Slashdot Mirror

How about this?

http://www.nrc.gov/about-nrc/radiation.html

You could look at a nuclear engineering text if you wanted to know more.

http://www.nuc.berkeley.edu/NE-39
http://www.nuc.berkeley.edu/NE-101
http://www.nuc.berkeley.edu/NE-124

How about this?

http://www.nrc.gov/about-nrc/radiation.html

You could look at a nuclear engineering text if you wanted to know more.

http://www.nuc.berkeley.edu/NE-39
http://www.nuc.berkeley.edu/NE-101
http://www.nuc.berkeley.edu/NE-124

Well, IANASE (Supercomputer Expert) but I *am* a programmer....

I'm assuming that you have a supercomputer when all those otherwise individual computers are working together in a coordinated fashion on a common problem.

A great example of a supercomputer is SETI @ Home which easily meets the definition of a "supercomputer" in many (most?) circles, although they usually refer to it as "distributed computing".

PILOT and LOGO:

RPILOT is a FOSS version of Pilot. Here is a list of LOGO resources find one that works for you. Berkely LOGO or Star LOGO or Net LOGO take your pick.

Grade Schools and Junior Highs used to teach PILOT or LOGO or both on the old 8 bit computers when I was that age. It was fun giving a virtual turtle commands to draw shapes and solve problems.

I've been following along this semester's CS61A Lectures by Dr. Brian Harvey out of UC Berkeley (audio and video podcast). He devotes two lectures to a 20 some year old videotape of Alan Kay talking about the coalescing of OOP principals in SmallTalk. Kay makes an important point: at different ages we learn differently. He also shows kids doing clever things with drawing and computer animation and they do it by writing programs. Look for Sept. 12 and 15.

I also found a book from Apress "Squeak, Learn Programming With Robots" which I think is very good, once one gets over the disappointment that the robots are graphics producers and not metal crushing monsters or lovable rogues, a la Bender Bender Rodriguez. Author: Stephane Ducasse. ISBN 1-59059-491-6

I recommend Computer Science Logo Style by Brian Harvey, one of the best CS instructors at UC Berkeley. You can get the books for free here (scroll down a bit).

To begin with let's take the computer out of programming Computer Science UnPlugged

There are three environments I'd recommend for teaching programming.

SmallTalk/Squeak coupled with Squeak by Example

Scheme/PLT Scheme coupled with How To Design Programs

Logo/Berkeley Logo coupled with Computer Science Logo Style

The big idea is not to teach them how to do something in one language but to teach them the concepts that will allow them to learn how to do it in any language.

If the kids really are gifted then the class time is too valuable to have them sat there pounding keyboards, that's manual labor. Treat the class as lectures/discussions with experimenting on computers left up to them.

Try and get them to learn about the lectures subject matter before the class e.g. publish a schedule and make it clear they are expected to have read so and so pages before the class.

If they are gifted and motivated this will be as close to the Platonic style of teaching as you can get.

The side effect is that you have prepared them to succeed in college if they can get the idea of using lectures as a review of what they already know.

To begin with let's take the computer out of programming Computer Science UnPlugged

There are three environments I'd recommend for teaching programming.

SmallTalk/Squeak coupled with Squeak by Example

Scheme/PLT Scheme coupled with How To Design Programs

Logo/Berkeley Logo coupled with Computer Science Logo Style

The big idea is not to teach them how to do something in one language but to teach them the concepts that will allow them to learn how to do it in any language.

If the kids really are gifted then the class time is too valuable to have them sat there pounding keyboards, that's manual labor. Treat the class as lectures/discussions with experimenting on computers left up to them.

Try and get them to learn about the lectures subject matter before the class e.g. publish a schedule and make it clear they are expected to have read so and so pages before the class.

If they are gifted and motivated this will be as close to the Platonic style of teaching as you can get.

The side effect is that you have prepared them to succeed in college if they can get the idea of using lectures as a review of what they already know.

Seconded. You can't beat designed for the task. It's got an extremely low learning curve, immediate feedback, and lends itself nicely to exploration. And contrary to popular belief, it's not exactly limited. Brian Harvey at UCB has 3 downloadable books suitable for varying skill levels here.

http://www.cs.berkeley.edu/~kubitron/asplos98/dilbert.html

Or really, lets just damn it all to hell and learn Lisp ... I can just blame someone else when my program is slow.

