You should perhaps check your calculation (and then gracefully rescind your proposition.) If the stated proabability is 37 * 10^-6, it cannot be 1.78%! This is absurd regardles of the size of the database. If not convinced, then I submit the following thought experiment: Does the proabability of error approach zero as the number of samples in the database approaches infinity?
All good scientific data is couched in probability, no? I would say that this very unlikely event is, even after its occurance, statistically insignificant. In any case, to increase the assay by a locus or two, or to even double the sampled loci, would not really increase its perceived accuracy in the courts. (By all means, this sort of analysis is far more reliable than a trivial fingerprint identification; and I would guess so by orders of magnitude.) By the way, the greatest random error is introduced at the stage of polymerase chain reaction. The inherent error in this step of the analysis vastly overwhelms the rather insignificant probability of computer error.
Face it, Intel is just not scared of Transmeta. I mean, the mobile chip Crusoe is neat but I can't believe it scares Intel that much. If AMD couldn't make Intel budge, how can scrappy Transmeta do any real damage. (They're so small they're evading our turbolasers! --sorry I couldn't help it.) I mean, Intel spends billions and billions of fab plants, R&D, marketing, etc. And anyone who believes that this is the death knell for Intel is either really young or just plain forgot the leapfrog way in which of CPUs are developed.
Slashdot Mirror
You should perhaps check your calculation (and then gracefully rescind your proposition.) If the stated proabability is 37 * 10^-6, it cannot be 1.78%! This is absurd regardles of the size of the database. If not convinced, then I submit the following thought experiment: Does the proabability of error approach zero as the number of samples in the database approaches infinity?
All good scientific data is couched in probability, no? I would say that this very unlikely event is, even after its occurance, statistically insignificant. In any case, to increase the assay by a locus or two, or to even double the sampled loci, would not really increase its perceived accuracy in the courts. (By all means, this sort of analysis is far more reliable than a trivial fingerprint identification; and I would guess so by orders of magnitude.) By the way, the greatest random error is introduced at the stage of polymerase chain reaction. The inherent error in this step of the analysis vastly overwhelms the rather insignificant probability of computer error.
Face it, Intel is just not scared of Transmeta. I mean, the mobile chip Crusoe is neat but I can't believe it scares Intel that much. If AMD couldn't make Intel budge, how can scrappy Transmeta do any real damage. (They're so small they're evading our turbolasers! --sorry I couldn't help it.) I mean, Intel spends billions and billions of fab plants, R&D, marketing, etc. And anyone who believes that this is the death knell for Intel is either really young or just plain forgot the leapfrog way in which of CPUs are developed.
no
Parvus sed potens.
One may dual boot between Linux and Windows 2000. Works like a charm.