I started exploring Aegis for managing a Linux
kernel tree a few weeks ago, I've found that it has a number of cases where the program deliberately aborts if you want to want to adopt a system administration policy different from the author's. Grep through the source for error messages like "too privileged", "AEGIS_MIN_UID",
"AEGIS_MIN_GID" for examples (there is even code to bomb out if you modify AEGIS_MIN_[U,G]ID in ways that Aegis doesn't like).
Whether Peter Miller's favorite policies are optimal is not the issue to me. For me, an important aspect of free software is that the owner of a computer is given maximum control. If the group maintaining a piece of software is basically trying to wrest that control from the computer owner, then my distrust of that group is enough so that the amount of work I would have to do in studying every line of their code and undoing their logic bombs is exceeds the productivity benefits that I would expect from using the software. For me this attitude problem is the critical issue. If this problem were fixed, I would probably deploy and contribute to Aegis.
For completenes, I'll mention a couple of other issues that other people considering using Aegis might find helpful to know about, although fixing these other issues probably would not tip the balance for me as to whether or not I'd choose Aegis.
Firstly, Aegis has a bit of an unncessary learning curve
because Aegis does not use commands compatible with cvs, sccs or rcs when this is possible. If you invest time in learning Aegis commands, you will not get a return on that investment elsewhere. In comparison, BitKeeper uses rcs and sccs command names and options when possible. I consider Aegis's freeness to be more important than BitKeeper's rcs/sccs command line familiarity, and would also consider incompatible command names and options to be worthwhile if the new syntax were a enough of a user interface improvement, but that doesn't seem to be the case here.
Secondly, I've become skeptical that Aegis really
needs to be a single software package (yes, I know that "cook", Aegis's recommended replacement for "make", is distributed separately from Aegis). I'd like a cvs variant (incompatible if need be) that would suport atomic operations, symlinks, inode information, renaming files, and maybe some distributed development features. I think Aegis's repository control based on regression testing is a great idea, but I also think that it could be implemented more simply as a separate package that used some kind of cvs successor (or, ideally, was retargetable to a number of software repository types).
I think Aegis has some great ideas, but I currently think it will be less work, especially in the long run, to find or make something that I like better.
Can you please try running your compilation benchmark with the BIOS attempting to turn the cache on and the BIOS attempting to turn the cache on, and verify that the compilation times match?
Please try doing the compilation a few times each way, so you can have some sense of the variation
not due to this change. That would help confirm whether the cache is indeed off in both cases.
Your page is too interesting to passingly comment on in an aging slashdot forum - please send me your email
address and I'll take some time and email you a thought-out response.
When you finish reading that.txt file from my
personal web, area, you'll see that my.signature, including email address, is at the bottom of it.
There was a minor error in the graphs in that file. "max price", should be in the lower
right rather than the upper left. I've corrected
those graphs if you want to download the file again. The rest of the file is unchanged.
Also note that I'll be gone Feb. 12-19 and may
not have email connectivity during that time,
so I apologize in advance if I'm unable to answer
during that time.
Actually, the data disagrees with you here. When the minimum wage goes up, the economy tends to get better.
For obvious reasons, too. Think about it.
I'm glad to hear that you have data, can recall a source for your data (in as much or as little detail as you can remember)?
Your use of the term "do better" could mean different things (for example, it's possible for the average wage of employed people to go up, while more people are unemployed because business that are not viable at a given minimum wage shut down or move). So, it's a litle difficult for me to see the "obvious reasons." For example, what would happen if we raised the minimum wage to $200/hour? A few jobs would exist, but mostly people would be wards of the state (or starve or turn to crime if the state could not afford it),
at least until $200/hour only bought what minimum wage buys today. Basically, the minimum wage makes capitalism illegal among people whose productivity is below that amount.
After spending several weeks in China (no, I'm not a fan of their authoratarian system), I've come to appreciate how people of much lower incomes start (entreprenuerial) businesses as readily as people of higher incomes in the US and how abundant jobs
become and how readily people change jobs when there is not a minimum wage impeding the creation of arbitrarily small businesses. It is true that there other elements to the trade-off. For example, in the US we seem to think that is better that people freeze to death in the streets than allow the building of cheaper homes (in many cases you're not allowed to even build homes with fewer than two independent sources of electricity in the kitchen) or most tricycle cars (which are widely used in China) when we still allow bicycles on the street, etc. There is a dignified capitalist economy by which people climb out of povery, but we've outlawed so much of it in the United States.
