Certainly there are problems with this definition of complexity.
Where to set the level zero player? "Human novice" is very fuzzy, but "random" could give many levels of 'complexity' within the 'idiotic playing' realm.
Where to set the top level player? "Best human" makes the complexity depend on how much effort we've put into it. Comparing to "perfect play" is usually computationally infeasible.
I'm a poor go player, and I've never tried the 3D Go in a cube, so what I say may be completely off base.
The problem is that in cubical go, you can't make walls, eyes, or captures.
In standard go, adding a stone to a group (but not connecting to another group) increases the group's number of liberties by at most two. In cubical go, it can increase by four. This means it is just too hard to restrict the number of liberties a group has.
If a player is trying to join two groups in standard go, and the opponent places a stone between them, the player has two ways around that stone (left or right). The opponent might be able to block both of these. In cubical go, they have four ways around - it will be very hard to stop them connecting.
An eye requires many more stones in cubical go. The opponent has many chances to disrupt it. Especially because of the above point, which means that stones placed deep in your territory can very likely still live.
I think you'll just end up with lots of intertwining filamentary structures until the board becomes so crowded that captures become possible, then the first person to get a capture of a sizable group will win.
This idea is original (I haven't copied it from anywhere) but it has probably been independently thought of before.
A natural thing to ask about a 2D game such as go, chess, naughts-and-crosses/tic-tac-toe is how can it be extended to three dimensions?
For go, the immediate thought is to play on a 3D grid of cubes. However, I believe (I haven't actually tried) that this will work very poorly. The number of neighbours that each point has will have a huge effect on play. With two many neighbours (6 in the case of a cubic grid) it will be impossible to defend territory, or to capture stones until nearly the entire board fills up. The game would be boring.
To capture the essence of go in 3D, we need each point to have four neighbours. This can be achieved with maximal symmetry by using either truncated octahedra or rhombic dodecahedra as the board's cells.
UPDATE: After writing the above, I did a quick search and discovered diamond go which has clearly used the same basic idea. I haven't figured out yet whether it is equivalent to one of my two options.
Game complexity
on
Cracking Go
·
· Score: 5, Interesting
On complexity: perhaps the best definition (for matching our intuitive feeling for how complex a game is) I've come across for game complexity is this: Imagine a novice player who avoids glaringly wrong moves (e.g. sacrificing a queen for no benefit in chess) but otherwise randomly choses from the legal moves available to them. Call this a 'level 0' player. Now a player who is smart enough to beat the level zero player 2/3 of the time is a level 1 player, someone who beats a level 1 player 2/3 of the time is level 2 etc. Then the game's complexity is the level of the best human player. If I recall correctly, this gives naughts-and-crosses/tic-tac-toe a complexity of 1, chess was about 20 and go was about 40. Unfortunately, this is all from memory, I haven't found the right magic words to feed Google to get more information.)
So yes, the game-state complexity can be a very poor measure of actual complexity, but go still has a very strong claim to being the most complex common human game. (An extreme example of high game-state complexity but trivial actual complexity would be a large game of "Brussel Sprouts".
On the Diplomacy/Axis and Allies comparison. The three games being compared are very different in nature (although they are all zero-sum.)
Go is two player, complete information, deterministic (as are chess, checkers, hex, naughts-and-crosses/tic-tac-toe) Such a game can, in theory, be completely analysed.
Diplomacy is multi-player, deterministic, incomplete information because moves are simultaneous. I'm not sure if or how this could be analysed.
Axis and Allies is (effectively) two player, probabilistic. It could (theoretically) be completely analysed as a very large dynamic programming problem. From a game-theoretical point of view, it resembles Yahtzee.
If even 10% of these consumers take up the Linux option, that would translate into a doubling of desktop Linux users, in effect, hastening the onset of an inflection point.
And yes, while more people know Windows, there are tens of millions now who also know Linux. Linux is fast approaching that first inflection-point. A move by regulators to ensure that there is a breathing-space for competition will likely see that inflection-point come sooner than later.
You use that word a lot. I don't think it means what you think it means.