Nope!

http://www.lrde.epita.fr/~didier/research/verna.06.ecoop.pdf

http://portal.acm.org/citation.cfm?doid=1143997.1144168

http://www.eecs.berkeley.edu/~fateman/papers/lispfloat.ps

Yeah. Learn Lisp first.

Good idea. :) I know /you/ were only joking but Lisp has been held back by a ton of widely believed (and massively ironic) mythology and it is very sad. The only thing really wrong with it is the lack of stuff written in/for it because of its grossly undeserved reputation.

Also seconded. Russel & Norvig. Artificial Intelligence: A Modern Approach is a good book, well illustrated, and generally lacks the undecipherable academia-speak that pervades lots of AI literature.

Here's an article that was particularly influential on me and some of my friends: Brooks, R. 1991. Intelligence Without Reason. MIT AI Memo num 1292. Even though it is 'just' a tech report, it is frequently cited. He had another one, Intelligence without Representation, which is also good.

Somebody else mentioned the McClelland and Rumelhart PDP (neural networks) book, and it is also still quite good in spite of its age.

The interesting thing about AI (to me) is the funny mix of domain expertise. You have philosophers, sociologists, cognitive scientists, psychologists, computer scientists, and mathematicians. That's not a complete list---I'm in human-computer interaction and design research.

But because of the motley crew of domains you have a hundred people speaking a hundred different dialects. Some people put everything in really mathy terms, and their journal articles look (to me) like they are written in Klingon. Then you have others who write in beautiful prose but don't give any specifics on how to implement things. Still others express everything in code or predicate logic.

The oldest school of AI holds that you can reduce intelligence to a series of rules that can operate on any input and make some deterministic and intelligent sense of it. That works to a degree, but it falls apart at some point partly because of the computational complexity (e.g. the algorithm works if you have a million years to wait for the answer). Another reason it falls apart is because there are some kinds of intelligence that can't be reduced to rational computation (e.g. I love my wife because of that thing she does...).

There's a newer kind of AI that is based on having relatively simple computational structures that eat lots of data, "learn" rules based on that data, and are capable of giving fairly convincing illusions of smartness when given additional data from the wild. Neural nets fall into this category.

A third kind of AI brings these two schools together in the belief that there are fundamental computational structures like Bayesian Networks that can model intelligence* but those structures by themselves are insufficient and must be able to adapt based on exposure to real data. So instead of having a static BN whose topology is defined at the start and remains the same throughout the life of the robot, we can have a dynamic BN whose structure changes based on the environment.

I remember reading a recent article by John McCarthy arguing that all this statistical business is hogwash, and that the old school positivist, reductionist approach will eventually win. He's a smart guy, inventor of LISP and a Turing Award recipient. It seems his view is in the minority, but I'm not one to say he's wrong. However, my inclination is that the third hybrid group is probably going to be the one to make the most progress in the years to come.

The reason for my preference to the hybrid school could probably be best explained by Lucy Suchman's Plans and Situated Actions . I can't really do her thesis justice in a few sentences, but the short version of her argument is that there are plans (the sequence of steps that we think we are about to carry out before performing some task) and actions, which is the set of things we actually do. In my mind, a plan corresponds roughly with the underlying computational mechanism, but the actions correspond with how that mechanism executes and what happens when the underlying structure is insufficient, wrong, misleading, or fails.

Hope that helps.

Gabe

* None of this is to say that computational structures that we implement with software/hardware ar

i took a AI course this semester, they recomended this book (AIMA) and gave the book's homepage: http://aima.cs.berkeley.edu/ some sample chapters and links there ...

Agreed. All the basic knowledge about the field is in this book. Part of these are available freely online. You can be judge : http://aima.cs.berkeley.edu/

Also, Bill Gates, to the best of my knowledge, has contributed exactly nothing to any field of science besides his money.

Actually, he has: http://www.cs.berkeley.edu/~christos/papers/Bounds%20For%20Sorting%20By%20Prefix%20Reversal.pdf

That Java code is only 10 lines longer, but it doesn't include the code for some other classes it uses to solve the problem.

But anyhow, you're missing his point. The basic backtracking algorithm for the problem is simple in any language, and could indeed be made much shorter in Java (w/o the entire search framework). He's talking about improving the search with a heuristic (in his case, what is known as a "minimum remaining values" search heuristic among the AI folks). But he's still probably wrong, as you'd only need to write a Comparator in Java, or overload the operator< in C++ to achieve the same effect, and my guess is it'd only take you about 2x the Python code for the same functionality.

