Ok, I had the trial Universal edition. If someone patched KDevelop to support DB/2, would that make it an IDE for KDE for PDE?
The proof is in the pudding
on
Mitnick on OSS
·
· Score: 1
It's easy to make a rational-sounding argument. I do that all the time. The question is whether it stands up to observation.
The sad thing is, the answer is yes.
There are plenty of commercial code verification tools (Coverity, which is a commercial version of the Stanford Checker, used with great effect on the Linux kernel, is an excellent example) and a substantial number of free tools designed for profiling and analysis (DAKOTA, KOJAK, Web100, TAHI, KTAU to name but a few) that can be mutilated/used to test for vulnerabilities.
As far as I know, the vendor of Coverity uses their checker to scan the Linux kernel from time to time, but I've never heard any mention of either Red Hat or OSDL doing likewise. Interesting, given that they actually do kernel development.
It's harder to assess who is using the free tools, but a glance at Freshmeat indicates that the number of total users is in the low hundreds. Of those, only a few are likely to be kernel developers with an interest in using the tools to debug/secure the Linux kernel. Indeed, the number is so low that the Linux Trace Toolkit has died, along with every single one of the enterprise-level event monitoring kernel patches.
Fortunately, I don't believe the number is zero. There are a few - a precious few - who do seem to be working on finding and fixing potentially (or actually) hazardous code in Linux. If I'd the cash, I'd nominate them for knighthoods for their fearless dragonslaying.
I've a copy, myself. Still have the snail mail CD & letter with trial code. My guess is that there's a bug in the timelock. I don't recall seeing the trial version available for a while, so my guess is that it got pulled, but I couldn't swear to that.
Somewhere, I've got the trial version of Informix, as well, which IBM now owns. I believe Informix for Linux has been totally pulled - I'm not even sure if they still sell Informix at all, given their efforts to push DB/2. I don't know if anyone high enough up in IBM reads Slashdot (they might - IBM has been Linux-friendly for a while now) but I'm sure there is still code from dead product lines of that kind that IBM could "recycle" into the Open Source community.
My guess is that IBM is reacting, to some extent, to Ingres being released as Open Source, as the prior Open Source databases (Postgres and MySQL) are not generally regarded as "enterprise ready" or usable at the "data warehouse" level, whereas Ingres was. IBM would undoubtably prefer people to be using their database (even if at zero cost) than a competitor. I'm not sure how many tools people developed for Ingres as a result of it being Open Sourced, but CA claimed a success. As IBM are also targeting developers, it seems reasonable to suppose that they are assuming that they can achieve similar results.
There are disorders (like Aspergers) where fMRI results are all screwed anyway. (Aspergers shows up as abnormalities in the pre-frontal lobes, other autistic disorders show up there and in parts of the mid-section of the brain.) Without some excellent baselines for assorted disorders, it will be much harder for those interpreting the results to know if they have a lie or an abnormality typical of a particular sufferer.
It is likely there are disorders which "disable" parts of the neurological response. Pathological liars who show no remorse or guilt - even using the best scientific equiptment available - may still show up nothing. Conversely, there may be disorders which abnormally trigger responses. Synesthesia, for example, routes data to completely the wrong part of the brain. If it is possible for a related disorder to shunt signals into this "lie indictator", then a lie will be declared even if no lie has been given.
These are going to be rare problems involving the most extremes in society. In fact, the very people most likely to be put through such tests. I could be wrong - I'm not a neurologist - but I'm not going to be convinced of its safety as a lie detector until it has been proven effective on people who are naturally on the fringe of society anyway.
I would point out something else here, too. This test is going to seriously screw with the insanity plea. As I said, some mental disorders are extremely visible on fMRIs - I believe acute depression is one. Prosecution psychs (who absolutely do NOT want people being declared insane) are likely to fight tooth-and-nail to not have such devices used in such cases. The data would be far more vauable to the defence if any level of insanity was shown, as juries are more likely to be swayed by pretty pictures of abnormalities than technobaffle from an expert. They also couldn't get away with accusing the defendent of copying Law & Order, as the defence would have them strapped to the fMRI in no time flat.
Prosecutors would also likely be wary of it. They want high success rates, media glory and a shot at promotion up the legal system's ladder. Anything that might show that many witnesses are liars themselves would hurt their chances. That goes double in the UK if the West Midlands Serious Crime Squad are involved.
A bit of history for those who don't know it: West Midland's Serious Crime Squad was caught altering "confessions" and witness statements after the fact, torturing suspects and other things generally considered not very nice. I believe almost 200 people were released on appeal, after that was discovered.
A bit of tech history: It was discovered by using a device that contained a magnetic resonator, along with some very fine powder that was affected by magnetic fields. I think it was iron, but I'm not certain. Anyway, the statements are all typed up and then signed at the end by the witness or defendent. Paper that should not have shown very faint depressions was, and paper that should have did not, indicating that the sheets had been added after the signature had been written.
Apparently some investigation showed that this was indeed the case, and that most of the signed statements were totally different from the statements presented in court. After that, as they say, all hell broke loose.
It is certain that corruption in the UK police runs far, far deeper than was ever discovered. It is equally certain that American police (where pay may be affected by performance, and where the poor have no legal aid to speak of, so nobody to speak for them) are far worse. Introduce a machine that can actually prove that in court, and you risk blowing the lid of the entire system.
Even if everyone is intending to play fair (ha!), the number of appeals courts ruling for a wrongful conviction will almost inevitably go up. That's going to be expensive, as most States pay up in such cases. If it turns out that such rulings are likely to be common, I susp
Let's start with Challenger, on the ground as that is the easiest. There are wires running to the cabin such that, in the event of a launch-time failure, the crew could slide down the wires to a concrete bunker which could survive a launchpad explosion. This would have required millisecond timing on everyone's part. The moment the flames came from the side of the boosters, the crew would have needed to simultaneously slammed open the hatch, shut down the main engines (rockets don't lift instantaneously and the less thrust you have, the longer it would take) and jump for the wires. No guarantee they'd have got to the bunker, or that they'd all have had time, but people have survived greater falls and even a few seconds on the wires would have given escapees a horizontal velocity that would have added to their chances of being frazzled by the booster engines.
Ok, so what about when they got to Max Q, when the breakup ocured? Tougher, but doable. The cabin in Challenger survived intact with at least one crewmember alive after the explosion. (No, I don't give a damn about the fact that some people do not consider a fireball an explosion - re: an earlier story. I take the line that a powerful outrush of hot gasses as the result of an uncontrolled reaction is an explosion, with a powerful inrush being an implosion. I would, however, agree that there is no proof they ALL survived. Adjustments made to the controls only prove that one person was concious.)