It indicates a point where the second derivative of a function is zero - less technically, where the curve is temporarily straight, or the transition point between acceleration and deceleration (either way). The author seems to think it is a good thing. I can only make sense of this if the author thinks that Linux adoption is currently decelerating, and so passing an inflection point will move into accelerating adoption.
The following is from memory from "Technology in World Civilization" by Arnold Pacey, which is a very readable book.
The Mongols were very influential in transfering technology between east and west. They gathered lots of scholars from their empire into their capital. An effect of this was that (for example) Chinese and and Turkish scholars could learn from each other. They would later go home with the new knowledge.
It is often the case that data analysis and interpretation for particular aspects of a research project (like 1-2 figures in a 7 figure paper) are up for vigorous debate.
A professor I work with commented to me about a paper he'd been asked to review. In his review, he said something like "The data analysis is completely wrong, but the data set is wonderful. The paper should be published - then somebody else will reanalyse it correctly."
That matches my IANAL understanding of the discussions at groklaw.
There are various classes of creditors in a bankrupcy. The highest priority class gets paid first, then the next etc. If there is enough money to pay only a fraction of the claims in a given priority class, everyone gets the same proportion of money owed.
The highest priority classes are paying the bankrupcy lawyers and paying back fraudulently acquired wealth. (I'm not sure which order these two are in.) As Novell's claim is in the 'fraudulent' class, they get paid first. If (as is likely) the amount owed is greater than SCO's total assets, Novell gets everything - all the assets and none of the liabilities of SCO.
I don't know whether the assets are just given to Novell, or if they are sold off and Novell gets the proceeds. I suspect they'll be sold - it is the only real way to determine the value of assets such as the Unix IP.
If they are sold off, Novell is likely to be in the bidding, and is likely to win the bidding. Whatever they bid isn't 'real money' to them - it comes straight back.* All they stand to lose is whatever amount of money somebody else was willing to pay for the Unix IP. Even if such a buyer exists, they would probably prefer to buy it from Novell after the liquidation, rather than as part of the SCO liquidation sale. This allows them to renegotiate what gets transfered, rather than having to follow the chain of old deals Novell -> old Sco -> Caldera/new SCO -> hypothetical buyer.
* Exception: if Novell bids so much that the SCO assets exceed what SCO owes them, then they'll pay some real money.
I expect the IBM case will pretty much evaporate. Novell would be successor-in-interest to SCO, and might have the option to keep fighting it, but obviously won't. I'm guessing (IASNAL) that if they did chose to keep fighting, they'd also have to accept some amount of liability for IBM's lawyer's fees if (when) they lost.
Note that I said "if your data format is clever enough". Evidently you know enough to say that the current formats are not clever enough. I'm happy to accept the current situation as falling into "not feasible" (at present.) From an information-theory point of view, it must be possible to use the information in the DVD data to reduce the storage requirement for the rest.
These are 40Gb disks, and a DVD is just under 4Gb, hence the 10% figure. Whether this is acceptable will differ according to who you are.
Stereo records played on mono turntables. Colour TV broadcasts displayed on B&W sets. Such compatibility is good, so long as you don't pay too much for it, hobbling the new format to the needs of the old.
So long as the manufacturing price of disks is low, this would allow DVD/VMD disks to be sold for the same price as a DVD. Then when you're buying DVD/VMD disks to play in your DVD player, you can later upgrade to a VMD player and instantly have a collection of HD content. This would appeal to consumers and the VMD drive manufacturers. The content sellers might want to charge more for HD, which could sink the proposition. (It relies on DVD/VMD combo disks costing the same as plain DVD.)
You lose about 10% of the disk capacity for DVD data which isn't used by the VMD playback. You might not even lose that 10% - if your data format is clever enough, you can use that 10% within the VMD playback rather than redundantly repeating it in the non-VMD data. I'm sure this is possible, but I'm not sure it is computationally feasible.
Coming late to the game, they need to establish a 'home ground', a niche that they can dominate and then grow out from. India could be that home ground.
