otherwise would be to be getting that actual experience they demand. An analogy to this would be tearing down old houses to get wood to build new houses because no one has the time, desire, or money, to plant new forests.
1 - Business wants it, but it doesn't want to pay for it. People come to work for a company, get a few years experience and a certification or two, and get somewhere between no and a trivial raise to compensate. Then the person gets hired away for much more money and the see they can't hire a replacement without having to offer more.
Anyone who has taken Economics 101 knows (if you didn't sleep through the course) the laws of supply and demand. When the demand goes up and/or the supply goes down, the prices go up. Correspondingly when the supply is up and the demand is down, prices go down. When the unemployment is high, business is happy because salaries and wages stay low. People are willing to work for less when the alternative is no work at all. Business fights to prevent efforts by the government to raise any salaries or wages, as well as fights efforts to impose more benefits like health care. Business insists that economics are at work and things should be left alone. Turn the tables on business and give it a situation where there are indeed fewer people, which would cause salaries and wages to go up, then business suddenly doesn't want to play by the same rules, and wants the government to step in and change things to their benefit.
I suggest that the real problem in the USA is a shortage of competent management.
First of all, has anyone gathered a list of site using doubleclick? This should be easy enough to get, given the fact that said sites will reference doubleclick in some IMG tag. Soon we could have a complete list of all their business relations, and potentially use that data for something, maybe. Anyway, I figure it just might be interesting to turn the tables on them, and since it would be a new compilation of information, the copyright ownership would not be theirs.
Now, if the browsers didn't pass cross-domain or cross-host HTTP referer information on requests that also had a cookie, we could could still get the benefits of cookies within a site, but the request for the ad image would have no referer data. What would DoubleClick do with no referer info? Refuse to give is the ad image?:-)
Since I just happen to be setting up a squid proxy this week, and I always compile primary services from the latest source code, I figure I'll take a peek under the hood and see how easy it would be to make it modify the request so that if the domain of the cookie and the domain of the referer do not match, discard one or the other, or both, of them before sending on the request.
Not to mention the fact that with Cisco it's also easy to automate your configurations so that you have a central store of exactly what configuration everything is supposed to have, and tools to make sure it does (alert you when a router is inconsistent, or just automatically correct it from the database). The 3Com equipment just couldn't do that. For "remote control" all the 3Com sales engineer (that's a contradiction of terms anyway) could deliver was a candy coated client program for that other OS that still required a person to be there to manually control everything (ok... so I'm an automation phreak).
3Com constantly changes the hardware API of their NICs without changing the models. That and when they change models, they don't release the specs for making a driver. Thus, they too often failed to work at all in Linux (and I suspect would be as much of a problem in BSD). I got totally fed up with their junk and switched to Netgear FA-310TX and have been happy with them ever since. And I'm paying a lower price.
As for underpowered? The 3Com NICs, when they did work, did deliver speed, so I won't knock them on that account. But my Netgear cards are doing it as well.
BTW, I don't need a wake-on-LAN function as my servers are always awake (they run Linux). Besides, my LANs are way too active anyway.
Cheating? That's what life is about...
on
Laptop Exams?
·
· Score: 1
...as long as you can get away with it. Looking up information hasn't been a form of cheating for ages.
I remember a physics lab exercise I did in 1975. We got the description of the lab we were supposed to do a week in advance in the previous lab. This one required setting up a table with various different weights pulling via strings in different directions on a small ring in the center around a little post. We had to adjust the weights until the ring did not contact the post. Considering how long it took the others, this wasn't all that easy to do.
I, however, had access to the mainframe, and whipped up a little program in Fortran that simulated the setup for various vectors and weights, and found a combination of six vectors that all cancelled out. I memorized those numbers and when time came to do the lab, I simply set up my arbitrary vector angles and weight distributions as I had gotten them from my program, and there was the post dead center in the middle of the ring.
The prof was running the lab and noticed I was done in 5 minutes and came over to check. When he asked how I did that, I told him I had run the simulation in advance. He asked what program I had used and I told him I wrote my own. He wanted to see it so I gave it to him. I got an A for the day and 2 hours 50 minutes of spare time to goof off:-)
Re:Don't that just make the questions harder?
on
Laptop Exams?
·
· Score: 1
Now days if it's a take home, it would end up being:
"Classify all simple rings of size less than or equal to order 60 up to isomorphism. Build web page."
Re:Laptops are not bad for economic diversity
on
Laptop Exams?