Anyhow, I will be interested in seeing the data that you cite.
I don't know if that is the case for California, but I vaguely recall that labor unions got a
federal "prevailing wage" law that basically says that for many jobs the federal government is not allowed to try to find the best deal, but instead has to pay the "prevaling wage", the calculation of which is probably a big political football. I believe that such policies are inflationary and potentially increase unemployment in cases where more would be done if labor costs were lower.
This is pure speculation, but it would not surprise me if California now has a similar
"prevailing wage" policy, given how pro-union Gray Davis is. I remember listening to NPR about a year after
Gray Davis first elected about how we was working
with the legislature to outlaw Price Club and CostCo becaue they used non-union labor. Davis raised the minimum wage, which I think contributes to unemployment and a higher cost of living.
(For example, with no minimum wage, perhaps it would be economical to recycle computers.) Davis also changed the law in California requiring paying 1.5X wages for any work done past eight hours in a day, which I think incents many companies to export jobs or try harder to reclassify many jobs as "exempt."
More recently the dock workers went on strike to protest computerization even after their jobs were guaranteed, and, on election day, the teamsters decided to picket a concert hall to impede delivery by non-union trucks to a facility that has never used union labor (i.e., using the implicit threat of violence to prevent scare off
non-union drivers from making deliveries, what a great job interview technique!), and, of course, nobody was talking about providing police protection for non-union workers in either case
and everyone was talking about how their might be violence. Anyhow, as you can imagine, it is rather difficult to argue that California is objectively the most profitable place to locate
jobs these days.
Sorry for ranting. I know I haven't cited any references in this posting. Feel free to correct my many errors and speculations. My main point is simply that I believe that Davis's policies, not just external factors, have made it a more profitable business decision to locate jobs elsewhere, downsize do both.
The benefits of hyperthreading are a function
of how many cycles would otherwise be wasted by
the CPU as it waits on a cache miss. Systems using regular Synchronous DRAM have more of these lost cycles than systems using Double Data Rate or Rambus. So, I think hyperthreading should narrow the performance gap between regular SDRAM versus DDR or Rambus.
Also, with DDR or Rambus costing nearly triple
what SDRAM costs, I wonder if some enterprising
company will develop a chipset that can interleave access to two SDRAM DIMM's for performance similar to DDR. Even if the two DIMM's have to be on
completely separate electrical busses, I would think that it would result in a lower total cost
for most popular combinations of performance levels and memory capacity.
Could someone please explain to me the benefit of delivering the food as a skin patch instead of as little food pills or something. For insulin and nicotine, I can understand the need for a controlled continuous release, but why the need for such precision with food?
The next generation of PCI, PCI Express (formerly 3GIO) features scalable bus width. The thinnest version is a single 250 megabyte per second "lane", which sounds like it could potentially replace USB, Firewire and Serial ATA with devices that are directly mapped into processor memory and IO space in a manner that is a bit reminiscent of the origin of IDE drives.
What does the "law of supply and demand" mean when you have more workers than jobs? It means that people in the US will starve.
Jobs are not a fixed quantity. Economic prescriptions
that treat a dependent variable as a fixed constant generally fail to produce their intended results.
I should have added that, yes, I know everyone
already has electricity, but I'm thinking in
terms of whatever repeaters and other infrastructure needs to be deployed. In a more densely populated area, I'd expect you'd need
fewer of these, have less distance for repair
trucks to travel on average, etc.
I don't know if the Japanese also use the term
"Feng Shui" to mean the same thing, but Feng
Shui is definitely a big thing in China. A
couple of people have told me that real estate
is cheaper in south Beijing partly because the
area is considered to have inferior Feng Shui.
Feng Shui are two Mandarin Chinese words. Feng ("fung") means wind and Shui ("shway") means water.
Do you have to invent a test for "This piece of code should be more elegant", or what?
I haven't actually used Aegis, but I assume that
since the submitter also submits a test case,
the submitter could "invent a test" as you put
it that roughly corresponded to the way the new
code is more "elegant."
I don't know if Aegis has this, but I'd like to
see "benchmarks" in addition to "test cases." A test case basically either passes or fails, but a benchmark would return a score, so one could then
submit a change that did not break anything but
resulted in smaller line count, fewer branches, smaller object size, faster exeution, whatever without breaking other test cases or
worsening other benchmarks. You'd probably also
need to define a utility function that would
combine benchmark scores to determine which combination
was better, and I expect people would frequently
submit changes or even branches to that utility function as their thinking evolved on how they'd
like to optimize various trade-offs.