A fine feature would be if it were possible to play the new HD VMD disks at DVD resolution on standard DVD players. Given they use the same lasers, it might be that DVD players will see one particular layer, on which the DVD data could be stored. This again would help greatly to break into the market.
However, they don't mention such a feature, and I'd hope they'd have thought about it, so probably it is technically infeasible.
You seem to be thinking about analyses where there are only two or three species included. In that circumstance, you can't test the clock, and have the problems you talk about.
Instead we'd do an analysis with say 12 primates. This is how we get the "sufficient data points" you worry about. (Some analyses have hundreds of species, but it becomes computationally difficult to reliably construct the trees in this circumstance.) Typically we have one or more species (the outgroup) which are known a priori to be more distantly related than the rest. In this case, we might use new world monkeys or some non-primates as the outgroup. Where the outgroup attaches to the rest (the ingroup) establishes the 'root' of the tree for the ingroup.
Now, within the ingroup, we look at the genetic distance from the present species back to the root. If all is well, this will be highly consistent (say all within 10% of each other.) Given that our ingroup may contain species which have diverged (say) over 30 million years ago, we have evidence that the clock is stable (within 10%) over periods of 30 million years. Therefore we can rely on it for the 5 million year human-chimp split.
It is always necessary to callibrate the clock against one or more well-established divergence times from the geological or fossil record. This might (for example) be the divergence between old and new world monkeys.
There are also "relaxed clock" analyses which allow the clock to vary over the tree, but only slowly. This is necessary to date deep divergences (e.g. amphibians vs other land vertebrates) as no genes will have maintained a steady clock over that time scale.
The human-chimp date will be based on many hundreds of genes. Less well studied species may have dates based on just a handful of genes.
We can get ancient DNA from suitable samples back to about 50,000 years. However, it would be hazardous to extrapolate to millions of years from this data. Most molecular clock analyses use contemporary data only.
This find perhaps sets back the date of chimp-gorilla split but not "human-ape".
The genetic human-gorilla divergence is only slightly larger than the human-chimp divergence. If we indeed had firm evidence that humans and gorillas diverged (say) 12 million years ago, then we would use this as a callibration point for our molecular clock, and it would put the human-chimp divergence back to about 10 million years ago. These dates are not independent.
Prior to [the KT boundary] the mammal line had already split much more than we previously gave it credit for, a lot of the main groups were developed.
Though, it's even weak evidence that it's actually a gorilla just that it had a diet like that of a gorilla.
These I agree with, although I don't have the expert knowledge for the second point, so I take my own belief with grains of salt.
IAA evolutionary molecular biologist*. There are some significant incorrect assumptions in your post.
First, the rate of genetic change used to date divergences is independent of the rate of morphological change. That cockroaches aren't changing much outwardly has no effect on the genetic drift that lies behind the 'molecular clock'. Some DNA is 'neutral' (changes have no effect on the fitness of the organism.) In genes for proteins, there will be many 'equally good' (so close to equal that evolution can't distinguish) solutions, and over time a species solution will 'drift' through this cloud of optimal solutions. These processes are independent of morphological change.
Second, we very well understand that the molecular clock is imperfect. It runs at different rates in different genes. In some genes, the rate is highly variable. In some lineages, there is an acceleration of rates. (E.g. rodents have a high rate compared to primates.) We look for these problems and correct for them or work around them.
Typically we draw phylogenies (family trees for species) with branch lengths proportional to genetic distance. (I.e. time, as measured by a putative molecular clock.) We know that, if we were able to plot the phylogeny measured by a real clock, the tips (current species) would all lie at the same time (the present.) If our phylogeny has its many tips nearly aligned, then we believe we have a good molecular clock, and we can estimate how good it is. If (as not infrequently happens) a particular gene or group of genes give a tree with poor tip alignment, we conclude there is no good molecular clock for this gene/group of genes, and we don't use it for dating divergences. If we see a lineage which systematically has branches too long or too short, we infer a change in the clock rate on that lineage, and take care to account for this when dating divergences within that lineage.
In short: We don't blindly assume the existence of a molecular clock. We hope for a clock, and look for the genes which act most clock-like, and we calibrate both the rate *and* the accuracy of the clock.