Since at least one person survived at that point, one could argue that the question becomes one of whether it would have been possible to extract any survivor(s) between the time of the disintegration and the time of impact with the water. I am going to argue that it was. It would have been hard. Very hard. And extremely dangerous. But impossible? No.
Ok, how could it have been done? There are two answers, depending on the angle you want to follow. If you assume EXACTLY the same resources and EXACTLY the same configuration (ie: no escape chutes, etc) then survivors would have needed to have opened the hatch at close to the maximum altitude (ie: when the cabin was no longer supersonic) and sky-dived. Hey, I didn't say it was going to be easy! The chase-planes would have needed to converge on the cabin during this time. They'd have had a very small window to pull this kind of stunt. They would have to get close enough at a high enough altitude that, on ejecting, they could hook up with the survivors and do a tandem descent on the pilots' parachutes.
In either case, do I think all 7 could have survived? Probably not. But even 1 survivor would have been a massive improvement.
Ok, what about adding equiptment? Well, since the booster rockets were connected with explosive bolts, all you really need to add is a guidance system in each rocket. Then, onm the launch pad, you could have jettisoned the rockets safely and escaped at your leisure. That's the absolute minimum, and again assumes people payed attention to the launchpad cameras.
After the explosion, you have two possibilities. One would be to have a parachute on the cabin, so that it could descend at a more controllable speed. The other would be to have a more shock-absorbing skin (good for surviving space junk strikes anyway) and an external air-bag similar to the ones used on the probe carrying the Mars rover. All you need is to reduce the shock of impact with the water by just enough to not jelly everyone. We're talking an instantaneous deceleration, the crew wouldn't need to remain concious or even completely intact. Oh, and you'd need a submarine with an escape hatch capable of hooking up with EITHER the door hatch OR the hatch that would have led to the payload bay.
So, there are certainly scenarios in which one or more of Challenger's crew would have lasted out the day.
Columbia is an easier one all-round. A space repair would have been impossible and I'll allow for the fact that they were in the wrong orbit
In this context, "tragedy" is being used to describe a fatal incident that was needless, entirely preventable and created through arrogance and pride. In that sense, it is very similar to "Romeo and Juliet" - the romantic aspect has nothing to do with the tragedy, the tragedy is a result of the self-serving, self-centered arrogance of the families involved leading to death after death, a chain of entirely stoppable events that nobody chooses to stop.
In that sense, Challenger followed by Columbia were of an identical nature. The chain was breakable at any time, NASA made the concious choice NOT to break it, the deaths were entirely preventable but a severe attitude problem made prevention impossible.
Oracle, DB/2, Sybase, Cold Fusion, Mathematica, Matlab, Veritas NetBackup, Alias' Maya, Pixar's Renderman Pro Server all ran under Linux.
Nobody, especially not people selling packages running into the thousands or tens of thousands of dollars per seat, is going to port a package to Linux if there isn't a market. These people are not only saying there's a market, but that that said market has a LOT of very rich people in it. It's not just students. Well, at least I don't know of any students that can afford to buy thousand-dollar software, and I certainly don't know of any that would have much of a need for an Enterprise-level database or high-end maths package. (I know "open book" exams tend to be generous, but somehow I can't see them letting you bring in a quad Xeon deshtop system with Mathematica loaded on it for a maths test...)
I think it would be good if someone ran a much more comprehensive survey, and also one that wasn't self-selecting. I also think it would be good to split the user types up a bit more - I somehow don't see the people with an interest in Maya 7 being the same people as those interested in Photoshop, for example.
Power supply units don't scale well. Double the power out will require FAR more than double the power in. Two computers with two high-efficiency PSUs will take LESS power than one computer with a single PSU that is less efficient.
Disks (and other mechanical parts) will consume a lot of energy, but you don't need to replicate every single physical disk - if the data is under two gigabytes, RAM disks should be fine. In the event of a hard drive failure, backing up off RAM disk is no different from backing up from physical disk, so what's the difference? A single SAN-based disk pack, copied into RAM on the servers, would be the least power-consuming design - especially if you powered the hard drive off except when syncing up.
It costs power to task swap, so the more active tasks there are, the more swapping (if the tasks are all being given fair time) and therefore the more CPU time is taken by kernel activity, therefore the more power is being used up on housekeeping. You should be able to reduce the power consumed by heavy kernel activity by load-balancing.
If you're going to load-balance, you don't need high-power server-rated or desktop-rated CPUs. Mobile CPUs will take less power, you'd just need a larger cluster to load-balance over. If using Linux, also look at CPUs other than Intel - many MIPS and MIPS64 implementations are pretty low-power.
Networks take power to run. There's no escaping that. Don't run more wire/fibre than you have to (that also includes not running longer cables than you need), and don't use more intermediate network devices than will get the job done properly. Oh, and don't overspec the network for a given technology. CAT6 is good stuff, but if your machines never exceed 10 mb/s on the network, you're going to lose efficiency. The "for a given technology" matters, as different technologies will consume different amounts of power for a given spec. Shop around.
Cooling systems are another mechanical system and so are necessarily power-hungry. You can't put those in RAM, however. Again, shop around. You want the best cooling power per unit of energy. This may turn out, for your system to involve having several fans on a single component. It might equally well work out that you can link ducting together such that a single fan can directly cool many components. Since the energy efficiency is what is important, go for the most energy efficient solution for your system.
Depenmding on the system, it MAY (this is not guaranteed) improve the efficiency to have a variable-speed fan, with the speed controllable by the CPU, and where all components cooled by this system have thermal sensors readable by the CPU. You can then vary the cooling as a function of both temperature and predicted load levels. (Varying according to temperature alone is useless, as the loads on the components will change faster than the sensor readings - but could change in either direction. Since the OS knows what tasks it is currently doing, it should be capable of predicting the likely loads for a much more reasonable timebase.)
Connectors are notorious for high resistance and therefore power loss. If there is something that you're unlikely to change for the productive lifetime of the computer, all power loss through all unnecessary connectors (whih are generally made from poor conductors anyway, just adding to the problem) is power you can conserve simply by improving the connection. If you insist on using connectors, make sure the wires that go to the connectors are soldered and not just held in place by pressure. Also, clean the connectors thoroughly, as buildups of oxide and dirt will increase the resistance. You WILL be better off by removing the connectors entirely and soldering anything that's not going to change in place.
Finally, the data center's power grid. You want very high voltage, very low current. (Power dissipation is proportional to voltage, but proportional to the square of the current.) The industrial powe
The cabin was intact (and designed to withstand sizable forces). In the instance, I believe it had completely separated from the rest of the orbiter, but there would presumably be easy ways to have it detatch if necessary. You'd have needed some sort of supersonic parachute system for the cabin only, then. (Essentially a classical parachute with holes, to stop it ripping up at those speeds.)