·
· Score: 1
In the days before personal computers (like when I went to school in the days of mainframes) open book exams did mean leafing through tons of useless material. But then, life meant leafing through tons of useless material, too, so part of the skill of the course was to learn how to find signal out of noise. Today, there is not only more signal, but also more noise. The sifting tools are more powerful to match, and we have to know how to use them.
Re:Ugh. I'd rather not hire you.
on
Laptop Exams?
·
· Score: 1
Then get together with a bunch of other students in the class and, as part of the whole studying process, build your own web site that is the gathering point for all the information you find. Now you have instant links to the vast web of knowledge and even all your class notes.
If you have all this trouble yourself with using the web, and you let it get to you like that, then you're probably not worth hiring for a high tech job.
As long as this is a school where all the students have equal access to laptops and internet, and the exams are properly designed with all this taken into consideration, then I'm all for it. If there is an issue of laptop access equality, then the exam could be given in a room with computer and network facilities. Since some schools do have a student laptop program, this would work well for them. Others would have to find a way around that.
As for the fairness issue in situations where some students have, and some don't, the answer to that is that those who have prepared themselves with computers and internet while in school will have an (unfair?... no) advantage over other out in the job market. When they get their first job, and nearly all that require a college education do involve use of a computer, and nearly all require that use to be extensive, how well they are prepared to handle their field of study and work with the tools of the information age, will define how well they do in work, and in life. That may not be equal, but it is fair, because that is life.
One of the issues I do think needs to be considered is what kinds of tools are on this computer. Surely students get to choose what web sites they get information from during the exam. They should be allowed to choose the software tools they get to use as well, within reason. In real life jobs, they may not get to choose all those tools, but more and more there is some choice even in the workplace. Further, when you consider the high tech jobs that many college science and engineering students go into, systems like those from Microsoft no longer prevail. And the BSD/Linux movement is even making inroads into the high tech workplace. Unless the coursework specifically involves a specific software tool, classes and exams should be computer system neutral at least to the extent that the students themselves are capable and willing to exercise a choice in the matter.
Re:More money = better grade at the end?
on
Laptop Exams?
·
· Score: 1
If the professors themselves forget to think, and don't make exams for laptop wielding students that work to require them to think in addition to knowledge, then, sure, this can be a problem. However, real life will require use to think, and think in a certain way that is based on access to vast information, both at our jobs and at work. The quality of the exam is important as it has to make people think about how to use the information they can get.
I guess we disagree in the gray fuzzy area between dervied and original work. The definition in USC 17 doesn't apply in my mind because I don't considering it to be "based upon" in the way the law is using it.
A compiler takes a work in the form of a source code program file, and produces an object code output file that can be (with maybe some other stuff added) executed in some way. The object code output is a derived work, but the compiler itself would not be considered derived from the input file just because it can do something with it, since it can also do stuff with anything else in the range of the language it compiles. Not all patch types are so simple as to have to be applied to one and only one. Indeed, a "patch" that totally deletes all the original code and inserts its own could be applied to just about any code and perform its intended function correctly. Why would this patch be any more a derived work than the compiler? At what point to you consider a patch to be a derived work? Where do you draw the line?
Sure, if you make the patch using the method most common in the UNIX/OSS world, which I assume is the only way you know how to do it, then, yes, that is a derived work.
A handwritten patch that doesn't include any context or other bits of the original is not derived from that work. It can modify it, and that result is derived from both, but this doesn't mean the patch must be derived from that work.
That has to be answered on a case by case basis. Most patches are because the usual way in the UNIX and OSS world is to edit a copy of the original, and generate the patch from a difference between original and the new work. That kind of patch is definitely derived from the original, and definitely must adhere to the rights granted by the original. GPL is clear on how derived works must be covered by the same license.
It is possible to generate a kind of patch that is not derived from the original. The singular case is a patch which deletes the entire original and replaces it with entirely new code. As long as that deletion is done in a way that has no prior knowledge of the original, then the license of the original clearly does not apply. The license of the patch would prevail.
If a patch is not derived from the original, but is written with the knowledge of the original, then this might be questionable (consider some of the reverse engineering cases). Context patches would have bits of the original so those don't fall under a non-derived patch. But would a patch that only identifies line or byte offsets, or file names, and provides replacement code, be a derived work? IMHO, the answer to that is no.