But, IMHO, the problem is not so much the increase in code, it's the shift in thinking which you have to undergo to make your code in Java. You just want to sort a bloody list based on a certain criteria, but now you have to make a class, encode your state data in it, and define a comparator function. Basically the brain -> program mapping is most of the time so much more direct in Python (and other similar languages) than the high-level assembly family of C languages that it isn't even funny. I sometimes feel like being put in a straitjacket when I have to write some Java or C++. Don't get me wrong, I definitely agree that a lousy programmer can make a mess in any programming language (I've written my share of bad code) and that a good programmer can write good code in any programming language (save for COBOL), but why torture yourself?

A good indicator for me is the source code for problems in the AIMA book, check out the different versions and see which ones convey the meaning and ideas more clearly.

No it isn't. Studies have shown that temperatures rose significantly across the US right after 9/11 when all planes where grounded for several days, because of clearer skies.

See here.

>Well the placental mammals, like us and almost every other mammal, did not evolve until after the Asteroid event.

Wrong. There were plenty of mammals in the Mesozoic. And according to the University of California Museum of Paleontology "Eutherians first became common in central Asia during the Upper Cretaceous." Eutherians being the technically correct name for placental mammals.

http://www.ucmp.berkeley.edu/mammal/eutheria/eutheriafr.html

And it'd implement the entire IEEE standard with every peculiarity and requirement?

When Woz did the same thing a year before 754 was first drafted, I don't think he worried about it. Can we now please quietly admit that 754 is a standard way of doing floating point, but isn't the only implementation? Sure we can!

SETI@home gets 495 teraFLOPS[...]

what a waste of teraflops

SETI@home gets 495 teraFLOPS, according to this site: http://boincstats.com/stats/project_graph.php?pr=sah

Sure, it's not one supercomputer, but it still does more calculations for one purpose than any other single supercomputer can.

Hello,
I think the discovery was made by the team led by Paul Kalas:
http://astro.berkeley.edu/~kalas/index.html

Well the GP of this post is not completely off. What they may be referring to is the fact that one prominent researcher's study came to the conclusion that HIV was not highly correlated to AIDS which suggested that HIV was not the causing the AIDS. Immediately he was kicked out of the scientific community as they stated that he should be jailed for life for trying to disprove a current hypothesis (which is the point of being scientific).

Part of your claim is mostly true. A prominent researcher (Peter Duesberg) did assert that HIV was not the cause of AIDS. His claims were not based on correlation between AIDS and HIV, however, because all prominent studies show an extremely high correlation -- people with AIDS invariably have HIV. Duesberg's argument was that correlation did not equal causation.

The second part of your claim is inflammatory and inaccurate. It is possible that some individuals expressed the sentiment that Duesberg should be jailed; this would be their own personal opinions. But Duesberg was not "drummed out of the scientific community" by any means. To this day he remains a professor of biochemistry and molecular biology at UC Berkeley.

Why create your own if instead you could use the decades old s/key (http://tools.ietf.org/rfc/rfc1760.txt)

You distro might have this in packages called opie. Debian packages:
opie-client - OPIE programs for generating OTPs on client machines
opie-server - OPIE programs for maintaining an OTP key file
libpam-opie - Use OTPs for PAM authentication

Java implementations can be found eg: http://math.berkeley.edu/~vojta/opiekey.html

Sorry, took me a while to find this picture, but this is what the device looks like: http://bankslab.berkeley.edu/Projects/visualHapticSchematic.html

Why not ask for some volunteers to write your application for BOINC, a popular, open-source distributed computing platform? This would give you the added benefit of global, free computing resources for your project(s). Several people are knowledgeable about writing applications for the BOINC platform, and several others have experience writing code for other BOINC-based scientific applications and might be interested in contributing to yours.

You can thank BSD...

http://www.eecs.berkeley.edu/Pubs/TechRpts/1983/5392.html

Another useful BSD utility is here...

http://www.daemonology.net/depenguinator/

;)

I've moved to Folding@home now as I hope it will have tangible benefits. My contribution is pretty minor as I don't have the hardware for GPU processing.