As I mentioned in replying to a previous
slashdot story, Covad welcomed line
sharing the last time I checked, including wireless,
including for-profit. This was the big selling point that the Covad sales rep used to convince
me to sign up, and I described in detail the
idea of reselling my covad connection over an
access point just to make sure there was no
misunderstanding.
In general, DSL providers seem to be less worried
about usage patterns than cable modem providers,
probably because having separate lines from the DSL
Access Multiplexer to each subscriber gives them a more reliable ability to
throttle individual customers if necessary,
as opposed to having a shared local loop in
the case of cable modems.
The web page implies that gforge requires
php4, which, last I checked, contained and
required the
GPL-incompatible zend "engine." Does anyone
know how much work would be involved in porting
gforge to (GPL compatible) php3?
I say we should fix copyright law so that it only works for seven years. After
those seven years we can use the source code of the program.
(I am not the anonymous poster who wrote that.)
Yes, I think that is exactly what we need. I've
believed we should have almost exactly that change
for years. I think seven years is a period of time
long enough to be reasonably profitable (contrary to what another respondent claimed). Go to a surplus store and compare the selling price of seven year old software with recently released software. Nevertheless, everyone I know who
buys software at all buys the latest versions of software in spite of the price difference. If you
update in a timely manner, your brand of software
should be profitable indifinitely this way.
Also, publishing cycle times have shrunk both
for physical packaging and, of course, by the addition of distribution over the internet.
If
your software solves such a fixed and narrowly defined problem that there really is nothing to update, then it's the sort of software that would
be cloned after about seven years anyway. Also, if people know that your software is going to be released in seven years, it may actually discourage cloning.
Under your proposal, right now the source code
to Windows 95 would have just been released and
I imagine people would be starting to beta free binary distributions of it.
I think this CPU is built with low-power consumption as its primary goal and performance second
I'm not semiconductor person, so my understanding
may be a bit confused, but I believe that the
CMOS technology used for any microprocessor built
in the last decade or two has power consumption
approximately proportional to its clock speed.
If your 2GHz CPU consumes 50W, it will consume
about 25W at 1GHz executing the same instructions (i.e.,
the power consumption due to current leakage
is negligible). If you only cared about
power consumption, you could just underclock.
The objective is improving the
MIPS/watt ratio (or rather MI/joule if
you divide out the time units) at a point
that still provides enough MIPS to make a
product that people want.
The case was settled before it reached trial. It was settled in part because of the opinion of the judge ruling of
the judge in the preliminary injunction stage. The judge found no harm at all, much less irreparable harm,
after Nusphere released the source.
The question of "no harm at all" would not have
been before the court for a preliminary injunction, only irreparability.
I could believe
that the judge might have determined that there
was no irreperable harm from the copyright
component and then might have decided for the purposes of the preliminary injunction hearing only to accept arguments about the irreparability of the
trademark infringement (and I think it would be
a lot easier to argue that trademark infringement
is irreperable). But if you claim the judge ruled
about something I don't think could have been considered to be before the court at that hearing,
then you I'd need to see specific references to be convinced that this isn't a case of your recollection being skewed.
I'm not a lawyer, so don't take this a legal advice.
There was a hearing for a preliminary injunction
in February 2002 where the question was whether Nusphere's use
of the code between then and the trial would
cause irreperable harm (i.e., so that
just assessing monetary damages at the end of the trial would not be good enough and an injunction was required before that). If the trial has actually occurred, as would have to be the case if the court truely ruled that Nusphere could remedy its breach by releasing all of its modified source, then could you post a reference for the actual trial? I see lots of documents about the preliminary hearing on google, but none about the
trial having occurred.
I am not a lawyer. Do not use this as legal advice.
Agreed. It wastes a lot of people's time when editors or story authors (in the absense of
competent editors) try to show
how "with it" they are by not expanding acronyms, especially when they have not recently been used in a slashot headline.
because the path the laser travels is so much smaller per RPM, the RPMs would have to be
porportionally faster to make up for it.
I think you're right, but linear bit density
is also a factor.
If the limiting factor is the acceleration experienced at the
edge of the disk, then
a = v**2/r
v = sqrt(a*r).
So, the maximum velocity that can be achieved
underneath the read head still
decreases as we make the media smaller,
as you correctly observed. By this reasoning,
it looks like a SFFO would have half the maximum
media velocity of DVD:
However, with SFFO, the bit density has been increased.