The objective of a parachute system would NOT have been to make for a soft landing in the water. It would need to be way too big to do that. On the other hand, why would it need to? The cabin had a servicable airlock and was airtight. So long as it struck the water slow enough so that deceleration was survivable, that's about all you'd need. Then it would just be a matter of escaping a submerged vehicle, which I believe NASA astronauts are specifically trained in doing.
The important thing is not that the escape system should be gentle, smooth and relaxing. It just needs to keep people alive, and even then only long enough for them to extract themselves, or be extracted, from the situation.
There may be other mechanisms, but the principle remains the same - NASA has always objected to an escape system on grounds of weight and space, but if you really go for the absolute minimal survival-only-never-mind-the-comfort escape system, it may well be within reasonable limits.
Anyone - absolutely anyone - who does not take the medication they need is being foolish. I take the meds I need, not because I regard them as safe, but because the potential results of NOT taking them would be many times worse. Odds are high I would have been dead many years ago.
If someone is unsure about the necessity or safety of their meds should discuss their concerns with a qualified person, NOT take the personal experiences of a single user of a single weblog. I earn enough that I can afford to get myself checked every so often - I've had three EEGs, two MRIs and a sleep study, and recent research would suggest I might benefit from an fMRI - but I recognize that I've been lucky from that perspective. I've also been broke for long stretches of time.
If someone is observing side-effects, record them and their severity, then forget about them for a while. If they persist, record them again. Go to your doctor with those records, so you can say "this is what happens, this is how often, these were the circumstances, what is your advice?"
There are bad doctors out there who will prescribe anything that'll shut the patient up. Those doctors are generally cheap, cynical and have long-since burned out.
More common are OK doctors who don't know everything (but feel as though they should act as though they do), but are generally good enough to fix the real problems and explain the non-issues.
There are also a few really good doctors who, when they don't know something, will give the best answer they can, research (within reason), then follow up if their previous answer missed anything important.
Oh, and there's also the co-dependent doctors who will give some answer (to please you), research in a frenzy (because they "should"), pester you to correct things, even if nothing needs correcting, then get mad at you because you "made" them do all that.
If you've an OK or a good doctor, side-effects shouldn't be a problem. They'll occur, but when they do, the doctor will either correct the meds or tell the person how to correct their behaviour. (If you take lithium, for example, you will suffer side effects from not drinking enough liquid. Almost any doctor will verbally beat up on someone for that.)
If you've a bad doctor, quit and go find an OK or good doctor.
If you've a co-dependent doctor, give them the meeting list for CoDA and -then- quit.
You mean, other than reading through the leaflets I get when I go get refills?
My personal experience with psych meds:
Lithium: It's effective, but lowers my seizure threshold. A failure to drink vast amounts of liquid could cause liver failure, kidney failure, brain damage (through seizures), or just regular death.
Zyprexa: Caused me to lose colour vision temporarily, increased blood pressure dangerously and I believe death is listed as a known side-effect.
Paxil: It's supposedly addictive, but I've not experienced that. It's known to cause suicidal thoughts in kids, though that doesn't apply to me. Doctors don't recommend it much, these days, because the list of side-effects is growing.
Neurontin: The side effects are nasty, but don't seem to apply to me. Long-term studies on seizure meds don't exist, because most haven't existed much beyond 10-15 years. (One sixth of a typical lifespan.)
Adavan: Not advised for regular use, because of potential long-term effects. I have to take the bloody stuff every day.
Ibilify: Gave me severe headaches, studies suggest that other side effects can include death. I was pulled off of it, as the risks were considered too high.
Personal experience. Probably the best basis short of a medical degree that I could have. (Between close friends and close relatives, I know 3 people who do have medical degrees, one of whom has "specialized" in psychology and neurology.)
Around that time, theory was a lot more advanced than practice. For example, there was a theory around that time that stated that seizures and some forms of mental illness were caused by malformed connections between brain cells - that all you needed to do was sever the connections and let them regrow. As theories went, that wasn't too bad.
Apparently what happened in practice is that doctors would use coat hangers or any other bits of wire they could find, and slash at the brain until the symptoms stopped.
Arguably, though, severe brain damage (through cutting chunks out or prodding them wildly with steel rods) was probably a better fate than those in Victorian asylums, which combined all the home comforts of a Soviet-era Siberian prison camp with the theraputic properties of a medieval torture chamber. At least the victims of the medical experiments were often incapable of suffering much. (Some, just not as much.)
Modern therapies for brain disorders are often highly dangerous, extremely toxic to the rest of the body, notorious for side-effects, often addictive, and many are poorly studied with completely unknown long-term consequences. That is many thousands of times better again than those who underwent the surgery.
With the newer discoveries being produced through fMRI and other next-generation scanning equiptment, I fully expect the next thirty to fourty years to produce as many radical changes to neurological treatments as the past thirty to fourty have. It'll be interesting to see how things change.
You see, Intel has a cunning plan to pepetuate Moore's Law. By making chips 45 nautical miles across, they can keep doubling the number of transistors for a very long time.
Aerogel sounds like an excellent insulator, while remaining porus. Presumably, there would be some way of using it to filter by temperature. Hmmm.
Pixar earned my contempt with killing off the Blue Moon Rendering Toolkit. How threatened can they be by free (as in beer) software that didn't even do the same stuff as Renderman? They earned my contempt further with this merger with Disney. Think about this - not long after Nemo came out, the two were at massive loggerheads over contractual and creative disputes. But give Steve Jobs a few million and he's suddenly all sweetness and light?
It seems to me that if the original disputes were real, then they'd still be real today, and Pixar's management sold out. If the disputes were attempts to manipulate, then Disney was suckered into the deal and animation fans were being used as so much bait. One way or another, ethics was busy in the next solar system.
Microsoft trying to sucker the DoJ into letting them violate the antitrust agreements is no great surprise, that they're doing it at the same time as trying to pull a similar stunt in Europe is perhaps more of one. It's hard to tell what the DoJ can or will do, given the current administration. On the other hand, the move may make the EU more cynical and more inclined to reject Microsoft's appeal. Depending on relative speed of action, if the EU does reject Microsoft's offering of "limited" code at a price and under a highly restrictive licence, the DoJ is likely to be a little tougher. Not too much - this IS election year - but enough that it won't create bad publicity from them going soft.
This is one of the very few times I'm actually more confused by TFA than I am by the Slashdot summary. So it's a text object with a semantic grep function? A collection of objects that can be used in a semantic wiki? A variant of Xanadu? A Java version of "Ask Jeeves"?
