The problem is that so much talent and effort (and money!) has gone into building fast x86 processors that it will take an enormous effort (and expenditure) to bring another architecture to a competitive level. IBM, I gather, is not pushing the POWER architecture as far as it could (eg., using highly modular designs to save time and money at some cost in performance).
You are quite right about the Alpha. If it had gotten a fraction of the Itanium's budget, it would be unbeatable by now.
This makes sense. The Itanic^Hum is actually quite good at running Fortran programs with enormous DO loops, but Intel and AMD x86oid processors are better at everything else -- including running Windows itself.
The Itanium was the only realistic chance we had to get away from the x86 for the forseeable future, and the designers blew it. So sad. Excuse me while I start one of my Leonard Cohen albums, I need something to cheer me up.
OpenDocument is an OASIS standard, but it comes from the StarOffice/OpenOffice people. They obviously put a lot of work into developing a good set of formats for office documents, as opposed to letting the coders design the format. (I'm a coder, but...) They make heavy use of W3C standards such as CSS, XSL-FO, SVG and MathML, so there's lots of potential for interopability. See http://en.wikipedia.org/wiki/OpenDocument for a good introduction. You can download the OpenDocument specification itself from http://www.oasis-open.org/committees/tc_home.php?w g_abbrev=office. From what I've read, it's an excellent piece of work.
Contrast this to Microsoft's poorly-documented new XML format, which is mired in the deep and dangerous swamps of backward compatibility with everything from OLE onwards.
Actually, the US Federal government required POSIX compliance, at least back in the early 1990s. But the POSIX standard was fairly loose -- as I understand, it made lots of stuff optional.
So Microsoft went through the POSIX standard and implemented the bare minimum to make NT pass the FIPS requirements, so that the bureaucrats could tick the "POSIX compliance" boxes on their forms. The result was not useful to programmers, nor was it intended to be. DEC added (real and useful) POSIX compatibility features to VMS at the same time for the same reasons. IBM did the same to MVS.
A little later, Microsoft started developing an add-on for NT which gave useful POSIX compatibility. I guess this is where Services For Unix came from.
My information comes from a talk at a Melbourne Linux User Group meeting over 10 years ago, so I may have some of the details wrong. The fun bit was that the talk was by a Microsoft engineer.
TFA is talking about something I've been noticing more and more: too many "scientists" don't understand (or tend to ignore) the basic principles of statistics. The classic example is that on average 1 in 100 tests will be more than 3 standard deviations out -- and therefore "significant at the 99% level". So looking (say) at 20 possible causes for 20 possible effects is expected to produce 4 'false positives', purely as a matter of chance. (People often report results that are 2 sigmas out -- the 95% confidence level -- which gives far more false positives.)
Basic statistics is not that hard, but getting things right does require a high level of clear thinking. Unfortunately, too many people don't understand the basic principles. Even worse, there are computer programs which enable people to use all sorts of statistical techniques, including the more advanced ones, without really understanding what is going on. Unfortunately, it seems to me, many scientists prefer publishing lots of 'soft' papers to taking the time to do rigorous statistics. This is only to be expected, because the incentive structure for academic careers tends to reward quantity of publications over quality, or is at least perceived to do so. (This is part of a much bigger and harder problem. Systems such as peer-reviewed journals worked well for the much smaller pre-WW2 academic communities, but seem inadequate for today's needs.)
Of course, it's not just scientists who have problems with misunderstanding and misuse of statistics. Perhaps we should regard "statistical literacy" as a fundamental attribute of a good education, along with "computer literacy".
Speaking of which, can anyone recommend a good book on statistics for non-dummies? I think I should do some reading...
In Australia, roundup-resistance was found in wild ryegrass in the early 90s. (Is this the case that LothDaddy mentioned?) So roundup-resistance
CAN develop naturally.
Slashdot Mirror
You are quite right about the Alpha. If it had gotten a fraction of the Itanium's budget, it would be unbeatable by now.
The Itanium was the only realistic chance we had to get away from the x86 for the forseeable future, and the designers blew it. So sad. Excuse me while I start one of my Leonard Cohen albums, I need something to cheer me up.
Contrast this to Microsoft's poorly-documented new XML format, which is mired in the deep and dangerous swamps of backward compatibility with everything from OLE onwards.
Which would you trust?
A little later, Microsoft started developing an add-on for NT which gave useful POSIX compatibility. I guess this is where Services For Unix came from.
My information comes from a talk at a Melbourne Linux User Group meeting over 10 years ago, so I may have some of the details wrong. The fun bit was that the talk was by a Microsoft engineer.
Basic statistics is not that hard, but getting things right does require a high level of clear thinking. Unfortunately, too many people don't understand the basic principles. Even worse, there are computer programs which enable people to use all sorts of statistical techniques, including the more advanced ones, without really understanding what is going on. Unfortunately, it seems to me, many scientists prefer publishing lots of 'soft' papers to taking the time to do rigorous statistics. This is only to be expected, because the incentive structure for academic careers tends to reward quantity of publications over quality, or is at least perceived to do so. (This is part of a much bigger and harder problem. Systems such as peer-reviewed journals worked well for the much smaller pre-WW2 academic communities, but seem inadequate for today's needs.)
Of course, it's not just scientists who have problems with misunderstanding and misuse of statistics. Perhaps we should regard "statistical literacy" as a fundamental attribute of a good education, along with "computer literacy".
Speaking of which, can anyone recommend a good book on statistics for non-dummies? I think I should do some reading ...
In Australia, roundup-resistance was found in wild ryegrass in the early 90s. (Is this the case that LothDaddy mentioned?) So roundup-resistance CAN develop naturally.