These discs are about 1/16th the
area of a DVD (1/4th the diameter: 12cm vs. 3cm), and 1/5th the capcity (4.7GB vs. 1GB), so they
have about triple the areal density (bits per unit
area), at least if
we assume that the unusable areas in the
center and outer margins will be proportional
to disc area.
If the density increase has
come equally from shrinking the distance between
bits on the track and shrinking the distance
between tracks (i.e., the aspect ratio of the
bits remains the same), then the change in
linear density of bits along a single track
will be proportional to the square root of
the change in areal density. In other words,
the bits are probably closer together by
1/sqrt(3). So, labelling the
density of adjacent bits within one track as
"linear_density", we get:
Other factors that may also determine how fast the disc can be spun are when the disc media starts to ripple and buckle, which I believe is helped by media thickness and hindered by media
diameter (SFFO is smaller but thinner), and
frictional and aerodynamic forces, which are
portional to v or v**2 respectively,
which would favor spinner small media faster.
Serial ATA is one drive per cable, which reduces
the cost of going at 150MB/second and safely supporting hot plugging (the analog properties of the cable are better, nobody will add a drive to your current cable, so no need to protect the drive from that, no need to build drives with jumpers, reduced support costs from users calling
in with misconfigured jumpers, etc.). The cables are thin and the connectors or small, so having
a lot of them is not as big a deal as with IDE ribbon cables. I expect that
once Serial ATA drives become common, you'll see
motherboards with four or eight SATA connectors.
If you want to connect a bunch of drives on a
common fast serial connection, there is already a plethora of options, all of which basically serialize SCSI commands: FireWire, Universal Serial Bus 2.0, Fibre Channel, Serial SCSI Architecture (SSA), InfiniBand, and iSCSI.
Slashdot Mirror
Whether Peter Miller's favorite policies are optimal is not the issue to me. For me, an important aspect of free software is that the owner of a computer is given maximum control. If the group maintaining a piece of software is basically trying to wrest that control from the computer owner, then my distrust of that group is enough so that the amount of work I would have to do in studying every line of their code and undoing their logic bombs is exceeds the productivity benefits that I would expect from using the software. For me this attitude problem is the critical issue. If this problem were fixed, I would probably deploy and contribute to Aegis.
For completenes, I'll mention a couple of other issues that other people considering using Aegis might find helpful to know about, although fixing these other issues probably would not tip the balance for me as to whether or not I'd choose Aegis.
Firstly, Aegis has a bit of an unncessary learning curve because Aegis does not use commands compatible with cvs, sccs or rcs when this is possible. If you invest time in learning Aegis commands, you will not get a return on that investment elsewhere. In comparison, BitKeeper uses rcs and sccs command names and options when possible. I consider Aegis's freeness to be more important than BitKeeper's rcs/sccs command line familiarity, and would also consider incompatible command names and options to be worthwhile if the new syntax were a enough of a user interface improvement, but that doesn't seem to be the case here.
Secondly, I've become skeptical that Aegis really needs to be a single software package (yes, I know that "cook", Aegis's recommended replacement for "make", is distributed separately from Aegis). I'd like a cvs variant (incompatible if need be) that would suport atomic operations, symlinks, inode information, renaming files, and maybe some distributed development features. I think Aegis's repository control based on regression testing is a great idea, but I also think that it could be implemented more simply as a separate package that used some kind of cvs successor (or, ideally, was retargetable to a number of software repository types).
I think Aegis has some great ideas, but I currently think it will be less work, especially in the long run, to find or make something that I like better.
Can you please try running your compilation benchmark with the BIOS attempting to turn the cache on and the BIOS attempting to turn the cache on, and verify that the compilation times match? Please try doing the compilation a few times each way, so you can have some sense of the variation not due to this change. That would help confirm whether the cache is indeed off in both cases.
When you finish reading that .txt file from my
personal web, area, you'll see that my .signature, including email address, is at the bottom of it.
There was a minor error in the graphs in that file. "max price", should be in the lower right rather than the upper left. I've corrected those graphs if you want to download the file again. The rest of the file is unchanged.
Also note that I'll be gone Feb. 12-19 and may not have email connectivity during that time, so I apologize in advance if I'm unable to answer during that time.
The slashdot lameness filter does not like the ascii graphs are essential to my response, so I've put my response in a text web page.
I'm glad to hear that you have data, can recall a source for your data (in as much or as little detail as you can remember)?
