Having MS I follow the research news (see National MS
Society website to find listings). The good news is
that there are many "promising" treatments, including
this one. However the press release descibes the result
of a "Phase I" study, which by definition is an experiment
to determine the _safety_ of a new treatment, not its
efficacy. Most candidate treatments that do OK
in their Phase I study, fail in their Phase II and Phase III
studies. Only 3(4) drugs have run the gauntlet and
are in use for non-experimental treatment (one is
available only under special conditions). The role
of various cells of the immune system, of myelin
destruction and other (unknown) neurodegenerative
processes, the cause, and much more are hotly
debated in the research literature. Myself, I'm
insisting on both a cure AND a method to
repair the damage already done. So there!
I find myself agreeing with everyone (a/. first?). Myself, learned to program on an early PDP (N less than 8 but I can't remember how much less). Wrote the binary program directly in a 3-ring binder (ever wonder where the term "page" came from re RAM?) since we had no assembler available. Entered the code with front panel switches and/or paper tape. Well I learned a LOT about how a very basic CPU works at the bit level. Modern CPUs are very very much more complex -- but the underlying mental model of languages like C and its offspring is exactly that primitive old CPU. So going down to the guts has some benefits. BUT, some years later concepts like data abstraction, functional languages, etc were invented. I found that the interpretive nature of LISP for example made it MUCH easier to understand, implement, and debug programs using these techniques. LISP does not, IMHO, try to present a program representation that looks at all like that old primitive CPU, or indeed any CPU. And yet data abstraction (of which OO is a subset) is certainly a vital thing for almost any computer literate person to understand, plausibly more important than undesanding indirect memory references or interrupts for most folks. I also later learned to work with programs that were not examples or toys, and they always turned out not to be "a program" at all but a largish collection of interrelated programs, headers, etc. That's where IDE's are really good. Last point. I last did programming "manually" (no IDE) when "make" was new. Nowadays everybody uses automatically generated make files that are basically not readable by humans. Personally I use an IDE on the few occasions I program at all anymore, so no make files. But I have to figure out and at times correct packages that come with a zillion c, cpp, h files etc plus unreadable make/imake files (or worse) -- there must be tools that will take these configuration control file things and translate them into something a human can understand, so that one has a chance of finding which one of the zillion modules one needs to fix and/or modify -- clue an old dino in: what's that tool called (Unix/Linux environment please).
Even laster: the whole point of python and friends is to hide all of this stuff, so just use whatever makes life easy if that's what you're teaching. Still won't simplify the hardest part: thinking hard before coding. To teach that, I suggest changing the requirements for medium size class projects every few days up to or even including the day the project is due. This will teach respect for the importance of clean, well-structured design in any language. It's also a lesson in how the real world of programming tends to work.
A glorious victory for US scientists -- we seem to have conquered the Canadian Herzberg Institute of Astrophysics. My Canadian colleagues will taste the "lite-beer of defeat" at last;->
"...a reforming of the system..."
Check out H.R. 2795, the "Patent Act of 2005." It may fix some of these obious problems. Stil time to get your 2 worth in.
Unless one envisions a future in which the bulk of the world's poulation is eternally mired in a under-developed nation limbo,
it is enevitable that the most populous countries will have the highest GDPs with all that entails. Rather than dreading this desirable (IMHO) and inevitable (I hope) change, the USA should be thinking about its _realistic_ place in a globalized future world. I would suggest the model of the Neatherlands: with a population of only 16 million they remain a major player in science, R&D, and commercial activities. Of course the Netherlands are not a "super power" like the USA currenlty is, but the citizens there have and can expect to maintain a very hgih standard of living (unless global warming drowns them all:-o).
IMHO the essential problem is not getting cold in but heat out. In most data center cooling schemes the exhaust hot air is allowed to mix with the cool room air; thermodynamically, this is a big mstake. What is lacking is simply a set of hot air ducts that can remove the hot exhaust air from each rack and expel it from the building (it will almost always be hotter than the outside air, so it makes more sense to cool ambient external air with heat pumps than to recool the exhaust air). If one designs one's data center this way, one does not even require a raised floor, since the entire room can now act as the cold air supply plenum.
In addition to having a much larger collecting area, the JWST will also be an infrared telescope rather than a visible light telescope like the Hubble. This will make the JWST better suited to studying objects at high redshift, but will also mean that the JWST will have less angular resolution than the Hubble did, Never fear, however -- adaptive optics on large ground based telescopes, such as the Kecks and the European VLT (in Chile) are already providing optical images with resolution greater than that of the Hubble despite our pesky atmosphere. So all in all, Hubble's day is done, and it should be retired sooner or later -- just when is worth arguing about, but is not really of fundamental importance.
The SETI detection tasks I wrote of are those used for the Phoenix system at present, and the Allen Telescope Array in future. The "SETI@home" project is a different, complementary effort run out of UC Berkeley (setiathome.ssl.berkeley.edu). They search the entire sky while Phoenix at present is searching only the nearest 1000 sun-like stars, Phoenix however searches about 700 times as large a range of frequencies with about 100 times the sensitivity. With the the Allen Telescope Array we plan to expand our high-sensitivity "targeted search" to the nearest 100,000 stars (maybe a million in time). We also hoping someday to build a detector that watches the whole sky all the time for bright but transient signals.