Returning to the case in hand: I expect that this is a case of convergent evolution. A 10 million years ago ape, not an ancestor to humans or gorillas, evolved gorilla-like teeth in response to similar evolutionary pressures. However, I don't have the knowledge (in paleontology and morphology) to assess the claims directly. I have reason to believe the molecular divergence date couldn't be so badly wrong, and the evidence against it seems thin to me.
* More accurately, I'm an astronomer pretending to be a mathematician pretending to be an evolutionary molecular biologist.
If we go back about 20 years, hard drives were for non-volatile fast-access storage, and tape drives were for backup, bulk data storage, archiving and sometimes data transfer (when there was too much for floppies.)
Now that flash is reaching the point where we can contemplate using it for the primary non-volatile storage niche, we may see hard drives being displaced into the backup/bulk storage/archiving niches. If so, expect to see increasing emphasis on ways to hot-plug hard drives into your computer, and increasing emphasis on price/GB and decreasing emphasis on performance and possibly per-drive capacity.
We'll really know we've reached this point when hard drives are used as a medium for delivering software.
A very few brilliant men have been called crackpots by their contemporaries but have ended up being exonerated by history.
The only example that springs to mind is continental drift, and now that I look it up in the Infallable Font Of All Knowledge I see that there were about half a dozen people pushing that. I'm sure there are other examples - perhaps even a few dozen.
By contrast, many thousands (probably millions) have been called crackpots by their contemporaries and ended proven so by history, or simply forgotten.
Minor correction: it is a red giant, not a red supergiant. Supergiants are very massive stars (on the order of 10 times the mass of the sun or more) which will eventually explode as supernovae. Ordinary red giants are evolved stars of more modest mass. Mira has a current mass of 1.2 solar masses (according to your Wikipedia link) and would have had an original (main sequence) mass of not much more.
It probably has the largest apparent size (angular diameter) of any star except the sun, but it isn't "one of the largest known." For comparison, here is a real red supergiant: http://en.wikipedia.org/wiki/Antares. This has about 75% larger radius than Mira.
I believe that Mira was the first star discovered to be variable, hence the name. (Same root as "miracle".)
I can't see this working - it relies on nobody else being in the market. Imagine the shares are "worth" $1.00, but SCO wants the share price to be $1.50. SCO shill Alice puts her parcel of 1000 shares up for sale at $1.50, and then SCO shill Bob places an order to buy 1000 shares at market - but if random independent shareholder Carol has put her 1000 shares to sell at $1.30, the result is the shills buy 1000 over-priced shares instead of just transfering them. Until they've bought out all the people willing to sell for less than $1.50, they can't set that as the price.
Can anyone with more knowledge of sharemarkets criticize this analysis?
Yes, but the desks will be worth something to Novell, not to the shareholders. SCO now owes Novell more money than they have assets.(The amount owed has not yet been calculated, and may need to be decided at trial.)
I find it interesting that since the start of 2005, the SCO stock price has pretty much followed a series of step functions, the price being very stable between abrupt adjustments: Jan 2005 - June 2006: $4.50 June 2006 - November 2006: $2.00 November 2006 - May 2007: $1.00 May 2007 - yesterday: $1.50 Today: $0.45
I don't remember what if anything happened in June 2006. November 2006 was when most of SCO's supposed evidence against IBM was thrown out because they hadn't been specific enough. The May 2007 rally is a mystery. Today's drop is obvious.
In their setup, each city has a delay line (i.e. optical fibre.) Each new city you add has to have a delay line twice as long as the previous one you added. The required amount of fibre grows exponentially with the number of cities.
It sounds like the package management software could deal with this. By default, mount all disks "noatime". For each package, note whether it uses atimes. If you install a package which uses atimes, the default mounting is changed (and you're warned about the performance hit.)
I suppose some common tools like 'find' would need to be modified to not require atimes, and to use them only if available.
Slashdot Mirror
Good points.
Certainly there are problems with this definition of complexity.