Your use of the term "do better" could mean different things (for example, it's possible for the average wage of employed people to go up, while more people are unemployed because business that are not viable at a given minimum wage shut down or move). So, it's a litle difficult for me to see the "obvious reasons." For example, what would happen if we raised the minimum wage to $200/hour? A few jobs would exist, but mostly people would be wards of the state (or starve or turn to crime if the state could not afford it), at least until $200/hour only bought what minimum wage buys today. Basically, the minimum wage makes capitalism illegal among people whose productivity is below that amount.
After spending several weeks in China (no, I'm not a fan of their authoratarian system), I've come to appreciate how people of much lower incomes start (entreprenuerial) businesses as readily as people of higher incomes in the US and how abundant jobs become and how readily people change jobs when there is not a minimum wage impeding the creation of arbitrarily small businesses. It is true that there other elements to the trade-off. For example, in the US we seem to think that is better that people freeze to death in the streets than allow the building of cheaper homes (in many cases you're not allowed to even build homes with fewer than two independent sources of electricity in the kitchen) or most tricycle cars (which are widely used in China) when we still allow bicycles on the street, etc. There is a dignified capitalist economy by which people climb out of povery, but we've outlawed so much of it in the United States.
Anyhow, I will be interested in seeing the data that you cite.
This is pure speculation, but it would not surprise me if California now has a similar "prevailing wage" policy, given how pro-union Gray Davis is. I remember listening to NPR about a year after Gray Davis first elected about how we was working with the legislature to outlaw Price Club and CostCo becaue they used non-union labor. Davis raised the minimum wage, which I think contributes to unemployment and a higher cost of living. (For example, with no minimum wage, perhaps it would be economical to recycle computers.) Davis also changed the law in California requiring paying 1.5X wages for any work done past eight hours in a day, which I think incents many companies to export jobs or try harder to reclassify many jobs as "exempt." More recently the dock workers went on strike to protest computerization even after their jobs were guaranteed, and, on election day, the teamsters decided to picket a concert hall to impede delivery by non-union trucks to a facility that has never used union labor (i.e., using the implicit threat of violence to prevent scare off non-union drivers from making deliveries, what a great job interview technique!), and, of course, nobody was talking about providing police protection for non-union workers in either case and everyone was talking about how their might be violence. Anyhow, as you can imagine, it is rather difficult to argue that California is objectively the most profitable place to locate jobs these days.
Sorry for ranting. I know I haven't cited any references in this posting. Feel free to correct my many errors and speculations. My main point is simply that I believe that Davis's policies, not just external factors, have made it a more profitable business decision to locate jobs elsewhere, downsize do both.
Also, with DDR or Rambus costing nearly triple what SDRAM costs, I wonder if some enterprising company will develop a chipset that can interleave access to two SDRAM DIMM's for performance similar to DDR. Even if the two DIMM's have to be on completely separate electrical busses, I would think that it would result in a lower total cost for most popular combinations of performance levels and memory capacity.
Could someone please explain to me the benefit of delivering the food as a skin patch instead of as little food pills or something. For insulin and nicotine, I can understand the need for a controlled continuous release, but why the need for such precision with food?
The next generation of PCI, PCI Express (formerly 3GIO) features scalable bus width. The thinnest version is a single 250 megabyte per second "lane", which sounds like it could potentially replace USB, Firewire and Serial ATA with devices that are directly mapped into processor memory and IO space in a manner that is a bit reminiscent of the origin of IDE drives.
Jobs are not a fixed quantity. Economic prescriptions that treat a dependent variable as a fixed constant generally fail to produce their intended results.
I should have added that, yes, I know everyone already has electricity, but I'm thinking in terms of whatever repeaters and other infrastructure needs to be deployed. In a more densely populated area, I'd expect you'd need fewer of these, have less distance for repair trucks to travel on average, etc.
Just a guess, but I would expect most networking services to become economical in areas with the highest population densities first.
Feng Shui are two Mandarin Chinese words. Feng ("fung") means wind and Shui ("shway") means water.
I haven't actually used Aegis, but I assume that since the submitter also submits a test case, the submitter could "invent a test" as you put it that roughly corresponded to the way the new code is more "elegant."
I don't know if Aegis has this, but I'd like to see "benchmarks" in addition to "test cases." A test case basically either passes or fails, but a benchmark would return a score, so one could then submit a change that did not break anything but resulted in smaller line count, fewer branches, smaller object size, faster exeution, whatever without breaking other test cases or worsening other benchmarks. You'd probably also need to define a utility function that would combine benchmark scores to determine which combination was better, and I expect people would frequently submit changes or even branches to that utility function as their thinking evolved on how they'd like to optimize various trade-offs.