Given my previous opinion standing, if you took an original work, and a patch, and applied the patch, the rights you have in the result would be the set of rights that is in common in both sources. Since GPL gives no right to distribute with any rights other exactly the set granted by GPL, the resultant rights would be the null set unless the patch also was GPL.
Assumption #1 is false. Therefore #2, #3, and #4 fail due to #1 not standing.
Patches USUALLY are based on the program they patch, and your perspective may be clouded by an incomplete experience with only such patches. Your logic seems sound for those cases (patches that are derived from the original), but it does not apply to cases where patches are not based on the program.
The RESULT of the patching, taken as a whole, can inherit only those rights derived from BOTH sources (the original the patch applied to, and the patch itself). For example, if the original does not grant a certain right, the patch cannot add on that right.
If the rights granted by the patch are mutually exclusive from the rights granted by the original (suppose the patch requires distribution in binary without source and the original requires distribution of source) then the result grants no rights whatsoever (null set).
GPL is safe in that a patch cannot add new rights (GPL is a license that essentialy denies a lot of rights, including denying the right to DISTRIBUTE with any rights changed). It can take some away, but that's just a deadend branch of the tree of distribution for said rights since distributing such a derived work isn't granted by GPL. However, a patch that is NOT derived from the original (and I have written such patches before) does not have to be licensed under GPL.
Therefore a patch with a license in conflict with GPL, and not derived from the original, can be distributed, and even applied, but the result most likely cannot be further distributed.
Slashdot Mirror
Even the LAWYER is lying to hide the facts. Quoting from the SJ Mercury:
That's obviously a pile of it in some attempt to recover from the majorly bad publicity they have been getting. Of course he also forgets to mention the real reason the loss is so low in the first place .... the product was a piece of crap to start with.
Re: attached word document
Your honor, they attached this, "word document" or whatever it is, but it's a gibberish inside. I tried to patch it up with vi, but it was totally hopeless.
Make the defining factor be whatever arms you can CARRY ... if you can CARRY it, you have a right to defend yourself with it.
How many employers in Philly are paying programmers at least $200K?
I'd like to see a web site that lists all these supposed sysadmin jobs.
Seriously, I find it hard to believe all this that recruiters are saying. I want so see some kind of evidence that these jobs really exist as opposed to recruiters merely collecting resumes.
Your problem is that you are a competent manager. Now if you were an INcompetent manager like most of them in business, then you, too, will get to see the same shortage that all the others see. So because you are competent and know what you are doing, you are missing out on all the opportunities to whine like your INcompetent counterparts in corporate American.
Businesses need to pay these programmers and order of magnitude more than the average. If they are that crucial to the corporate success then give them an offer they cannot refuse: $250,000/yr plus $250,000 worth of stock options a year. THAT programmer will f------ a-- STAY there!
Show where your job opening is being advertised.
If a software developer DOES know how to delete a node from a doubly linked list AND learned that from on the job experience, then I do NOT want to hire them because they will waste to much tim re-inventing things that have already been done and nicely packaged up in many languages already.
Such details are generally NOT of concern for programmers in higher level languages like Perl or Java or even C++. Standard resources and classes provide that for you. If I were hiring to code in those languages, those details would be unimportant.
OTOH, it might still be good to ask that question anyway and see if they answer in a way I just did in the above 2 paragraphs. If they did, I'd probably hire them.
Were you hiring for a C or assembly programming position?
Head hunters and recruiters (99.9% of them) could never have asked the doubly-linked-list question. So that's how their resume slipped through.
Now can you tell me how to search for the node in a binary tree which is has the next higher key than the given key when the given key is not actually present in the tree?
You want efficient and Java at the same time? No wonder you aren't getting any resumes.
Web designers are all over the place. Why did you need to hire someone from Belarus? What kind of special skills did you need, that he managed to acquire in that former Soviet state? Or what out of the way town did you locate in just so you could go hiring people from outside the country?
The TOP TEN reasons why business says there is an IT labor shortage are:
Anyone who has taken Economics 101 knows (if you didn't sleep through the course) the laws of supply and demand. When the demand goes up and/or the supply goes down, the prices go up. Correspondingly when the supply is up and the demand is down, prices go down. When the unemployment is high, business is happy because salaries and wages stay low. People are willing to work for less when the alternative is no work at all. Business fights to prevent efforts by the government to raise any salaries or wages, as well as fights efforts to impose more benefits like health care. Business insists that economics are at work and things should be left alone. Turn the tables on business and give it a situation where there are indeed fewer people, which would cause salaries and wages to go up, then business suddenly doesn't want to play by the same rules, and wants the government to step in and change things to their benefit.
