As far as site selection, why not go with the Japanese? After all, back in 1979 Tomino did the first Gundam anime series
You know, Gundam is a pretty amazing and important piece of anime, but you don't cite it when you bring up something scientific because it's science-fiction.
Much science-fiction is based on hard science; now, this is a bit old fashioned in the US, but this is one of the many ways the Japanese are fortunate in being "behind the times".
A good amount of the Tomino Gundam Universal Century (UC) future history was lifted straight from current and still valid hard science and engineering theories, aneutronic helium 3 fusion and O'Neil colonies being the most obvious examples. They also pay a lot of attention to details like heat dissipation. (Let's not talk about Minoskivy (sp.) physics or Newtypes, though.:-) This is especially amazing since it's just background that the show itself doesn't bother to explain; it's expected the audience will look up the details if they're interested.
Citing Gundam in this case is like saying something must be true because you read it on Slashdot.
Personal insult like the above also shouldn't have a place on Slashdot.... I cited Gundam UC mostly in humor --- I'm sorry I didn't make that obvious to you --- but also because of the intersection of Japan being a possible site and how remarkable UC is for its attention to scientific and engineering detail, including the critical and relatively obscure answer to D-T neutronic fusion. And perhaps the fact a culture is more serious about getting these details right is worth consideration?
(Although unfortunately pretty much only for refinement; the Japanese system tends to be stifled by seniority, and it has not escaped their notice that every one of their Nobelists had to leave the country to do his research....)
The MBTA and Amtrak have already begun construction of an underground rail-link tunnel between North Station and South Station. Fear not.
Wow! It's not well known history, but North and South Station have never been really connected. There was a "short haul" line to service factories and the like, but ventilation problems back when locomotives were coal fired prevented all but that sort of light traffic from using it.
That link was shut down in 1911 or so as I remember, and the last time I checked moving rail cars from north to south requires running them right through MIT and ~ 12 miles out to Framingham and then back. For cargo, this will probably remain the case (cargo and passenger trains have different requirements based on cargo weight and passenger speed).
It's going to be great for the areas north of Boston if they pull it off (especially with airline trips getting ever more obnoxious). The Boston area has a great commuter rail system (as of a dozen years ago), but the difficulty connecting from North to South Station (you would have to get off a commuter train, get on the Green line, then transfer to the Red to get to South Station and its set of commuter lines) is a severe problem. And if Amtrack can extend service up north....
Also, wasn't the interstate system built to quickly access different areas of the country, should it ever be invaded?
Or if we ever had some other need to move lots of troops from A to B. One of the men pushing it was President Eisenhower, who before WWII had personally experienced the nightmare of trying to move through our old style highway system, with massive congestion at every small town. Our men who took Europe also noted how useful Hitler's Autobahns were. You want as many movement options as possible, especially land based ones.
Also note that even if you plan on using trains for troop and equipment moving, you must have alternatives in case of sabotage, accidents, etc. Note how difficult it was for the 1AD and 1ID to escape from the lands of the weasels when various railroads were denied to them.
I'll add that I lived in the Boston area for a dozen years before leaving just before this began; I can't comment on the cost et. al., but I can say the benefits as stated are massive. A lot of transportation nightmares have been mitigated.
Wouldn't fusion have to have been made practical for terrestrial power generation before anything like this should be started on?
Indeed, but this project is explicitly designed to be the next "scale up" towards that goal. A design goal of 500MW of fusion power output is nothing to sneeze at....
On the other than, practical fusion is much further away than is advertised, since it requires fusing helium 3, which doesn't produce neutrons, but is a lot harder to fuse. Otherwise your reactor's atoms are slowly transmuted into other (frequently radioactive) elements as it runs. We also have to get a good source of helium 3 ("They're going to strip mine the moon!" the enviros are already whining).
As far as site selection, why not go with the Japanese? After all, back in 1979 Tomino did the first Gundam anime series, and part of its background was small fusion reactors running on helium 3, allowing for a) lots of power, b) big explosions if one can't shut down properly (this is explicit from the beginning), and c) a Jupiter Energy Fleet for the helium 3 (modeled on the petro-rich Arabs --- remember that Japan has to import all of its oil, and I think most of that comes from the Middle East) which is always behind each war, pulling the strings (with the exception of Zeta Gundam where one of them actually showed up; get it when it comes out next year in the US, it's very good)).