In general, DSL providers seem to be less worried about usage patterns than cable modem providers, probably because having separate lines from the DSL Access Multiplexer to each subscriber gives them a more reliable ability to throttle individual customers if necessary, as opposed to having a shared local loop in the case of cable modems.
The interview link is down right now, so please forgive me if this is already discussed there.
The web page implies that gforge requires php4, which, last I checked, contained and required the GPL-incompatible zend "engine." Does anyone know how much work would be involved in porting gforge to (GPL compatible) php3?
(I am not the anonymous poster who wrote that.)
Yes, I think that is exactly what we need. I've believed we should have almost exactly that change for years. I think seven years is a period of time long enough to be reasonably profitable (contrary to what another respondent claimed). Go to a surplus store and compare the selling price of seven year old software with recently released software. Nevertheless, everyone I know who buys software at all buys the latest versions of software in spite of the price difference. If you update in a timely manner, your brand of software should be profitable indifinitely this way. Also, publishing cycle times have shrunk both for physical packaging and, of course, by the addition of distribution over the internet.
If your software solves such a fixed and narrowly defined problem that there really is nothing to update, then it's the sort of software that would be cloned after about seven years anyway. Also, if people know that your software is going to be released in seven years, it may actually discourage cloning.
Under your proposal, right now the source code to Windows 95 would have just been released and I imagine people would be starting to beta free binary distributions of it.
I'm not semiconductor person, so my understanding may be a bit confused, but I believe that the CMOS technology used for any microprocessor built in the last decade or two has power consumption approximately proportional to its clock speed. If your 2GHz CPU consumes 50W, it will consume about 25W at 1GHz executing the same instructions (i.e., the power consumption due to current leakage is negligible). If you only cared about power consumption, you could just underclock.
The objective is improving the MIPS/watt ratio (or rather MI/joule if you divide out the time units) at a point that still provides enough MIPS to make a product that people want.
The question of "no harm at all" would not have been before the court for a preliminary injunction, only irreparability.
I could believe that the judge might have determined that there was no irreperable harm from the copyright component and then might have decided for the purposes of the preliminary injunction hearing only to accept arguments about the irreparability of the trademark infringement (and I think it would be a lot easier to argue that trademark infringement is irreperable). But if you claim the judge ruled about something I don't think could have been considered to be before the court at that hearing, then you I'd need to see specific references to be convinced that this isn't a case of your recollection being skewed.
I'm not a lawyer, so don't take this a legal advice.
I am not a lawyer. Do not use this as legal advice.
Agreed. It wastes a lot of people's time when editors or story authors (in the absense of competent editors) try to show how "with it" they are by not expanding acronyms, especially when they have not recently been used in a slashot headline.
I think aerial is short for aerial antenna, meaning an antenna that is in the air.
I think you're right, but linear bit density is also a factor. If the limiting factor is the acceleration experienced at the edge of the disk, then
So, the maximum velocity that can be achieved underneath the read head still decreases as we make the media smaller, as you correctly observed. By this reasoning, it looks like a SFFO would have half the maximum media velocity of DVD:However, with SFFO, the bit density has been increased. These discs are about 1/16th the area of a DVD (1/4th the diameter: 12cm vs. 3cm), and 1/5th the capcity (4.7GB vs. 1GB), so they have about triple the areal density (bits per unit area), at least if we assume that the unusable areas in the center and outer margins will be proportional to disc area.
If the density increase has come equally from shrinking the distance between bits on the track and shrinking the distance between tracks (i.e., the aspect ratio of the bits remains the same), then the change in linear density of bits along a single track will be proportional to the square root of the change in areal density. In other words, the bits are probably closer together by 1/sqrt(3). So, labelling the density of adjacent bits within one track as "linear_density", we get:Other factors that may also determine how fast the disc can be spun are when the disc media starts to ripple and buckle, which I believe is helped by media thickness and hindered by media diameter (SFFO is smaller but thinner), and frictional and aerodynamic forces, which are portional to v or v**2 respectively, which would favor spinner small media faster.
If you want to connect a bunch of drives on a common fast serial connection, there is already a plethora of options, all of which basically serialize SCSI commands: FireWire, Universal Serial Bus 2.0, Fibre Channel, Serial SCSI Architecture (SSA), InfiniBand, and iSCSI.