Where to set the level zero player? "Human novice" is very fuzzy, but "random" could give many levels of 'complexity' within the 'idiotic playing' realm.
Where to set the top level player? "Best human" makes the complexity depend on how much effort we've put into it. Comparing to "perfect play" is usually computationally infeasible.
I'm a poor go player, and I've never tried the 3D Go in a cube, so what I say may be completely off base.
The problem is that in cubical go, you can't make walls, eyes, or captures.
In standard go, adding a stone to a group (but not connecting to another group) increases the group's number of liberties by at most two. In cubical go, it can increase by four. This means it is just too hard to restrict the number of liberties a group has.
If a player is trying to join two groups in standard go, and the opponent places a stone between them, the player has two ways around that stone (left or right). The opponent might be able to block both of these. In cubical go, they have four ways around - it will be very hard to stop them connecting.
An eye requires many more stones in cubical go. The opponent has many chances to disrupt it. Especially because of the above point, which means that stones placed deep in your territory can very likely still live.
I think you'll just end up with lots of intertwining filamentary structures until the board becomes so crowded that captures become possible, then the first person to get a capture of a sizable group will win.
This idea is original (I haven't copied it from anywhere) but it has probably been independently thought of before.
A natural thing to ask about a 2D game such as go, chess, naughts-and-crosses/tic-tac-toe is how can it be extended to three dimensions?
For go, the immediate thought is to play on a 3D grid of cubes. However, I believe (I haven't actually tried) that this will work very poorly. The number of neighbours that each point has will have a huge effect on play. With two many neighbours (6 in the case of a cubic grid) it will be impossible to defend territory, or to capture stones until nearly the entire board fills up. The game would be boring.
To capture the essence of go in 3D, we need each point to have four neighbours. This can be achieved with maximal symmetry by using either truncated octahedra or rhombic dodecahedra as the board's cells.
UPDATE: After writing the above, I did a quick search and discovered diamond go which has clearly used the same basic idea. I haven't figured out yet whether it is equivalent to one of my two options.
On complexity: perhaps the best definition (for matching our intuitive feeling for how complex a game is) I've come across for game complexity is this: Imagine a novice player who avoids glaringly wrong moves (e.g. sacrificing a queen for no benefit in chess) but otherwise randomly choses from the legal moves available to them. Call this a 'level 0' player. Now a player who is smart enough to beat the level zero player 2/3 of the time is a level 1 player, someone who beats a level 1 player 2/3 of the time is level 2 etc. Then the game's complexity is the level of the best human player. If I recall correctly, this gives naughts-and-crosses/tic-tac-toe a complexity of 1, chess was about 20 and go was about 40. Unfortunately, this is all from memory, I haven't found the right magic words to feed Google to get more information.)
So yes, the game-state complexity can be a very poor measure of actual complexity, but go still has a very strong claim to being the most complex common human game. (An extreme example of high game-state complexity but trivial actual complexity would be a large game of "Brussel Sprouts".
On the Diplomacy/Axis and Allies comparison. The three games being compared are very different in nature (although they are all zero-sum.)
Go is two player, complete information, deterministic (as are chess, checkers, hex, naughts-and-crosses/tic-tac-toe) Such a game can, in theory, be completely analysed.
Diplomacy is multi-player, deterministic, incomplete information because moves are simultaneous. I'm not sure if or how this could be analysed.
Axis and Allies is (effectively) two player, probabilistic. It could (theoretically) be completely analysed as a very large dynamic programming problem. From a game-theoretical point of view, it resembles Yahtzee.
If even 10% of these consumers take up the Linux option, that would translate into a doubling of desktop Linux users, in effect, hastening the onset of an inflection point.
And yes, while more people know Windows, there are tens of millions now who also know Linux. Linux is fast approaching that first inflection-point. A move by regulators to ensure that there is a breathing-space for competition will likely see that inflection-point come sooner than later.
You use that word a lot. I don't think it means what you think it means.
It indicates a point where the second derivative of a function is zero - less technically, where the curve is temporarily straight, or the transition point between acceleration and deceleration (either way). The author seems to think it is a good thing. I can only make sense of this if the author thinks that Linux adoption is currently decelerating, and so passing an inflection point will move into accelerating adoption.