If you just have a CPU, then your spare cycles would probably be spent on some BOINC-based projects. I'm especially a fan of Rosetta and uFluids. Rosetta is another protein folding program, but unlike Folding@home it focuses on predicting the final protein structure from the genetic code, rather than simulating the folding process itself. And if you think labs-on-a-chip are cool, uFluids is designing better microfluidic devices with some enormous genetic algorithm. Those are harder to speed up with GPUs, so you could do more good there. The clients are also pretty convenient.

distributed.net used to have a very vibrant community, with several projects on-going at one time. But lately, things haven't been going so well for them. The prize funds for their RC5-72 challenge were recently yanked. And the only other project they had on-going was this OGR-25 project.

Does anyone know if they'll offer further projects in the near future? Many people I know have moved on to BOINC-based distributed computing projects, instead of sticking with distributed.net.

I've always been fond of this one, which has a lower bound of 5x10^-7.

But it's pretty clear from my own (as yet unpublished) work that given a proper statistical analysis both the lower bound and the expectation value of N are 0, (or more accurately that the log of the value is negative infinity). The upper bound is about 10k. These limits are mostly set by the probability of intelligence evolving on a planet where life originates.

The main issue is that if intelligence is common, it's very difficult for us to have arrived as late in the history of life on Earth as we did. Using our history as a guide, if you ran the last 4.5 billion years over a thousand times, even if you got life here every time, you'd end up with intelligent life once, at most. And life here only has between 100 million and half a billion years left before it starts to get too warm.

I'll stick with small numbers unless someone comes up with a good study of the evolution of complex organisms that indicates that it is likely.

How, pray tell, did cyanobacteria consume carbon dioxide and release nitrogen? Biological creatures (that we know of) do not perform fission or fusion.

From an article at MIT:"Many Proterozoic oil deposits are attributed to the activity of cyanobacteria. They are also important providers of nitrogen fertilizer in the cultivation of rice and beans. The cyanobacteria have also been tremendously important in shaping the course of evolution and ecological change throughout earth's history. The oxygen atmosphere that we depend on was generated by numerous cyanobacteria during the Archaean and Proterozoic Eras. Before that time, the atmosphere had a very different chemistry, unsuitable for life as we know it today."

Chill out, we all know what he meant.

There are actually several implementations of IPv6 for very low power, embedded devices. Arch Rock (www.archrock.com) has developed a commercial implementation. At Berkeley, we've developed open-source hardware (http://www.cs.berkeley.edu/~prabal/projects/epic/) and software (ttp://smote.cs.berkeley.edu:8000/tracenv/wiki/b6loWPAN) for low power IPv6 networking. Both the Arch Rock and Berkeley stacks support robust multi-hop IPv6 routing to support a large number of devices.

There are actually several implementations of IPv6 for very low power, embedded devices. Arch Rock (www.archrock.com) has developed a commercial implementation. At Berkeley, we've developed open-source hardware (http://www.cs.berkeley.edu/~prabal/projects/epic/) and software (ttp://smote.cs.berkeley.edu:8000/tracenv/wiki/b6loWPAN) for low power IPv6 networking. Both the Arch Rock and Berkeley stacks support robust multi-hop IPv6 routing to support a large number of devices.

Can it be removed instantly without solvents and with no damage to the materials bonded together?

Yes, that's the whole point of this.
Think about a gecko, does it rip the paint off walls with each step? The bond is easily released by peeling but sticks strongly if you try to slide it, like a post-it note but stronger. Apparently it is self-cleaning too.
It doesn't bond the way glue or other adhesives do, the adhesion is from the addition of the van der Waals force from millions/billions of tiny "hairs". more info

They don't? The methodology isn't perfect, but we're not talking about calculating tolerances for a bridge or determining what hunk of tissue to excise during brain surgery here. In other words, a blunt metric is all we're looking for (and is certainly more reliable than a quote from an individual with a direct interest in promoting the language).

And it's not like there aren't corroborating sources like Google trends, Sourceforge statistics (a bit old but still relevent), book sales and so forth.

The thread model has some fundamental problems, but since they seem here to stay there are some things you should keep in mind, nicely summarized in this article(pdf).

Article also available in html if you click on the first computer.org link from google. Hmm, why does it work from google and not from slashot?