This was originally discussed in the UK in 1926 and by 1970 you could study from the material transmitted and actually earn a degree from it. (Yes, 46 years is a long time to go from theory to fact, but the British Civil Service are notorious for delaying any good ideas politicians might have.)
The biggest differences seem to be that a lot of the iTunes material is audio only, is not examinable material, but is available on the Internet. (There would be nothing to stop the BBC from simulcasting the OU material over the Internet - NASA Select does - but the BBC aren't always guilty of having much in the way of intelligence.)
I see the future of "extramural" education of this kind as being Internet-based but much more along the OU lines in terms of quality of material and the option of taking an examination at the end.
39 lectures to 461 random tunes/gossip... Sounds about the ratio most students go for. If anything, it might be a little heavy on the course material.
(I don't think anybody was seriously looking at iTunes as a rival to the UK's Open University program, where they've been doing remote broadcasts of lectures for a long time now.)
They had to put together a team for the Pentium IV - and had to use Pentium Pro people to do it. So there wasn't a Pentium III verification team? Hey, these are THEIR docs, they should know where they got the people from, and if they had no P3 verification team, I would certainly want to know why.
Next up, they decided that they couldn't verify the whole chip, that this was "beyond" the tools of the time, so they only tested the bits that they felt like testing. So much for black-box engineering. For that matter, so much for thorough testing. They also ONLY did integrated validation, not module validation, which is a big big no-no in the validation world.
They did "Random Instruction Testing" to make sure the different ways of calling an instruction worked, as it is impossible to test every possible calling path. Uhhh - crappy design. So long as a segment of logic is tested and proven, and so long as it is possible to prove that there are no side-effects between segments, you don't have to check every possible calling path for every instruction, you only have to check that each segment of logic complies 100% with the specification and does NOTHING else. This means defining ALL of your invariants and proving that they have the predicted states at all points.
If Intel wants to do stoccastic testing, then it is no wonder that they produce crappy chips, have been dumped by Microsoft (Itanium 2 support was dropped from 64-bit Windows) and are losing market share to other manufacturers. The Monte Carlo method is great for random simulations but will never be useful in producing high-reliability products.
Time is money. The more time that is spent looking things up, the more money is being spent, so the more it costs to get whatever it is done. Ergo, the most profitable way to get things done is to have maximum information to hand, because then the least time is used in learning how to be productive, rather than being productive.
Companies are there to make money. You are there to make a pay check. The pay check is pre-determined by whatever you signed onto, the profit of the company is determined solely by return on investment (over the long haul). You are not there to make the company money, that is the job of the company. If you were there to make the company money, you would be making the decisions, not following them. You'd also have a budget, as investments aren't free.
Training is an investment. You can choose to train yourself (by going to courses, etc) but the only time that makes sense is if you're getting a ROI (return on investment) such as a raise or a job somewhere else that pays enough more that you'll get the investment back and more before you'd get anything more than a cost-of-living raise as things stand.
In general, though, investments are the business of the company because they are the ones who are looking for return. No investment, no return. In practice, companies won't do this because they're cheap. It's much more cost-effective to hire someone at a dirt-cheap rate, force THEM to make the investment, but ensure they never get any return from it. Many companies will even regard training as using up vacation time (which is usually unpaid) so you get ripped off three ways at once and essentially end up paying your employer for the dubious priviledge of doing their work for them.
In another sense, since the work goes to support the national ecomony, the training ALSO goes to support the national ecomony, AND since a skilled workforce is likely to attract more jobs, I'd argue that the Government actually has a greater responsibility in paying the costs than you do. A highly skilled, highly educated workforce is far more beneficial to them than it is to you personally.
However, theory is immaterial if it isn't how things work in practice. How things work in practice is that employees have to do not only their own jobs but everyone else's job too. It sucks, it's a crappy system, it's inherently unstable and will eventually collapse, but it is the way it is done. That means that you pay for your training, whether you ever see an ounce of benefit or not.
The best I can suggest to anyone in that situation is to grab some used textbooks, download a trial version of whatever it is, and practice at home. It'll cost a lot less. You might not do as good a job at work, but if they don't care, then let them suffer with the long-term consequences. If the consequences aren't on your job description, they're not yours to worry about. Sure, that's self-centered, but copmpanies (and Governments) will never learn good conduct if ethical employees keep enabling them. Treat them like they're a drug addict. Don't enable. If you do, you just become part of the problem.
Most people like reassurance. They go towards the calming, soothing voice of the shephard like all good sheep. (Well, maybe that's a bit harsh. I'm sure there are some rebellious sheep out there.) In other words, people tend to prefer their horrors at a distance. They don't like the idea that THEY might be next.
The survey itself should be examined, though. It is very easy to put in leading questions, especially in a subject area that invites more tales of bravado than facts. It would be good if a more comprehensive survey could be done across Europe, not only looking at risky behaviour, but also looking at limits of knowledge.
eg: One antiviral used to treat AIDS in Africa was banned by the FDA in America as it is extremely toxic and rapidly kills the person taking it. The FDA is also sponsoring the use of the drug in Africa, which got some media attention recently. How many people read those reports? What is the general awareness like of the toxicity of the available treatments?
The problem with AIDS is that it isn't as dramatic as, say, Ebola, or as colourful as the Black Death. Unlike, say, Spanish Flu, the death rates are given in decades rather than days. A year, for a teenager, is forever. The incubation period for HIV is variable but 5 years is not unusual. What's five times eternity?
It's hard to get a handle on how many people are infected, or what their distribution is, but if you were to start with five hundred million, concentrated in areas that have reached one extreme or another, you'd probably be reasonably close.
Long answer: There have been plenty of studies with Rhesus Monkeys that do indeed show that injecting the HIV virus does cause AIDS. The alternative theory was devised by a French scientist whose name escapes me, but appears to have been motivated more for fame, glory and nationalism than anything. The argument is often repeated, but repeating it doesn't make it valid, it simply makes it heard more often.
With the HIV virus, we know the mechanism by which infection originates, spreads, disables the immune system, etc. There isn't a vast amount we don't know. The HIV virus took a while to isolate and sequence, but when compared to other viruses, it was damn quick.
What we don't know is the history prior to the first recorded case, whether or not a guy in England really DID somehow eliminate the HIV virus from his body (he refuses to get re-tested after he got a negative), why some people do not produce HIV antibodies when exposed to the virus (are they immune, as some claim, or is their immune system just not capable of detecting it?) and how a virus so astronomically unstable can function (one problem with producing a vaccine is that de-activated HIV can re-activate itself, becoming extremely dangerous).
Now, there are indeed cases where medical science seems to have jumped to conclusions. BSE and CJDnv are supposedly caused by prions, but infected brain tissue retains its ability to transfer the deadly agent after being cooked at high temperatures. Also, it is unclear how proteins (a prion is just a protein) can get through the stomach wall AND the blood-brain barrier in order to cause damage.