Of course when everyone was on dialup alot of ISPs got killed by unlimited plans. As im sure you know (maybe others don't) the problem was the rule of thumb was around 7 accounts per modem or so. So if one person stayed on for 24 hours straight, they tied up a phone line and modem for the entire24 hours and completely throw off your numbers.... in fact it ends up costing more in equipment and telephone fees to keep that user than you make from them.
BZZT!
This turns out not to be the case, at least for dialup. With the inherent limited bandwidth requirements, their cost was almost entirely up front, in their build out for peak use. It cost them exactly as much to provision a 24x7 user as the archetypical grandmother... who was online during their peak period(s).
Now, it's a bit different. IP addresses are perhaps more scarce (although that wouldn't seem to be an issue in this case, since there aren't many large bandwidth users), and peak bandwidth demands are obviously much higher, and can be really troublesome to build out (especially for cable).
The answer to this is obviously competition; e.g. telephone xDSL companies know they have a competitive advantage (as a rule) in reliability and keeping up with growth due to the different network designs, and people like me are willing to pay for that.
And obviously some companies realize that going to war with your customers is never good business, as outlined by many others.
Funny you should mention I will Fear no Evil; Jerry Pournelle commented on his blog that at the time it was published, Heinlein was so concerned about money that he didn't have it edited; it was rushed to publication (RAH though it needed some serious work).
Pournelle is pretty sure that Heinlein never desired this book to be published, because he would have otherwise done so when money was tight....
Which is not to say that it isn't worthwhile, especially as a snapshot of the author when he wrote it.
[...] he really does go all Ayn Rand at the end there... is I think an oversimplification; there's major evolution in his writings (that got published) starting in the '60s, and with a couple of exceptions (Mistress, Lazarus Long) not at all for the better, I think.
I agree his best works were from before, especially almost all of the "juveniles" (of which Troopers was the last),
but I don't think his basic philosophy changed very much, although it's probably more apparent. I think the quality of his work peaked in the '50s, but the stuff before then is definitely worth checking out. Above all, get a copy of the trade paperback Expanded Universe (published in the early '80s or so); you will learn more about Heinlein and a lot of other things (the E.E. "Doc" Smith micro-biography is priceless) from it than any other single source.
(Pournelle commented perhaps a decade and a half ago on BIX that Heinlein really was a philosopher, but the US simply doesn't have a niche for real philosophers as philosophers in the modern era.)
One thing I find interesting is that I grew up 100 miles south of him, and I often understand exactly where he's coming from; I have the same (apparent) visceral reactions to certain things that RAH did....
I know I wouldn't want to work in an open office space, with no cover, knowing there were hundreds of disgruntled, anti-social nerds out there with nothing to do...nerds that had been fired from the job that I was doing...
Note: cover is that which will stop bullets, shell splinters, etc. I.e. a good earth berm, a ditch, a really strong wall (normal brick/cinder block is not enough, at least not for 7.62 NATO). I wouldn't want to depend on modern (cheap) desks, cubicles, or interior walls or most doors for cover....
Concealment is that which keeps you hidden... but on the battlefield with riflemen with semi-autos (or machine guns, etc.) it has less utility... e.g. squads of WWII US riflemen with M1 Garand rifles could often saturate an area of cover behind which they knew the enemy was (and don't forget the BAR men, of course). For a modern illustration, watch e.g. the beginning of The Days of the Condor.
I agree... such workspaces are to be avoided.... Worst are the ones with good salesmen. who by definition are talking in a way that's hard to avoid listening to. I like them... if they're in their own office(s). ^_^
He's like, "oh, I took a bunch of math and physics courses, but I forgot to my humanities!"
Simson exists (hell, he inspired a "Three Minute Hate" at MIT's biggest auditorium (1300 seats) when Berkley Breathed was asking for suggestions on naming a new character (rather frightening to be there, seeing the other sort of mob "action")).
As for his academic creds, he graduated from MIT with a degree in Chemistry and two in Humanities
(he told me he could have gotten a third of the latter type with another course or two, but someone told him "enough was enough":-).
And as previously mentioned/implied, he then graduated from the Columbia School of Journalism.
Try again, or better yet, lose your "MIT envy".
(And for the record, MIT requires exactly two math (calculus) and physics (mechanics, E&M, optics, etc.) courses. Basically the way we understood the world to work as of the end of the 19th century.