I suggest that the real problem in the USA is a shortage of competent management.
First of all, has anyone gathered a list of site using doubleclick? This should be easy enough to get, given the fact that said sites will reference doubleclick in some IMG tag. Soon we could have a complete list of all their business relations, and potentially use that data for something, maybe. Anyway, I figure it just might be interesting to turn the tables on them, and since it would be a new compilation of information, the copyright ownership would not be theirs.
Now, if the browsers didn't pass cross-domain or cross-host HTTP referer information on requests that also had a cookie, we could could still get the benefits of cookies within a site, but the request for the ad image would have no referer data. What would DoubleClick do with no referer info? Refuse to give is the ad image? :-)
Since I just happen to be setting up a squid proxy this week, and I always compile primary services from the latest source code, I figure I'll take a peek under the hood and see how easy it would be to make it modify the request so that if the domain of the cookie and the domain of the referer do not match, discard one or the other, or both, of them before sending on the request.
Not to mention the fact that with Cisco it's also easy to automate your configurations so that you have a central store of exactly what configuration everything is supposed to have, and tools to make sure it does (alert you when a router is inconsistent, or just automatically correct it from the database). The 3Com equipment just couldn't do that. For "remote control" all the 3Com sales engineer (that's a contradiction of terms anyway) could deliver was a candy coated client program for that other OS that still required a person to be there to manually control everything (ok ... so I'm an automation phreak).
3Com constantly changes the hardware API of their NICs without changing the models. That and when they change models, they don't release the specs for making a driver. Thus, they too often failed to work at all in Linux (and I suspect would be as much of a problem in BSD). I got totally fed up with their junk and switched to Netgear FA-310TX and have been happy with them ever since. And I'm paying a lower price.
As for underpowered? The 3Com NICs, when they did work, did deliver speed, so I won't knock them on that account. But my Netgear cards are doing it as well.
BTW, I don't need a wake-on-LAN function as my servers are always awake (they run Linux). Besides, my LANs are way too active anyway.
...as long as you can get away with it. Looking up information hasn't been a form of cheating for ages.
:-)
I remember a physics lab exercise I did in 1975. We got the description of the lab we were supposed to do a week in advance in the previous lab. This one required setting up a table with various different weights pulling via strings in different directions on a small ring in the center around a little post. We had to adjust the weights until the ring did not contact the post. Considering how long it took the others, this wasn't all that easy to do.
I, however, had access to the mainframe, and whipped up a little program in Fortran that simulated the setup for various vectors and weights, and found a combination of six vectors that all cancelled out. I memorized those numbers and when time came to do the lab, I simply set up my arbitrary vector angles and weight distributions as I had gotten them from my program, and there was the post dead center in the middle of the ring.
The prof was running the lab and noticed I was done in 5 minutes and came over to check. When he asked how I did that, I told him I had run the simulation in advance. He asked what program I had used and I told him I wrote my own. He wanted to see it so I gave it to him. I got an A for the day and 2 hours 50 minutes of spare time to goof off
Now days if it's a take home, it would end up being:
"Classify all simple rings of size less than or equal to order 60 up to isomorphism. Build web page."
In the days before personal computers (like when I went to school in the days of mainframes) open book exams did mean leafing through tons of useless material. But then, life meant leafing through tons of useless material, too, so part of the skill of the course was to learn how to find signal out of noise. Today, there is not only more signal, but also more noise. The sifting tools are more powerful to match, and we have to know how to use them.
Then get together with a bunch of other students in the class and, as part of the whole studying process, build your own web site that is the gathering point for all the information you find. Now you have instant links to the vast web of knowledge and even all your class notes.
If you have all this trouble yourself with using the web, and you let it get to you like that, then you're probably not worth hiring for a high tech job.
As long as this is a school where all the students have equal access to laptops and internet, and the exams are properly designed with all this taken into consideration, then I'm all for it. If there is an issue of laptop access equality, then the exam could be given in a room with computer and network facilities. Since some schools do have a student laptop program, this would work well for them. Others would have to find a way around that.
As for the fairness issue in situations where some students have, and some don't, the answer to that is that those who have prepared themselves with computers and internet while in school will have an (unfair? ... no) advantage over other out in the job market. When they get their first job, and nearly all that require a college education do involve use of a computer, and nearly all require that use to be extensive, how well they are prepared to handle their field of study and work with the tools of the information age, will define how well they do in work, and in life. That may not be equal, but it is fair, because that is life.