Even in this case, although there are plenty of skeptics of the prion theory, I know of nobody who is seeking to ridicule the work. Rather, they are pursuing their own lines of enquiry with some measure of dignity. That's how you can tell the good from the great. The merely good will sometimes bolster their egos by proclaiming themselves the One True Word, denouncing everyone else. The great let the results speak for themselves.
I'd expect very close to no bugs in either. The costs involved in carrying out comprehensive design analysis, specification verification and implementation verification are virtually zero compared to the cost of producing the initial run of actual silicon.
You also have to bear in mind that designs are modular and have limited connections, so N transistors is not a meaningful number - you should only be concerned with the number of modules and the number of interconnects. (eg: a 32-bit register will obviously take more transistors than an 8-bit register, but both are simply cut-and-paste copies of a 1-bit register. So long as you have the 1-bit form correct, there is no increase in complexity no matter how wide the register becomes.)
As for the interconnects - if you have N modules, you have an upper limit of !N possible interactions, if you can string any possible combination together. That's a big number, even for small values of N. But most of those don't exist. You cannot feed the output of one operation directly into the input of another. There are some special cases where there is a chain of events, but it is not something you can program with total freedom. Many operations just produce a result which is pushed back into the registers. Thus, N modules will produce only a little more than N interactions of interest. That is a much more managable number.
Then you need to consider that processors aren't "open floor plan". They are highly segmented. The term "floating point unit" literally does refer to a definable segment of the chip that is designed for floating point work. Again, from the standpoint of reliability, you can test each unit independently before doing an integrated test, so unit tests don't need to concern themselves with overall complexity or the number of other units out there.
Next up is the cost of a recall. Recalls are expensive. From a pure profit standpoint, you want to spend less on QA than you'd spend on a recall, but the less you spend on QA, the more you are likely to end up spending on that recall. The ideal is to reduce the number of potentially serious bugs to the point where any further initial clean-up will cost more than the money lost in cleaning up afterwards. Less QA than that will cost more than it saves. More QA than that will also cost more than it saves unless it expands the market (ie: the chip becomes good enough to be used in mission-critical systems such as life-support or fly-by-wire systems), but is sometimes good to do anyway for PR reasons.
Finally, not all transistors are "important". Once you know the cache algorithm works, the actual cache memory is irrelevent - memory is rarely implemented "incorrectly", it doesn't "do" anything (the active part is the algorithm), it's just heap.
With modern software verification tools, chip validation suites and the high level of understanding of microelectronics, an average of one bug for every four or five instructions is high. I would consider a chip with a third as many bugs to be only just acceptable for home use, and a thirtieth as many for operations in which any significant number of people would be put at risk. The extra cost would be minimal (compared to all the other costs) and would still be much less than the cost to Intel of the Pentium divide bug or to Transmeta of the flaws in their initial Crusoe chips.
Slashdot Mirror
Ok, I had the trial Universal edition. If someone patched KDevelop to support DB/2, would that make it an IDE for KDE for PDE?
The sad thing is, the answer is yes.
There are plenty of commercial code verification tools (Coverity, which is a commercial version of the Stanford Checker, used with great effect on the Linux kernel, is an excellent example) and a substantial number of free tools designed for profiling and analysis (DAKOTA, KOJAK, Web100, TAHI, KTAU to name but a few) that can be mutilated/used to test for vulnerabilities.
As far as I know, the vendor of Coverity uses their checker to scan the Linux kernel from time to time, but I've never heard any mention of either Red Hat or OSDL doing likewise. Interesting, given that they actually do kernel development.
It's harder to assess who is using the free tools, but a glance at Freshmeat indicates that the number of total users is in the low hundreds. Of those, only a few are likely to be kernel developers with an interest in using the tools to debug/secure the Linux kernel. Indeed, the number is so low that the Linux Trace Toolkit has died, along with every single one of the enterprise-level event monitoring kernel patches.
Fortunately, I don't believe the number is zero. There are a few - a precious few - who do seem to be working on finding and fixing potentially (or actually) hazardous code in Linux. If I'd the cash, I'd nominate them for knighthoods for their fearless dragonslaying.
Somewhere, I've got the trial version of Informix, as well, which IBM now owns. I believe Informix for Linux has been totally pulled - I'm not even sure if they still sell Informix at all, given their efforts to push DB/2. I don't know if anyone high enough up in IBM reads Slashdot (they might - IBM has been Linux-friendly for a while now) but I'm sure there is still code from dead product lines of that kind that IBM could "recycle" into the Open Source community.
My guess is that IBM is reacting, to some extent, to Ingres being released as Open Source, as the prior Open Source databases (Postgres and MySQL) are not generally regarded as "enterprise ready" or usable at the "data warehouse" level, whereas Ingres was. IBM would undoubtably prefer people to be using their database (even if at zero cost) than a competitor. I'm not sure how many tools people developed for Ingres as a result of it being Open Sourced, but CA claimed a success. As IBM are also targeting developers, it seems reasonable to suppose that they are assuming that they can achieve similar results.
It is likely there are disorders which "disable" parts of the neurological response. Pathological liars who show no remorse or guilt - even using the best scientific equiptment available - may still show up nothing. Conversely, there may be disorders which abnormally trigger responses. Synesthesia, for example, routes data to completely the wrong part of the brain. If it is possible for a related disorder to shunt signals into this "lie indictator", then a lie will be declared even if no lie has been given.
These are going to be rare problems involving the most extremes in society. In fact, the very people most likely to be put through such tests. I could be wrong - I'm not a neurologist - but I'm not going to be convinced of its safety as a lie detector until it has been proven effective on people who are naturally on the fringe of society anyway.
I would point out something else here, too. This test is going to seriously screw with the insanity plea. As I said, some mental disorders are extremely visible on fMRIs - I believe acute depression is one. Prosecution psychs (who absolutely do NOT want people being declared insane) are likely to fight tooth-and-nail to not have such devices used in such cases. The data would be far more vauable to the defence if any level of insanity was shown, as juries are more likely to be swayed by pretty pictures of abnormalities than technobaffle from an expert. They also couldn't get away with accusing the defendent of copying Law & Order, as the defence would have them strapped to the fMRI in no time flat.
Prosecutors would also likely be wary of it. They want high success rates, media glory and a shot at promotion up the legal system's ladder. Anything that might show that many witnesses are liars themselves would hurt their chances. That goes double in the UK if the West Midlands Serious Crime Squad are involved.
A bit of history for those who don't know it: West Midland's Serious Crime Squad was caught altering "confessions" and witness statements after the fact, torturing suspects and other things generally considered not very nice. I believe almost 200 people were released on appeal, after that was discovered.