Lots better than, say, Harvard, where you just have to prove that you can do algebra....)
Re:Simson Garfinkel is a real *journalist*
on
Mafia Tech Support
·
· Score: 1
(For whatever that is worth:-)
Simson has been a friend of mine for many years; after graduating, he went to the Columbia School of Journalism, and while still doing a lot of fun programming make a respectable career in science writing. (rough quote in late '80s, much more modest when he said it (not that he's very modest:-) "I knew I had 'arrived' when I called the President of the AMA and he recognized me by name").
So you can trust him and the article as much as you'd trust any honest journalist (and his editor(s); Simson does not have the final say on what's under his byline).
And he has the tech to back it; e.g. he mastered the 2-4th CD-ROMs in the US (two tests, and then all surviving Ancient Greek, using his own file system, which is *major* prior art in the area). He walks the walk, at a very high level. (If the ISO guys had listened to him, we wouldn't have suffered the horror of the multi-session CD-R transition....)
(Maybe this time I can convince him to buy a home defense firearm, but I doubt he's in much danger, unless "the mob" wants to find his source, and even then they'd ask "politely" before making him an offer he couldn't refuse....)
Me: (We'll ignore for the duration of this discussion the horrific raw odds of any tech merger working (i.e. Novell buying SUSE).)
Well, I guess we can't ignore it.
Why? Novell buying SuSE can be a good thing for this - my college uses [...]. That CAN become NoSE Server/[...]
Emphasis added only in the bolding of "can". which is the operative word as you implicitly acknowledge.
The harsh, brutal fact is that most high-tech mergers and acquisitions fail. I seem to remember one figure of 90%, and that sounds about right; the record is really grim. Only Cisco managed to consistently do it, by virtue of viewing a company acquisition as the best way of hiring good people....
The sad fact is that anyone with a knowledge of high-tech industry history is going to write off SUSE Linux for several years until we see if it's going to be a one out of ten exception....
One other really unfortunate detail is that SUSE has been the "European Linux" up to now; there is perhaps enough anti-American bigotry that this purchase by a company in a notoriously religious state will cause SUSE to loose a lot of its current market share.
(Understand, I assume everyone has the best of intentions, the purchase makes great sense in terms of filling a gaping hole in Novell's offerings and in theory providing SUSE with many good things, etc. etc.... but that doesn't mean they'll be able to execute it.)
[...] CS majors will want a linux distro that offers support.
Which is why Red Hat's sustained new attitude of "servers are for the elite" continues to puzzle me.
Some respectable fraction of these CS Majors need/want to work on servers (i.e. the RHEL ES version). They need what it offers, they want to be able to put it on their resume (e.g. "provided 'this useful campus service' using RHEL ES"), etc. etc.
At my school, companies fell over themselves to try to put their products in front of students, who would soon enough be influencing and then making buying decisions.
Once again, something about Red Hat's new business model does not compute... and SUSE continues to look good on the surface, especially with non-profit discounts (they have to give the same discount to the Federal government, so why not non-profits, which probably represent much lower sales?).
(We'll ignore for the duration of this discussion the horrific raw odds of any tech merger working (i.e. Novell buying SUSE).)
A good RNG. Unless you're talking about someone flipping coins or rolling dice, every way to do so with a computer will use *gasp* an algorithm.
Or use thermal noise....
Last decade the company I worked for wanted some very random numbers, and with a $300 small pod like device that attached to a parallel port and got entropy from (as I remember) a hot diode we got them.
Don't know the current state of the market for these devices, or if they would apply to this architecture, but it can be easily and cheaply done.
Does the DDK work with compilers other than Microsoft Visual $1000++?
Heh.
The "last time I checked" in earnest was many years ago; back then, the question would have been "Does the DDK work with MVC/C++, instead of the MSDN supplied compiler and debugger (NT was not originally developed with MSVC++:-).
(While I followed the DDK story for a while longer, "in earnest" was a decade ago, when through simple lunch seating luck I arranged with a Microsoft I/O technical lead that the company I worked for would help MS define a driver class for high capacity scanners such as the Kodak Imagelink 900 family (2 pages/second, the sort for which TWAIN et. al. are not a solution). My boss shot this down on the day Clinton was first elected, along with the entire project for us to do our scanner system on NT; pretty obviously he managed to "shoot down" his entire company within a few years....)