Slashdot Mirror
Having MS I follow the research news (see National MS Society website to find listings). The good news is that there are many "promising" treatments, including this one. However the press release descibes the result of a "Phase I" study, which by definition is an experiment to determine the _safety_ of a new treatment, not its efficacy. Most candidate treatments that do OK in their Phase I study, fail in their Phase II and Phase III studies. Only 3(4) drugs have run the gauntlet and are in use for non-experimental treatment (one is available only under special conditions). The role of various cells of the immune system, of myelin destruction and other (unknown) neurodegenerative processes, the cause, and much more are hotly debated in the research literature. Myself, I'm insisting on both a cure AND a method to repair the damage already done. So there!
I find myself agreeing with everyone (a /. first?). Myself, learned to program on an early PDP (N less than 8 but I can't remember how much less). Wrote the binary program directly in
a 3-ring binder (ever wonder where the term "page" came from re RAM?) since we had no
assembler available. Entered the code with front panel switches and/or paper tape.
Well I learned a LOT about how a very basic CPU works at the bit level. Modern CPUs
are very very much more complex -- but the underlying mental model of languages
like C and its offspring is exactly that primitive old CPU. So going down to the guts
has some benefits. BUT, some years later concepts like data abstraction, functional
languages, etc were invented. I found that the interpretive nature of LISP for example
made it MUCH easier to understand, implement, and debug programs using these techniques. LISP does not, IMHO, try to present a program representation that looks at all like that old primitive CPU, or indeed any CPU. And yet data abstraction (of which
OO is a subset) is certainly a vital thing for almost any computer literate person to
understand, plausibly more important than undesanding indirect memory references or interrupts for most folks. I also later learned to work with programs that were not
examples or toys, and they always turned out not to be "a program" at all but a largish
collection of interrelated programs, headers, etc. That's where IDE's are really good.
Last point. I last did programming "manually" (no IDE) when "make" was new. Nowadays everybody uses automatically generated make files that are basically not readable
by humans. Personally I use an IDE on the few occasions I program at all anymore, so no make files. But I have to figure out and at times correct packages that come with a zillion c, cpp, h files etc plus unreadable make/imake files (or worse) -- there must be tools that will take these configuration control file things and translate them into something a human can understand, so that one has a chance of finding which one of the zillion modules one needs to fix and/or modify -- clue an old dino in: what's that tool called (Unix/Linux
environment please).
Even laster: the whole point of python and friends is to hide all of this stuff, so just use whatever makes life easy if that's what you're teaching. Still won't simplify the hardest part: thinking hard before coding. To teach that, I suggest changing the requirements for medium size class projects every few days up to or even including the day the project is due. This will teach respect for the importance of clean, well-structured design in any language. It's also a lesson in how the real world of programming tends to work.
A glorious victory for US scientists -- we seem to have conquered the Canadian Herzberg Institute of Astrophysics. My Canadian colleagues will taste the "lite-beer of defeat" at last ;->
"...a reforming of the system..." Check out H.R. 2795, the "Patent Act of 2005." It may fix some of these obious problems. Stil time to get your 2 worth in.
Unless one envisions a future in which the bulk of the world's poulation is eternally mired in a under-developed nation limbo, it is enevitable that the most populous countries will have the highest GDPs with all that entails. Rather than dreading this desirable (IMHO) and inevitable (I hope) change, the USA should be thinking about its _realistic_ place in a globalized future world. I would suggest the model of the Neatherlands: with a population of only 16 million they remain a major player in science, R&D, and commercial activities. Of course the Netherlands are not a "super power" like the USA currenlty is, but the citizens there have and can expect to maintain a very hgih standard of living (unless global warming drowns them all :-o).
IMHO the essential problem is not getting cold in but heat out. In most data center cooling schemes the exhaust hot air is allowed to mix with the cool room air; thermodynamically, this is a big mstake. What is lacking is simply a set of hot air ducts that can remove the hot exhaust air from each rack and expel it from the building (it will almost always be hotter than the outside air, so it makes more sense to cool ambient external air with heat pumps than to recool the exhaust air). If one designs one's data center this way, one does not even require a raised floor, since the entire room can now act as the cold air supply plenum.
A long discussion of many of the points raised in this thread can be found here.
In addition to having a much larger collecting area, the JWST will also be an infrared telescope rather than a visible light telescope like the Hubble. This will make the JWST better suited to studying objects at high redshift, but will also mean that the JWST will have less angular resolution than the Hubble did, Never fear, however -- adaptive optics on large ground based telescopes, such as the Kecks and the European VLT (in Chile) are already providing optical images with resolution greater than that of the Hubble despite our pesky atmosphere. So all in all, Hubble's day is done, and it should be retired sooner or later -- just when is worth arguing about, but is not really of fundamental importance.
"Oh, Rumania, Rumania, Rumania, Rumania, Rumania, Rumania, Rumania. Was such a lovely place, I just can't explain ya..."
The SETI detection tasks I wrote of are those used for the Phoenix system at present, and the Allen Telescope Array in future. The "SETI@home" project is a different, complementary effort run out of UC Berkeley (setiathome.ssl.berkeley.edu). They search the entire sky while Phoenix at present is searching only the nearest 1000 sun-like stars, Phoenix however searches about 700 times as large a range of frequencies with about 100 times the sensitivity. With the the Allen Telescope Array we plan to expand our high-sensitivity "targeted search" to the nearest 100,000 stars (maybe a million in time). We also hoping someday to build a detector that watches the whole sky all the time for bright but transient signals.
http://math.ucr.edu/home/baez/physics/Administrivi a/booklist.html