The following is from memory from "Technology in World Civilization" by Arnold Pacey, which is a very readable book.
The Mongols were very influential in transfering technology between east and west. They gathered lots of scholars from their empire into their capital. An effect of this was that (for example) Chinese and and Turkish scholars could learn from each other. They would later go home with the new knowledge.
It is often the case that data analysis and interpretation for particular aspects of a research project (like 1-2 figures in a 7 figure paper) are up for vigorous debate.
A professor I work with commented to me about a paper he'd been asked to review. In his review, he said something like "The data analysis is completely wrong, but the data set is wonderful. The paper should be published - then somebody else will reanalyse it correctly."
That matches my IANAL understanding of the discussions at groklaw.
There are various classes of creditors in a bankrupcy. The highest priority class gets paid first, then the next etc. If there is enough money to pay only a fraction of the claims in a given priority class, everyone gets the same proportion of money owed.
The highest priority classes are paying the bankrupcy lawyers and paying back fraudulently acquired wealth. (I'm not sure which order these two are in.) As Novell's claim is in the 'fraudulent' class, they get paid first. If (as is likely) the amount owed is greater than SCO's total assets, Novell gets everything - all the assets and none of the liabilities of SCO.
I don't know whether the assets are just given to Novell, or if they are sold off and Novell gets the proceeds. I suspect they'll be sold - it is the only real way to determine the value of assets such as the Unix IP.
If they are sold off, Novell is likely to be in the bidding, and is likely to win the bidding. Whatever they bid isn't 'real money' to them - it comes straight back.* All they stand to lose is whatever amount of money somebody else was willing to pay for the Unix IP. Even if such a buyer exists, they would probably prefer to buy it from Novell after the liquidation, rather than as part of the SCO liquidation sale. This allows them to renegotiate what gets transfered, rather than having to follow the chain of old deals Novell -> old Sco -> Caldera/new SCO -> hypothetical buyer.
* Exception: if Novell bids so much that the SCO assets exceed what SCO owes them, then they'll pay some real money.
I expect the IBM case will pretty much evaporate. Novell would be successor-in-interest to SCO, and might have the option to keep fighting it, but obviously won't. I'm guessing (IASNAL) that if they did chose to keep fighting, they'd also have to accept some amount of liability for IBM's lawyer's fees if (when) they lost.
Note that I said "if your data format is clever enough". Evidently you know enough to say that the current formats are not clever enough. I'm happy to accept the current situation as falling into "not feasible" (at present.) From an information-theory point of view, it must be possible to use the information in the DVD data to reduce the storage requirement for the rest.
These are 40Gb disks, and a DVD is just under 4Gb, hence the 10% figure. Whether this is acceptable will differ according to who you are.
Stereo records played on mono turntables. Colour TV broadcasts displayed on B&W sets. Such compatibility is good, so long as you don't pay too much for it, hobbling the new format to the needs of the old.
So long as the manufacturing price of disks is low, this would allow DVD/VMD disks to be sold for the same price as a DVD. Then when you're buying DVD/VMD disks to play in your DVD player, you can later upgrade to a VMD player and instantly have a collection of HD content. This would appeal to consumers and the VMD drive manufacturers. The content sellers might want to charge more for HD, which could sink the proposition. (It relies on DVD/VMD combo disks costing the same as plain DVD.)
You lose about 10% of the disk capacity for DVD data which isn't used by the VMD playback. You might not even lose that 10% - if your data format is clever enough, you can use that 10% within the VMD playback rather than redundantly repeating it in the non-VMD data. I'm sure this is possible, but I'm not sure it is computationally feasible.
Coming late to the game, they need to establish a 'home ground', a niche that they can dominate and then grow out from. India could be that home ground.
A fine feature would be if it were possible to play the new HD VMD disks at DVD resolution on standard DVD players. Given they use the same lasers, it might be that DVD players will see one particular layer, on which the DVD data could be stored. This again would help greatly to break into the market.