Today, I have no idea, and getting a legal copy of the DDK (as opposed to "borrowing" someone's old release from MSDN) is not free. I was just pointing out that device driver development is not closed (and that if you have some home brew hardware I really doubt that Microsoft is going to come after you with a rusty knife if you didn't get a MSDN subscription...).
Also, I have to ask when you're realistically going to have hardware that needs the DDK (which wasn't much if anything more than header files, tools, and documentation, after all), and not be able to afford the software. Can one make a PCI interface circuit board really cheaply? (I don't know, or know if there are e.g. FPGA based ones that you'd then need to do driver work for, or their cost.)
Since [ Windows is ] closed source, you are FUCKED if there is no drivers.
Err, unless things have changed drastically since the last time I checked, writing Windows drivers is anything but closed: you get a copy of the DDK and go to town.
(If you think about it, it's in Microsoft's interest to make it as easy as possible for anyone to write a driver for one of the operating systems; the more hardware supported, the more sales.)
Slashdot Mirror
You know, Gundam is a pretty amazing and important piece of anime, but you don't cite it when you bring up something scientific because it's science-fiction.
Much science-fiction is based on hard science; now, this is a bit old fashioned in the US, but this is one of the many ways the Japanese are fortunate in being "behind the times".
A good amount of the Tomino Gundam Universal Century (UC) future history was lifted straight from current and still valid hard science and engineering theories, aneutronic helium 3 fusion and O'Neil colonies being the most obvious examples. They also pay a lot of attention to details like heat dissipation. (Let's not talk about Minoskivy (sp.) physics or Newtypes, though. :-) This is especially amazing since it's just background that the show itself doesn't bother to explain; it's expected the audience will look up the details if they're interested.
Citing Gundam in this case is like saying something must be true because you read it on Slashdot.
Personal insult like the above also shouldn't have a place on Slashdot.... I cited Gundam UC mostly in humor --- I'm sorry I didn't make that obvious to you --- but also because of the intersection of Japan being a possible site and how remarkable UC is for its attention to scientific and engineering detail, including the critical and relatively obscure answer to D-T neutronic fusion. And perhaps the fact a culture is more serious about getting these details right is worth consideration?
(Although unfortunately pretty much only for refinement; the Japanese system tends to be stifled by seniority, and it has not escaped their notice that every one of their Nobelists had to leave the country to do his research....)Wow! It's not well known history, but North and South Station have never been really connected. There was a "short haul" line to service factories and the like, but ventilation problems back when locomotives were coal fired prevented all but that sort of light traffic from using it.
That link was shut down in 1911 or so as I remember, and the last time I checked moving rail cars from north to south requires running them right through MIT and ~ 12 miles out to Framingham and then back. For cargo, this will probably remain the case (cargo and passenger trains have different requirements based on cargo weight and passenger speed).
It's going to be great for the areas north of Boston if they pull it off (especially with airline trips getting ever more obnoxious). The Boston area has a great commuter rail system (as of a dozen years ago), but the difficulty connecting from North to South Station (you would have to get off a commuter train, get on the Green line, then transfer to the Red to get to South Station and its set of commuter lines) is a severe problem. And if Amtrack can extend service up north....
Or if we ever had some other need to move lots of troops from A to B. One of the men pushing it was President Eisenhower, who before WWII had personally experienced the nightmare of trying to move through our old style highway system, with massive congestion at every small town. Our men who took Europe also noted how useful Hitler's Autobahns were. You want as many movement options as possible, especially land based ones.
Also note that even if you plan on using trains for troop and equipment moving, you must have alternatives in case of sabotage, accidents, etc. Note how difficult it was for the 1AD and 1ID to escape from the lands of the weasels when various railroads were denied to them.
I'll add that I lived in the Boston area for a dozen years before leaving just before this began; I can't comment on the cost et. al., but I can say the benefits as stated are massive. A lot of transportation nightmares have been mitigated.
Indeed, but this project is explicitly designed to be the next "scale up" towards that goal. A design goal of 500MW of fusion power output is nothing to sneeze at....
On the other than, practical fusion is much further away than is advertised, since it requires fusing helium 3, which doesn't produce neutrons, but is a lot harder to fuse. Otherwise your reactor's atoms are slowly transmuted into other (frequently radioactive) elements as it runs. We also have to get a good source of helium 3 ("They're going to strip mine the moon!" the enviros are already whining).