A bit of tech history: It was discovered by using a device that contained a magnetic resonator, along with some very fine powder that was affected by magnetic fields. I think it was iron, but I'm not certain. Anyway, the statements are all typed up and then signed at the end by the witness or defendent. Paper that should not have shown very faint depressions was, and paper that should have did not, indicating that the sheets had been added after the signature had been written.
Apparently some investigation showed that this was indeed the case, and that most of the signed statements were totally different from the statements presented in court. After that, as they say, all hell broke loose.
It is certain that corruption in the UK police runs far, far deeper than was ever discovered. It is equally certain that American police (where pay may be affected by performance, and where the poor have no legal aid to speak of, so nobody to speak for them) are far worse. Introduce a machine that can actually prove that in court, and you risk blowing the lid of the entire system.
Even if everyone is intending to play fair (ha!), the number of appeals courts ruling for a wrongful conviction will almost inevitably go up. That's going to be expensive, as most States pay up in such cases. If it turns out that such rulings are likely to be common, I susp
There was an attempt to patch humans, but they kept putting the patches on denim instead.
Ok, so what about when they got to Max Q, when the breakup ocured? Tougher, but doable. The cabin in Challenger survived intact with at least one crewmember alive after the explosion. (No, I don't give a damn about the fact that some people do not consider a fireball an explosion - re: an earlier story. I take the line that a powerful outrush of hot gasses as the result of an uncontrolled reaction is an explosion, with a powerful inrush being an implosion. I would, however, agree that there is no proof they ALL survived. Adjustments made to the controls only prove that one person was concious.)
Since at least one person survived at that point, one could argue that the question becomes one of whether it would have been possible to extract any survivor(s) between the time of the disintegration and the time of impact with the water. I am going to argue that it was. It would have been hard. Very hard. And extremely dangerous. But impossible? No.
Ok, how could it have been done? There are two answers, depending on the angle you want to follow. If you assume EXACTLY the same resources and EXACTLY the same configuration (ie: no escape chutes, etc) then survivors would have needed to have opened the hatch at close to the maximum altitude (ie: when the cabin was no longer supersonic) and sky-dived. Hey, I didn't say it was going to be easy! The chase-planes would have needed to converge on the cabin during this time. They'd have had a very small window to pull this kind of stunt. They would have to get close enough at a high enough altitude that, on ejecting, they could hook up with the survivors and do a tandem descent on the pilots' parachutes.
In either case, do I think all 7 could have survived? Probably not. But even 1 survivor would have been a massive improvement.
Ok, what about adding equiptment? Well, since the booster rockets were connected with explosive bolts, all you really need to add is a guidance system in each rocket. Then, onm the launch pad, you could have jettisoned the rockets safely and escaped at your leisure. That's the absolute minimum, and again assumes people payed attention to the launchpad cameras.
After the explosion, you have two possibilities. One would be to have a parachute on the cabin, so that it could descend at a more controllable speed. The other would be to have a more shock-absorbing skin (good for surviving space junk strikes anyway) and an external air-bag similar to the ones used on the probe carrying the Mars rover. All you need is to reduce the shock of impact with the water by just enough to not jelly everyone. We're talking an instantaneous deceleration, the crew wouldn't need to remain concious or even completely intact. Oh, and you'd need a submarine with an escape hatch capable of hooking up with EITHER the door hatch OR the hatch that would have led to the payload bay.
So, there are certainly scenarios in which one or more of Challenger's crew would have lasted out the day.
Columbia is an easier one all-round. A space repair would have been impossible and I'll allow for the fact that they were in the wrong orbit
In that sense, Challenger followed by Columbia were of an identical nature. The chain was breakable at any time, NASA made the concious choice NOT to break it, the deaths were entirely preventable but a severe attitude problem made prevention impossible.
Nobody, especially not people selling packages running into the thousands or tens of thousands of dollars per seat, is going to port a package to Linux if there isn't a market. These people are not only saying there's a market, but that that said market has a LOT of very rich people in it. It's not just students. Well, at least I don't know of any students that can afford to buy thousand-dollar software, and I certainly don't know of any that would have much of a need for an Enterprise-level database or high-end maths package. (I know "open book" exams tend to be generous, but somehow I can't see them letting you bring in a quad Xeon deshtop system with Mathematica loaded on it for a maths test...)
I think it would be good if someone ran a much more comprehensive survey, and also one that wasn't self-selecting. I also think it would be good to split the user types up a bit more - I somehow don't see the people with an interest in Maya 7 being the same people as those interested in Photoshop, for example.
However, I guess you have to start somewhere.
Disks (and other mechanical parts) will consume a lot of energy, but you don't need to replicate every single physical disk - if the data is under two gigabytes, RAM disks should be fine. In the event of a hard drive failure, backing up off RAM disk is no different from backing up from physical disk, so what's the difference? A single SAN-based disk pack, copied into RAM on the servers, would be the least power-consuming design - especially if you powered the hard drive off except when syncing up.
It costs power to task swap, so the more active tasks there are, the more swapping (if the tasks are all being given fair time) and therefore the more CPU time is taken by kernel activity, therefore the more power is being used up on housekeeping. You should be able to reduce the power consumed by heavy kernel activity by load-balancing.
If you're going to load-balance, you don't need high-power server-rated or desktop-rated CPUs. Mobile CPUs will take less power, you'd just need a larger cluster to load-balance over. If using Linux, also look at CPUs other than Intel - many MIPS and MIPS64 implementations are pretty low-power.
Networks take power to run. There's no escaping that. Don't run more wire/fibre than you have to (that also includes not running longer cables than you need), and don't use more intermediate network devices than will get the job done properly. Oh, and don't overspec the network for a given technology. CAT6 is good stuff, but if your machines never exceed 10 mb/s on the network, you're going to lose efficiency. The "for a given technology" matters, as different technologies will consume different amounts of power for a given spec. Shop around.
Cooling systems are another mechanical system and so are necessarily power-hungry. You can't put those in RAM, however. Again, shop around. You want the best cooling power per unit of energy. This may turn out, for your system to involve having several fans on a single component. It might equally well work out that you can link ducting together such that a single fan can directly cool many components. Since the energy efficiency is what is important, go for the most energy efficient solution for your system.
Depenmding on the system, it MAY (this is not guaranteed) improve the efficiency to have a variable-speed fan, with the speed controllable by the CPU, and where all components cooled by this system have thermal sensors readable by the CPU. You can then vary the cooling as a function of both temperature and predicted load levels. (Varying according to temperature alone is useless, as the loads on the components will change faster than the sensor readings - but could change in either direction. Since the OS knows what tasks it is currently doing, it should be capable of predicting the likely loads for a much more reasonable timebase.)