However, they don't mention such a feature, and I'd hope they'd have thought about it, so probably it is technically infeasible.
Maybe it didn't phone home, and Rick Rubin (a music producer, not a computer geek) just doesn't understand what the root kit did.
You seem to be thinking about analyses where there are only two or three species included. In that circumstance, you can't test the clock, and have the problems you talk about.
Instead we'd do an analysis with say 12 primates. This is how we get the "sufficient data points" you worry about. (Some analyses have hundreds of species, but it becomes computationally difficult to reliably construct the trees in this circumstance.) Typically we have one or more species (the outgroup) which are known a priori to be more distantly related than the rest. In this case, we might use new world monkeys or some non-primates as the outgroup. Where the outgroup attaches to the rest (the ingroup) establishes the 'root' of the tree for the ingroup.
Now, within the ingroup, we look at the genetic distance from the present species back to the root. If all is well, this will be highly consistent (say all within 10% of each other.) Given that our ingroup may contain species which have diverged (say) over 30 million years ago, we have evidence that the clock is stable (within 10%) over periods of 30 million years. Therefore we can rely on it for the 5 million year human-chimp split.
It is always necessary to callibrate the clock against one or more well-established divergence times from the geological or fossil record. This might (for example) be the divergence between old and new world monkeys.
There are also "relaxed clock" analyses which allow the clock to vary over the tree, but only slowly. This is necessary to date deep divergences (e.g. amphibians vs other land vertebrates) as no genes will have maintained a steady clock over that time scale.
The human-chimp date will be based on many hundreds of genes. Less well studied species may have dates based on just a handful of genes.
We can get ancient DNA from suitable samples back to about 50,000 years. However, it would be hazardous to extrapolate to millions of years from this data. Most molecular clock analyses use contemporary data only.
This find perhaps sets back the date of chimp-gorilla split but not "human-ape".
The genetic human-gorilla divergence is only slightly larger than the human-chimp divergence. If we indeed had firm evidence that humans and gorillas diverged (say) 12 million years ago, then we would use this as a callibration point for our molecular clock, and it would put the human-chimp divergence back to about 10 million years ago. These dates are not independent.
Prior to [the KT boundary] the mammal line had already split much more than we previously gave it credit for, a lot of the main groups were developed.
Though, it's even weak evidence that it's actually a gorilla just that it had a diet like that of a gorilla.
These I agree with, although I don't have the expert knowledge for the second point, so I take my own belief with grains of salt.
IAA evolutionary molecular biologist*. There are some significant incorrect assumptions in your post.
First, the rate of genetic change used to date divergences is independent of the rate of morphological change. That cockroaches aren't changing much outwardly has no effect on the genetic drift that lies behind the 'molecular clock'. Some DNA is 'neutral' (changes have no effect on the fitness of the organism.) In genes for proteins, there will be many 'equally good' (so close to equal that evolution can't distinguish) solutions, and over time a species solution will 'drift' through this cloud of optimal solutions. These processes are independent of morphological change.
Second, we very well understand that the molecular clock is imperfect. It runs at different rates in different genes. In some genes, the rate is highly variable. In some lineages, there is an acceleration of rates. (E.g. rodents have a high rate compared to primates.) We look for these problems and correct for them or work around them.
Typically we draw phylogenies (family trees for species) with branch lengths proportional to genetic distance. (I.e. time, as measured by a putative molecular clock.) We know that, if we were able to plot the phylogeny measured by a real clock, the tips (current species) would all lie at the same time (the present.) If our phylogeny has its many tips nearly aligned, then we believe we have a good molecular clock, and we can estimate how good it is. If (as not infrequently happens) a particular gene or group of genes give a tree with poor tip alignment, we conclude there is no good molecular clock for this gene/group of genes, and we don't use it for dating divergences. If we see a lineage which systematically has branches too long or too short, we infer a change in the clock rate on that lineage, and take care to account for this when dating divergences within that lineage.
In short: We don't blindly assume the existence of a molecular clock. We hope for a clock, and look for the genes which act most clock-like, and we calibrate both the rate *and* the accuracy of the clock.