As far as site selection, why not go with the Japanese? After all, back in 1979 Tomino did the first Gundam anime series, and part of its background was small fusion reactors running on helium 3, allowing for a) lots of power, b) big explosions if one can't shut down properly (this is explicit from the beginning), and c) a Jupiter Energy Fleet for the helium 3 (modeled on the petro-rich Arabs --- remember that Japan has to import all of its oil, and I think most of that comes from the Middle East) which is always behind each war, pulling the strings (with the exception of Zeta Gundam where one of them actually showed up; get it when it comes out next year in the US, it's very good)).
BZZT!
This turns out not to be the case, at least for dialup. With the inherent limited bandwidth requirements, their cost was almost entirely up front, in their build out for peak use. It cost them exactly as much to provision a 24x7 user as the archetypical grandmother ... who was online during their peak period(s).
Now, it's a bit different. IP addresses are perhaps more scarce (although that wouldn't seem to be an issue in this case, since there aren't many large bandwidth users), and peak bandwidth demands are obviously much higher, and can be really troublesome to build out (especially for cable).
The answer to this is obviously competition; e.g. telephone xDSL companies know they have a competitive advantage (as a rule) in reliability and keeping up with growth due to the different network designs, and people like me are willing to pay for that.
And obviously some companies realize that going to war with your customers is never good business, as outlined by many others.
Pournelle is pretty sure that Heinlein never desired this book to be published, because he would have otherwise done so when money was tight....
Which is not to say that it isn't worthwhile, especially as a snapshot of the author when he wrote it.
[...] he really does go all Ayn Rand at the end there... is I think an oversimplification; there's major evolution in his writings (that got published) starting in the '60s, and with a couple of exceptions (Mistress, Lazarus Long) not at all for the better, I think.
I agree his best works were from before, especially almost all of the "juveniles" (of which Troopers was the last), but I don't think his basic philosophy changed very much, although it's probably more apparent. I think the quality of his work peaked in the '50s, but the stuff before then is definitely worth checking out. Above all, get a copy of the trade paperback Expanded Universe (published in the early '80s or so); you will learn more about Heinlein and a lot of other things (the E.E. "Doc" Smith micro-biography is priceless) from it than any other single source.
(Pournelle commented perhaps a decade and a half ago on BIX that Heinlein really was a philosopher, but the US simply doesn't have a niche for real philosophers as philosophers in the modern era.)
One thing I find interesting is that I grew up 100 miles south of him, and I often understand exactly where he's coming from; I have the same (apparent) visceral reactions to certain things that RAH did....
Note: cover is that which will stop bullets, shell splinters, etc. I.e. a good earth berm, a ditch, a really strong wall (normal brick/cinder block is not enough, at least not for 7.62 NATO). I wouldn't want to depend on modern (cheap) desks, cubicles, or interior walls or most doors for cover....
Concealment is that which keeps you hidden ... but on the battlefield with riflemen with semi-autos (or machine guns, etc.) it has less utility ... e.g. squads of WWII US riflemen with M1 Garand rifles could often saturate an area of cover behind which they knew the enemy was (and don't forget the BAR men, of course). For a modern illustration, watch e.g. the beginning of The Days of the Condor.
I agree ... such workspaces are to be avoided.... Worst are the ones with good salesmen. who by definition are talking in a way that's hard to avoid listening to. I like them ... if they're in their own office(s). ^_^
Simson exists (hell, he inspired a "Three Minute Hate" at MIT's biggest auditorium (1300 seats) when Berkley Breathed was asking for suggestions on naming a new character (rather frightening to be there, seeing the other sort of mob "action")).
As for his academic creds, he graduated from MIT with a degree in Chemistry and two in Humanities (he told me he could have gotten a third of the latter type with another course or two, but someone told him "enough was enough" :-).
And as previously mentioned/implied, he then graduated from the Columbia School of Journalism.
Try again, or better yet, lose your "MIT envy".
(And for the record, MIT requires exactly two math (calculus) and physics (mechanics, E&M, optics, etc.) courses. Basically the way we understood the world to work as of the end of the 19th century.
Lots better than, say, Harvard, where you just have to prove that you can do algebra....)
Simson has been a friend of mine for many years; after graduating, he went to the Columbia School of Journalism, and while still doing a lot of fun programming make a respectable career in science writing. (rough quote in late '80s, much more modest when he said it (not that he's very modest :-) "I knew I had 'arrived' when I called the President of the AMA and he recognized me by name").