One of the issues I do think needs to be considered is what kinds of tools are on this computer. Surely students get to choose what web sites they get information from during the exam. They should be allowed to choose the software tools they get to use as well, within reason. In real life jobs, they may not get to choose all those tools, but more and more there is some choice even in the workplace. Further, when you consider the high tech jobs that many college science and engineering students go into, systems like those from Microsoft no longer prevail. And the BSD/Linux movement is even making inroads into the high tech workplace. Unless the coursework specifically involves a specific software tool, classes and exams should be computer system neutral at least to the extent that the students themselves are capable and willing to exercise a choice in the matter.
If the professors themselves forget to think, and don't make exams for laptop wielding students that work to require them to think in addition to knowledge, then, sure, this can be a problem. However, real life will require use to think, and think in a certain way that is based on access to vast information, both at our jobs and at work. The quality of the exam is important as it has to make people think about how to use the information they can get.
I guess we disagree in the gray fuzzy area between dervied and original work. The definition in USC 17 doesn't apply in my mind because I don't considering it to be "based upon" in the way the law is using it.
A compiler takes a work in the form of a source code program file, and produces an object code output file that can be (with maybe some other stuff added) executed in some way. The object code output is a derived work, but the compiler itself would not be considered derived from the input file just because it can do something with it, since it can also do stuff with anything else in the range of the language it compiles. Not all patch types are so simple as to have to be applied to one and only one. Indeed, a "patch" that totally deletes all the original code and inserts its own could be applied to just about any code and perform its intended function correctly. Why would this patch be any more a derived work than the compiler? At what point to you consider a patch to be a derived work? Where do you draw the line?
Sure, if you make the patch using the method most common in the UNIX/OSS world, which I assume is the only way you know how to do it, then, yes, that is a derived work.
A handwritten patch that doesn't include any context or other bits of the original is not derived from that work. It can modify it, and that result is derived from both, but this doesn't mean the patch must be derived from that work.
Is a patch derived from the original?
That has to be answered on a case by case basis. Most patches are because the usual way in the UNIX and OSS world is to edit a copy of the original, and generate the patch from a difference between original and the new work. That kind of patch is definitely derived from the original, and definitely must adhere to the rights granted by the original. GPL is clear on how derived works must be covered by the same license.
It is possible to generate a kind of patch that is not derived from the original. The singular case is a patch which deletes the entire original and replaces it with entirely new code. As long as that deletion is done in a way that has no prior knowledge of the original, then the license of the original clearly does not apply. The license of the patch would prevail.
If a patch is not derived from the original, but is written with the knowledge of the original, then this might be questionable (consider some of the reverse engineering cases). Context patches would have bits of the original so those don't fall under a non-derived patch. But would a patch that only identifies line or byte offsets, or file names, and provides replacement code, be a derived work? IMHO, the answer to that is no.
Given my previous opinion standing, if you took an original work, and a patch, and applied the patch, the rights you have in the result would be the set of rights that is in common in both sources. Since GPL gives no right to distribute with any rights other exactly the set granted by GPL, the resultant rights would be the null set unless the patch also was GPL.
Assumption #1 is false. Therefore #2, #3, and #4 fail due to #1 not standing.
Patches USUALLY are based on the program they patch, and your perspective may be clouded by an incomplete experience with only such patches. Your logic seems sound for those cases (patches that are derived from the original), but it does not apply to cases where patches are not based on the program.
The RESULT of the patching, taken as a whole, can inherit only those rights derived from BOTH sources (the original the patch applied to, and the patch itself). For example, if the original does not grant a certain right, the patch cannot add on that right.
If the rights granted by the patch are mutually exclusive from the rights granted by the original (suppose the patch requires distribution in binary without source and the original requires distribution of source) then the result grants no rights whatsoever (null set).
GPL is safe in that a patch cannot add new rights (GPL is a license that essentialy denies a lot of rights, including denying the right to DISTRIBUTE with any rights changed). It can take some away, but that's just a deadend branch of the tree of distribution for said rights since distributing such a derived work isn't granted by GPL. However, a patch that is NOT derived from the original (and I have written such patches before) does not have to be licensed under GPL.
Therefore a patch with a license in conflict with GPL, and not derived from the original, can be distributed, and even applied, but the result most likely cannot be further distributed.