Returning to the case in hand: I expect that this is a case of convergent evolution. A 10 million years ago ape, not an ancestor to humans or gorillas, evolved gorilla-like teeth in response to similar evolutionary pressures. However, I don't have the knowledge (in paleontology and morphology) to assess the claims directly. I have reason to believe the molecular divergence date couldn't be so badly wrong, and the evidence against it seems thin to me.
* More accurately, I'm an astronomer pretending to be a mathematician pretending to be an evolutionary molecular biologist.
If we go back about 20 years, hard drives were for non-volatile fast-access storage, and tape drives were for backup, bulk data storage, archiving and sometimes data transfer (when there was too much for floppies.)
Now that flash is reaching the point where we can contemplate using it for the primary non-volatile storage niche, we may see hard drives being displaced into the backup/bulk storage/archiving niches. If so, expect to see increasing emphasis on ways to hot-plug hard drives into your computer, and increasing emphasis on price/GB and decreasing emphasis on performance and possibly per-drive capacity.
We'll really know we've reached this point when hard drives are used as a medium for delivering software.
A very few brilliant men have been called crackpots by their contemporaries but have ended up being exonerated by history.
The only example that springs to mind is continental drift, and now that I look it up in the Infallable Font Of All Knowledge I see that there were about half a dozen people pushing that. I'm sure there are other examples - perhaps even a few dozen.
By contrast, many thousands (probably millions) have been called crackpots by their contemporaries and ended proven so by history, or simply forgotten.
Minor correction: it is a red giant, not a red supergiant. Supergiants are very massive stars (on the order of 10 times the mass of the sun or more) which will eventually explode as supernovae. Ordinary red giants are evolved stars of more modest mass. Mira has a current mass of 1.2 solar masses (according to your Wikipedia link) and would have had an original (main sequence) mass of not much more.
It probably has the largest apparent size (angular diameter) of any star except the sun, but it isn't "one of the largest known." For comparison, here is a real red supergiant: http://en.wikipedia.org/wiki/Antares. This has about 75% larger radius than Mira.
I believe that Mira was the first star discovered to be variable, hence the name. (Same root as "miracle".)
I can't see this working - it relies on nobody else being in the market. Imagine the shares are "worth" $1.00, but SCO wants the share price to be $1.50. SCO shill Alice puts her parcel of 1000 shares up for sale at $1.50, and then SCO shill Bob places an order to buy 1000 shares at market - but if random independent shareholder Carol has put her 1000 shares to sell at $1.30, the result is the shills buy 1000 over-priced shares instead of just transfering them. Until they've bought out all the people willing to sell for less than $1.50, they can't set that as the price.
Can anyone with more knowledge of sharemarkets criticize this analysis?
By this logic, you might as well buy lotto tickets.
If the expected return is less than the purchase price, it is a bad buy, even if the tickets only cost $0.45 each.
Yes, but the desks will be worth something to Novell, not to the shareholders. SCO now owes Novell more money than they have assets .(The amount owed has not yet been calculated, and may need to be decided at trial.)
I find it interesting that since the start of 2005, the SCO stock price has pretty much followed a series of step functions, the price being very stable between abrupt adjustments:
Jan 2005 - June 2006: $4.50
June 2006 - November 2006: $2.00
November 2006 - May 2007: $1.00
May 2007 - yesterday: $1.50
Today: $0.45
I don't remember what if anything happened in June 2006. November 2006 was when most of SCO's supposed evidence against IBM was thrown out because they hadn't been specific enough. The May 2007 rally is a mystery. Today's drop is obvious.
Five year graph
One year graph
One week graph
I can't see how there would be.
In their setup, each city has a delay line (i.e. optical fibre.) Each new city you add has to have a delay line twice as long as the previous one you added. The required amount of fibre grows exponentially with the number of cities.
It sounds like the package management software could deal with this. By default, mount all disks "noatime". For each package, note whether it uses atimes. If you install a package which uses atimes, the default mounting is changed (and you're warned about the performance hit.)
I suppose some common tools like 'find' would need to be modified to not require atimes, and to use them only if available.