Connectors are notorious for high resistance and therefore power loss. If there is something that you're unlikely to change for the productive lifetime of the computer, all power loss through all unnecessary connectors (whih are generally made from poor conductors anyway, just adding to the problem) is power you can conserve simply by improving the connection. If you insist on using connectors, make sure the wires that go to the connectors are soldered and not just held in place by pressure. Also, clean the connectors thoroughly, as buildups of oxide and dirt will increase the resistance. You WILL be better off by removing the connectors entirely and soldering anything that's not going to change in place.
Finally, the data center's power grid. You want very high voltage, very low current. (Power dissipation is proportional to voltage, but proportional to the square of the current.) The industrial powe
No, not by Nostradamus, but by the prophet John Wyndham.
The objective of a parachute system would NOT have been to make for a soft landing in the water. It would need to be way too big to do that. On the other hand, why would it need to? The cabin had a servicable airlock and was airtight. So long as it struck the water slow enough so that deceleration was survivable, that's about all you'd need. Then it would just be a matter of escaping a submerged vehicle, which I believe NASA astronauts are specifically trained in doing.
The important thing is not that the escape system should be gentle, smooth and relaxing. It just needs to keep people alive, and even then only long enough for them to extract themselves, or be extracted, from the situation.
There may be other mechanisms, but the principle remains the same - NASA has always objected to an escape system on grounds of weight and space, but if you really go for the absolute minimal survival-only-never-mind-the-comfort escape system, it may well be within reasonable limits.
If someone is unsure about the necessity or safety of their meds should discuss their concerns with a qualified person, NOT take the personal experiences of a single user of a single weblog. I earn enough that I can afford to get myself checked every so often - I've had three EEGs, two MRIs and a sleep study, and recent research would suggest I might benefit from an fMRI - but I recognize that I've been lucky from that perspective. I've also been broke for long stretches of time.
If someone is observing side-effects, record them and their severity, then forget about them for a while. If they persist, record them again. Go to your doctor with those records, so you can say "this is what happens, this is how often, these were the circumstances, what is your advice?"
If you've an OK or a good doctor, side-effects shouldn't be a problem. They'll occur, but when they do, the doctor will either correct the meds or tell the person how to correct their behaviour. (If you take lithium, for example, you will suffer side effects from not drinking enough liquid. Almost any doctor will verbally beat up on someone for that.)
If you've a bad doctor, quit and go find an OK or good doctor.
If you've a co-dependent doctor, give them the meeting list for CoDA and -then- quit.
My personal experience with psych meds:
Personal experience. Probably the best basis short of a medical degree that I could have. (Between close friends and close relatives, I know 3 people who do have medical degrees, one of whom has "specialized" in psychology and neurology.)
Apparently what happened in practice is that doctors would use coat hangers or any other bits of wire they could find, and slash at the brain until the symptoms stopped.
Arguably, though, severe brain damage (through cutting chunks out or prodding them wildly with steel rods) was probably a better fate than those in Victorian asylums, which combined all the home comforts of a Soviet-era Siberian prison camp with the theraputic properties of a medieval torture chamber. At least the victims of the medical experiments were often incapable of suffering much. (Some, just not as much.)
Modern therapies for brain disorders are often highly dangerous, extremely toxic to the rest of the body, notorious for side-effects, often addictive, and many are poorly studied with completely unknown long-term consequences. That is many thousands of times better again than those who underwent the surgery.
With the newer discoveries being produced through fMRI and other next-generation scanning equiptment, I fully expect the next thirty to fourty years to produce as many radical changes to neurological treatments as the past thirty to fourty have. It'll be interesting to see how things change.
It means he never has to put up with re-runs on television and got to escape the entire disco era unscathed.
You see, Intel has a cunning plan to pepetuate Moore's Law. By making chips 45 nautical miles across, they can keep doubling the number of transistors for a very long time.
Pixar earned my contempt with killing off the Blue Moon Rendering Toolkit. How threatened can they be by free (as in beer) software that didn't even do the same stuff as Renderman? They earned my contempt further with this merger with Disney. Think about this - not long after Nemo came out, the two were at massive loggerheads over contractual and creative disputes. But give Steve Jobs a few million and he's suddenly all sweetness and light?
It seems to me that if the original disputes were real, then they'd still be real today, and Pixar's management sold out. If the disputes were attempts to manipulate, then Disney was suckered into the deal and animation fans were being used as so much bait. One way or another, ethics was busy in the next solar system.
Microsoft trying to sucker the DoJ into letting them violate the antitrust agreements is no great surprise, that they're doing it at the same time as trying to pull a similar stunt in Europe is perhaps more of one. It's hard to tell what the DoJ can or will do, given the current administration. On the other hand, the move may make the EU more cynical and more inclined to reject Microsoft's appeal. Depending on relative speed of action, if the EU does reject Microsoft's offering of "limited" code at a price and under a highly restrictive licence, the DoJ is likely to be a little tougher. Not too much - this IS election year - but enough that it won't create bad publicity from them going soft.
This is one of the very few times I'm actually more confused by TFA than I am by the Slashdot summary. So it's a text object with a semantic grep function? A collection of objects that can be used in a semantic wiki? A variant of Xanadu? A Java version of "Ask Jeeves"?
The biggest differences seem to be that a lot of the iTunes material is audio only, is not examinable material, but is available on the Internet. (There would be nothing to stop the BBC from simulcasting the OU material over the Internet - NASA Select does - but the BBC aren't always guilty of having much in the way of intelligence.)
I see the future of "extramural" education of this kind as being Internet-based but much more along the OU lines in terms of quality of material and the option of taking an examination at the end.
(I don't think anybody was seriously looking at iTunes as a rival to the UK's Open University program, where they've been doing remote broadcasts of lectures for a long time now.)
Next up, they decided that they couldn't verify the whole chip, that this was "beyond" the tools of the time, so they only tested the bits that they felt like testing. So much for black-box engineering. For that matter, so much for thorough testing. They also ONLY did integrated validation, not module validation, which is a big big no-no in the validation world.
They did "Random Instruction Testing" to make sure the different ways of calling an instruction worked, as it is impossible to test every possible calling path. Uhhh - crappy design. So long as a segment of logic is tested and proven, and so long as it is possible to prove that there are no side-effects between segments, you don't have to check every possible calling path for every instruction, you only have to check that each segment of logic complies 100% with the specification and does NOTHING else. This means defining ALL of your invariants and proving that they have the predicted states at all points.
If Intel wants to do stoccastic testing, then it is no wonder that they produce crappy chips, have been dumped by Microsoft (Itanium 2 support was dropped from 64-bit Windows) and are losing market share to other manufacturers. The Monte Carlo method is great for random simulations but will never be useful in producing high-reliability products.