So you can trust him and the article as much as you'd trust any honest journalist (and his editor(s); Simson does not have the final say on what's under his byline).
And he has the tech to back it; e.g. he mastered the 2-4th CD-ROMs in the US (two tests, and then all surviving Ancient Greek, using his own file system, which is *major* prior art in the area). He walks the walk, at a very high level. (If the ISO guys had listened to him, we wouldn't have suffered the horror of the multi-session CD-R transition....)
(Maybe this time I can convince him to buy a home defense firearm, but I doubt he's in much danger, unless "the mob" wants to find his source, and even then they'd ask "politely" before making him an offer he couldn't refuse....)
Well, I guess we can't ignore it.
Why? Novell buying SuSE can be a good thing for this - my college uses [...]. That CAN become NoSE Server/[...]
Emphasis added only in the bolding of "can". which is the operative word as you implicitly acknowledge.
The harsh, brutal fact is that most high-tech mergers and acquisitions fail. I seem to remember one figure of 90%, and that sounds about right; the record is really grim. Only Cisco managed to consistently do it, by virtue of viewing a company acquisition as the best way of hiring good people....
The sad fact is that anyone with a knowledge of high-tech industry history is going to write off SUSE Linux for several years until we see if it's going to be a one out of ten exception....
One other really unfortunate detail is that SUSE has been the "European Linux" up to now; there is perhaps enough anti-American bigotry that this purchase by a company in a notoriously religious state will cause SUSE to loose a lot of its current market share.
(Understand, I assume everyone has the best of intentions, the purchase makes great sense in terms of filling a gaping hole in Novell's offerings and in theory providing SUSE with many good things, etc. etc. ... but that doesn't mean they'll be able to execute it.)
Which is why Red Hat's sustained new attitude of "servers are for the elite" continues to puzzle me.
Some respectable fraction of these CS Majors need/want to work on servers (i.e. the RHEL ES version). They need what it offers, they want to be able to put it on their resume (e.g. "provided 'this useful campus service' using RHEL ES"), etc. etc.
At my school, companies fell over themselves to try to put their products in front of students, who would soon enough be influencing and then making buying decisions.
Once again, something about Red Hat's new business model does not compute ... and SUSE continues to look good on the surface, especially with non-profit discounts (they have to give the same discount to the Federal government, so why not non-profits, which probably represent much lower sales?).
(We'll ignore for the duration of this discussion the horrific raw odds of any tech merger working (i.e. Novell buying SUSE).)
Or use thermal noise....
Last decade the company I worked for wanted some very random numbers, and with a $300 small pod like device that attached to a parallel port and got entropy from (as I remember) a hot diode we got them.
Don't know the current state of the market for these devices, or if they would apply to this architecture, but it can be easily and cheaply done.
Heh.
The "last time I checked" in earnest was many years ago; back then, the question would have been "Does the DDK work with MVC/C++, instead of the MSDN supplied compiler and debugger (NT was not originally developed with MSVC++ :-).
(While I followed the DDK story for a while longer, "in earnest" was a decade ago, when through simple lunch seating luck I arranged with a Microsoft I/O technical lead that the company I worked for would help MS define a driver class for high capacity scanners such as the Kodak Imagelink 900 family (2 pages/second, the sort for which TWAIN et. al. are not a solution). My boss shot this down on the day Clinton was first elected, along with the entire project for us to do our scanner system on NT; pretty obviously he managed to "shoot down" his entire company within a few years....)
Today, I have no idea, and getting a legal copy of the DDK (as opposed to "borrowing" someone's old release from MSDN) is not free. I was just pointing out that device driver development is not closed (and that if you have some home brew hardware I really doubt that Microsoft is going to come after you with a rusty knife if you didn't get a MSDN subscription...).
Also, I have to ask when you're realistically going to have hardware that needs the DDK (which wasn't much if anything more than header files, tools, and documentation, after all), and not be able to afford the software. Can one make a PCI interface circuit board really cheaply? (I don't know, or know if there are e.g. FPGA based ones that you'd then need to do driver work for, or their cost.)
Err, unless things have changed drastically since the last time I checked, writing Windows drivers is anything but closed: you get a copy of the DDK and go to town.
(If you think about it, it's in Microsoft's interest to make it as easy as possible for anyone to write a driver for one of the operating systems; the more hardware supported, the more sales.)