...and this might begin to answer your question but I still find it hard to understand!

http://cosmology.berkeley.edu/Education/BHfaq.html#q4

It's fair to criticize Linpack for being a one-trick pony. It measures system performance for dense linear algebra, and nothing else. Jack Dongarra (the guy who wrote Linpack and maintains the top 500 lists) is quite up-front about Linpack's limitations, and he thinks that using a single number as the end-all-be-all of a computer's performance is a bad idea. It's a simple fact of life that certian kinds of computers do better on certain problems. The good guys out at Berkeley even sat down a couple years ago and enumerated all of the problems they found in real-world HPC applications (See the tables on pages 11-12). The real truth here is that people should stop treating Linpack like it's the final measure of system performance. If you are doing pure linear algebra problems, it's a pretty good measurement for your purposes; if you are not, then you use it at your own peril.

UC Berkeley has their archived webcasts also... http://webcast.berkeley.edu/

You forgot the links.

http://berkeley.edu/multimedia/index.php

http://inst.eecs.berkeley.edu/classes-eecs.html

You forgot the links.

http://berkeley.edu/multimedia/index.php

http://inst.eecs.berkeley.edu/classes-eecs.html

Diesel contains approximately 30% more energy per volume than gasoline does.

No it doesn't. It contains between 0-10% more energy than gasoline.

"The density of petroleum diesel is about 0.85 kg/l (7.09 lbs/gallon) whereas petrol (gasoline) has a density of about 0.72 kg/l (6.01 lbs/gallon), about 15% less. When burnt, diesel typically releases about 38.6 MJ/l (138,700 BTU per US gallon), whereas gasoline releases 34.9 MJ/l (125,000 BTU per US gallon), 10% lear[2] by energy density, but 45.41 MJ/kg and 48.47 MJ/kg, 6.7% more by specific energy." "Fuel Energy Density" says automotive gasoline has a density of 34.2 MJ/litre whereas automotive diesel has a density of 38.6 MJ/litre.

Falcon

Snow plows have used magnetic guidance such as this for years now: http://www.path.berkeley.edu/PATH/Research/snowplow/

CalTrans started using a lane marker system on I-80 over Donner Pass in 1998: http://www.sfgate.com/cgi-bin/article.cgi?file=/chronicle/archive/2001/10/29/MN172839.DTL

There's still an operator, but it's not the airport shuttle either. You probably want a person to operate the blade, and make other decisions that vary with the conditions created by each storm. Keeping the operator safe (and keeping an expensive asset out of the ditch) is worthwhile, especially when time is a factor.

First, this is basically Demo '97 technology. The CALTRANS PATH people have been fooling around with this for years. I saw this around 1990 or so up at the CALTRANS Richmond test facility. Automated lane following was demonstrated in 1959 by General Motors with Firebird III.

About the only justification for this is to improve stop accuracy at bus stops so the bus can get close to the curb without scraping the tires. A bit of automated parking assistance there might be helpful. A neat trick would be to use rear wheel steering so that when the bus pulls up to the curb, the bus ends up parallel to the curb. Let the driver drive the front end, and put the back end on autopilot. This would be a big help for articulated buses, which tend to stop with the trailer hanging out in an adjacent lane, and might allow for smaller bus stop zones.

This is far more primitive than DARPA Grand Challenge technologies.

While I find the grid at Cern impressive with their claim that "Cern will be using one of the biggest computer grids this summer to pool the processing power of about 100,000 CPUs worldwide", I find the SETI project even more impressive, which according to Berkley boasts "Currently the largest distributed computing effort with over 3 million users".

Granted, Cern claims that it processing its information at 1Gbps, I wonder how that stacks up against SETI

I'm still waiting to see the "proof" of evolution that many people claim exists...out of either innocent ignorance of the facts, or intentional distortion of the truth. Surely one of you can provide a few links to some of this supposedly abundant "proof". If you, IgnoramusMaximus, have "observed the process of evolution", as you claim, then please, by all means, share some of the details of your unique experience.

I do not get paid for educating you.

However, you can start here.

As to evolution-in-progress observations, not only that site and others like it have many, even Slashdot carried stories, such as this, recently.

BORING! They could have Neural Networks, or or some upper bounded "Advanced Beginner" and acheive the same result ::

When you define perfection(tm), you can acheive it. Then you realise your perfection(tm) is not actual perfection, but some management person's project signoff of perfection.

Seems like the same old consultant $$$ trick. The dificult portion is picking to best heuristics, and is trivial to game.

Actually I did. Just how stupid are you? Try reading. it. again.