Companies are there to make money. You are there to make a pay check. The pay check is pre-determined by whatever you signed onto, the profit of the company is determined solely by return on investment (over the long haul). You are not there to make the company money, that is the job of the company. If you were there to make the company money, you would be making the decisions, not following them. You'd also have a budget, as investments aren't free.
Training is an investment. You can choose to train yourself (by going to courses, etc) but the only time that makes sense is if you're getting a ROI (return on investment) such as a raise or a job somewhere else that pays enough more that you'll get the investment back and more before you'd get anything more than a cost-of-living raise as things stand.
In general, though, investments are the business of the company because they are the ones who are looking for return. No investment, no return. In practice, companies won't do this because they're cheap. It's much more cost-effective to hire someone at a dirt-cheap rate, force THEM to make the investment, but ensure they never get any return from it. Many companies will even regard training as using up vacation time (which is usually unpaid) so you get ripped off three ways at once and essentially end up paying your employer for the dubious priviledge of doing their work for them.
In another sense, since the work goes to support the national ecomony, the training ALSO goes to support the national ecomony, AND since a skilled workforce is likely to attract more jobs, I'd argue that the Government actually has a greater responsibility in paying the costs than you do. A highly skilled, highly educated workforce is far more beneficial to them than it is to you personally.
However, theory is immaterial if it isn't how things work in practice. How things work in practice is that employees have to do not only their own jobs but everyone else's job too. It sucks, it's a crappy system, it's inherently unstable and will eventually collapse, but it is the way it is done. That means that you pay for your training, whether you ever see an ounce of benefit or not.
The best I can suggest to anyone in that situation is to grab some used textbooks, download a trial version of whatever it is, and practice at home. It'll cost a lot less. You might not do as good a job at work, but if they don't care, then let them suffer with the long-term consequences. If the consequences aren't on your job description, they're not yours to worry about. Sure, that's self-centered, but copmpanies (and Governments) will never learn good conduct if ethical employees keep enabling them. Treat them like they're a drug addict. Don't enable. If you do, you just become part of the problem.
The survey itself should be examined, though. It is very easy to put in leading questions, especially in a subject area that invites more tales of bravado than facts. It would be good if a more comprehensive survey could be done across Europe, not only looking at risky behaviour, but also looking at limits of knowledge.
eg: One antiviral used to treat AIDS in Africa was banned by the FDA in America as it is extremely toxic and rapidly kills the person taking it. The FDA is also sponsoring the use of the drug in Africa, which got some media attention recently. How many people read those reports? What is the general awareness like of the toxicity of the available treatments?
The problem with AIDS is that it isn't as dramatic as, say, Ebola, or as colourful as the Black Death. Unlike, say, Spanish Flu, the death rates are given in decades rather than days. A year, for a teenager, is forever. The incubation period for HIV is variable but 5 years is not unusual. What's five times eternity?
It's hard to get a handle on how many people are infected, or what their distribution is, but if you were to start with five hundred million, concentrated in areas that have reached one extreme or another, you'd probably be reasonably close.
With the HIV virus, we know the mechanism by which infection originates, spreads, disables the immune system, etc. There isn't a vast amount we don't know. The HIV virus took a while to isolate and sequence, but when compared to other viruses, it was damn quick.
What we don't know is the history prior to the first recorded case, whether or not a guy in England really DID somehow eliminate the HIV virus from his body (he refuses to get re-tested after he got a negative), why some people do not produce HIV antibodies when exposed to the virus (are they immune, as some claim, or is their immune system just not capable of detecting it?) and how a virus so astronomically unstable can function (one problem with producing a vaccine is that de-activated HIV can re-activate itself, becoming extremely dangerous).
Now, there are indeed cases where medical science seems to have jumped to conclusions. BSE and CJDnv are supposedly caused by prions, but infected brain tissue retains its ability to transfer the deadly agent after being cooked at high temperatures. Also, it is unclear how proteins (a prion is just a protein) can get through the stomach wall AND the blood-brain barrier in order to cause damage.
Even in this case, although there are plenty of skeptics of the prion theory, I know of nobody who is seeking to ridicule the work. Rather, they are pursuing their own lines of enquiry with some measure of dignity. That's how you can tell the good from the great. The merely good will sometimes bolster their egos by proclaiming themselves the One True Word, denouncing everyone else. The great let the results speak for themselves.
You also have to bear in mind that designs are modular and have limited connections, so N transistors is not a meaningful number - you should only be concerned with the number of modules and the number of interconnects. (eg: a 32-bit register will obviously take more transistors than an 8-bit register, but both are simply cut-and-paste copies of a 1-bit register. So long as you have the 1-bit form correct, there is no increase in complexity no matter how wide the register becomes.)
As for the interconnects - if you have N modules, you have an upper limit of !N possible interactions, if you can string any possible combination together. That's a big number, even for small values of N. But most of those don't exist. You cannot feed the output of one operation directly into the input of another. There are some special cases where there is a chain of events, but it is not something you can program with total freedom. Many operations just produce a result which is pushed back into the registers. Thus, N modules will produce only a little more than N interactions of interest. That is a much more managable number.
Then you need to consider that processors aren't "open floor plan". They are highly segmented. The term "floating point unit" literally does refer to a definable segment of the chip that is designed for floating point work. Again, from the standpoint of reliability, you can test each unit independently before doing an integrated test, so unit tests don't need to concern themselves with overall complexity or the number of other units out there.
Next up is the cost of a recall. Recalls are expensive. From a pure profit standpoint, you want to spend less on QA than you'd spend on a recall, but the less you spend on QA, the more you are likely to end up spending on that recall. The ideal is to reduce the number of potentially serious bugs to the point where any further initial clean-up will cost more than the money lost in cleaning up afterwards. Less QA than that will cost more than it saves. More QA than that will also cost more than it saves unless it expands the market (ie: the chip becomes good enough to be used in mission-critical systems such as life-support or fly-by-wire systems), but is sometimes good to do anyway for PR reasons.
Finally, not all transistors are "important". Once you know the cache algorithm works, the actual cache memory is irrelevent - memory is rarely implemented "incorrectly", it doesn't "do" anything (the active part is the algorithm), it's just heap.
With modern software verification tools, chip validation suites and the high level of understanding of microelectronics, an average of one bug for every four or five instructions is high. I would consider a chip with a third as many bugs to be only just acceptable for home use, and a thirtieth as many for operations in which any significant number of people would be put at risk. The extra cost would be minimal (compared to all the other costs) and would still be much less than the cost to Intel of the Pentium divide bug or to Transmeta of the flaws in their initial Crusoe chips.