The link is there now. It wasn't when I responded to your posting earlier. Why? I don't know. My account is configured to insert the domain name after linked text (to avoid the URL trolls), and it wasn't there when I pasted your quote into my reply.

But, I went to look at your link. To be clear, WSWS.org is the World Socialist Web Site. I prefer to avoid citation of clearly partisan sources, but I read through the article and found the assertion you quoted:

The top .1% of Americans earned almost as much as the bottom 150 million Americans.

Based on other data I've seen, something didn't add up. So, I went to look at the actual report. I was only able to find the version dated through 2002, but the home page of one of the authors has the Excel data, updated through 2006. I wasn't able to find the statement in question in the report. I also wasn't able to find the NY Times article that the WSWS article cites. So, I don't know the source for the statement.

But, we should be able to verify it ourselves. Download the Excel workbook and take a look at the worksheet named "Table0". There, you can see the average income (including capital gains) for the top 0.1% is $3.7M. Multiply that by the number of families (133,325) and you get a total of $495B. In the same table, the total number of families is 148M. Dividing 495B by 74M (1/2 of 148M) yields an average income per family for the bottom 50% that must be less than $6,672, if the statement in question were correct.

The workbook doesn't contain information by percentile, for less than 90%. But, this graph was derived from Table A-3: Selected Measures of Household Income Dispersion: 1967 to 2003. The table on the same page shows the same data, and in 2003, the average income of the lowest quintile is $10,536 -- substantially higher than the implied average of $6,672 for the lowest 50% that is claimed above.

I'm not claiming that the report is in error, although there is certainly some controvery about it. However, it appears that someone's interpretation doesn't meet the smell test. You might want to take some time reading the entire report and corroborate it against other sources.

The graph from Wikipedia (derived from a US Census report) appears to support part of your claim: the gap between the 95th percentile and the 10th percentile has certainly gotten wider since 1967. The gap between the 10th and 50th percentiles also has gotten wider, although to a lesser extent. However, the gap has leveled off or even declined slightly since 1999 -- ironically since Bush 43 took office.

Actually I did. Just how stupid are you? Try reading. it. again.

The link is there now. It wasn't when I responded to your posting earlier. Why? I don't know. My account is configured to insert the domain name after linked text (to avoid the URL trolls), and it wasn't there when I pasted your quote into my reply.

But, I went to look at your link. To be clear, WSWS.org is the World Socialist Web Site. I prefer to avoid citation of clearly partisan sources, but I read through the article and found the assertion you quoted:

The top .1% of Americans earned almost as much as the bottom 150 million Americans.

Based on other data I've seen, something didn't add up. So, I went to look at the actual report. I was only able to find the version dated through 2002, but the home page of one of the authors has the Excel data, updated through 2006. I wasn't able to find the statement in question in the report. I also wasn't able to find the NY Times article that the WSWS article cites. So, I don't know the source for the statement.

But, we should be able to verify it ourselves. Download the Excel workbook and take a look at the worksheet named "Table0". There, you can see the average income (including capital gains) for the top 0.1% is $3.7M. Multiply that by the number of families (133,325) and you get a total of $495B. In the same table, the total number of families is 148M. Dividing 495B by 74M (1/2 of 148M) yields an average income per family for the bottom 50% that must be less than $6,672, if the statement in question were correct.

The workbook doesn't contain information by percentile, for less than 90%. But, this graph was derived from Table A-3: Selected Measures of Household Income Dispersion: 1967 to 2003. The table on the same page shows the same data, and in 2003, the average income of the lowest quintile is $10,536 -- substantially higher than the implied average of $6,672 for the lowest 50% that is claimed above.

I'm not claiming that the report is in error, although there is certainly some controvery about it. However, it appears that someone's interpretation doesn't meet the smell test. You might want to take some time reading the entire report and corroborate it against other sources.

The graph from Wikipedia (derived from a US Census report) appears to support part of your claim: the gap between the 95th percentile and the 10th percentile has certainly gotten wider since 1967. The gap between the 10th and 50th percentiles also has gotten wider, although to a lesser extent. However, the gap has leveled off or even declined slightly since 1999 -- ironically since Bush 43 took office.

Perhaps the GP should have included a link to "The Problem with Threads" by Edward A. Lee, IEEE Computer, 39(5):33-42, May 2006, also available here as Electrical Engineering and Computer Sciences University of California at Berkeley Technical Report No. UCB/EECS-2006-1 .