Anyway, other major BSD-like projects don't suffer from private forks, and there's no reason to think LLVM will either.
That's kind of funny. I have a dark suspicion that Apple created LLVM *precisely* so that they could maintain a private fork concerning some premium technologies (OpenCL, GCD, H.264 all come to mind). Generally, however, maintaining a private fork is a PITA so it happens far less than GPL bigots promulgate. (Non bigots don't promulgate. That's how you tell.)
Why do you think Intel bought Kuck and Associates? To make it optimize their own code more better, or optimize someone else's code less better? KCC before it was Intelified used to really kick ass, *especially* on AMD.
Unlike GCC, outsiders can successfully provide patches and bug reports to LLVM. The process isn't so incredibly oppressive that people just don't bother.
Well, the last time I reported a edge case error in template support for GCC, it was fixed within *two days* of receiving my report. I guess the response time depends on the nature of what you report. But you have a point. A comparable edge case bug that I reported on the R users mailing list was fixed in 15 minutes by one of the core developers, so GCC does kind of suck. There was a bug I used to follow on Ubuntu that was still in triage three years after filing and had hundreds of people following, as resolution slipped from one release to the next. It's hard to speak in generalities about these things.
Finally, to all those people who think that only the generated code speed matters, grab a brain! The rate limiting factor on software quality and complexity is the tooling ecology. For example, suppose some day Eclipse CDT adopts the LLVM C++ parser and error message components. If this advantage were to enable me to write more powerful algorithms, *even if* the generated code under LLVM was 10% slower than some other compiler, my superior solution would still kick ass.
If the tooling advantage of LLVM is *already* making it more productive to add complex optimization passes to LLVM, somehow I doubt that 10% penalty continues to exist much longer.
The most important feature in a compiler is making the software developer smarter and more efficient.
How did that retort get moderated insightful? It's far more clueless than the post he's responding to, which as least has its heart in the right place. Every second podcast at Econtalk has a long seventh inning stretch on a Hayekian view of capitalism cut _exactly_ from this mold.
If you're taking the grand view of what capitalism requires, small government is not on the list. Twenty years ago it used to be said that Russians understood capitalism better than Americans, because they could actually define it, and list the institutions it entails (in a negative light).
These days no one actively debates the grand view of capitalism. The active debate is about capitalism as a mainspring of wealth creation and the role of government to A) abet or B) hinder the golden goose. In the blue trunks: free market fundamentalism. In the red trunks: liberal society and justice for all.
Its a dearly held tenant of the invisible-hand contingent that markets are able to solve allocation problems though the pricing system that a centralized system could never properly manage, because the required information can't be collected at a central point, unless one waves a magic wand to approximate the utility function of people not present to speak for themselves. That kind of sucks.
It was Stiglitz who showed that the magical ability of markets to solve allocation problems through the price mechanism breaks down under conditions of asymmetrical information. *If* you have price transparency (and a few other things) markets can do an excellent job where government can't.
What you end up with is a system where the vigorous new enterprise favours price transparency (which permits greater economic mobility) while the incumbent corporations do everything in their power to debase price transparency (telecoms industry, media industry, to name just a few).
I don't trust the views of anyone who doesn't think that information transparency leads to a more effective and vigorous market economy. But then I believe that wealth should be earned rather than squatted upon. I know, it's a radical idea.
I was reading some commentary on the media business, including How to Save the News which is interesting, but didn't impress me. One of the articles mentioned Bertrand competition, which suggests that in the absence of product differentiation, the product will end up selling at marginal production cost. (I'm not an economist, so sue me if I didn't get that phrase quite right.)
The Atlantic article goes on an on without mentioning the core point: why do people volunteer themselves to have their purchasing preferences manipulated by visual images in the first place? If ad revenue represents 80% of a newspaper's income, how does the effect the nature of the story reported? Is it to inform the reader, or to create a warm context for associated display ads? The theory of advertising impressions is that you get the viewer into a receptive emotional state, and then burn your image into the viewers amygdala while under the influence of the warm glow. Hence all the Superbowl ads, which are beamed at men awash in vicarious sexual potency. Not such a good model for funding an insightful report on genocide in Somalia.
I'm all for a world with far greater price transparency. It would weed out many of the people who wish to live fat lifestyles without ever creating much of value. Opportunities for value creation have never been better. Personally, I wouldn't mind seeing more of the carpet baggers bagging carpets until they change their ways.
I think a marketplace which maximizes informed choice on *both* sides of every transactions could work small economic miracles. Big business believes in such a market until they don't. Big business believes in small government until they require a big bailout. This is just wealthy peopl
In fact, knowing the size of the flow is kind of crucial to stopping it.
The biggest point you make should be the most obvious: We aren't doing what we have already have proven to work, boom and capture.
I thought the premise is that first you have to know the size of the problem, before you can determine if the previously successful solutions are applicable or not.
If "what worked before" is irreproachably scale invariant, then BP is right in their conjecture that there's no scope for data to drive a more effective response. Wasn't part of the problem with Katrina more water than pumps?
What's completely stupid about the BP position on this is that there's no contention between assessing the scale of the problem and mounting the most effective response. People who aren't working on the clean up can work on assessing the data.
What's the downside to enlarging participation? The only aspect of the response that would be stretched thin by this approach is spin management by the BP PR flacks, who would view having a disaster *and* open data to contend with as too much like Germany fighting a war on two fronts.
For what purpose? The entire PR effort could be replaced with a simple pro-capitalist capitulation: "We screwed up, big time, and unlike the banks, we're solvent enough to pay restitution." As a sop to their shareholders they could also add "and we're going to litigate the hell out of our partners in crime after the tar settles". Capitalism is not such a bad system when the gears are allowed to mesh.
I've done a lot of embedded development, some of it in the security sector, which is not tolerant of sloppiness.
One of the products had a web interface implemented in PHP, which I partly coded. I don't use PHP regularly, so I was entirely at the mercy of the PHP online documentation. This was a few years back, but my god was it a struggle to glean from the documentation the bounds of validity of the different PHP function calls. In many cases I had to resort to untangling a long and often contradictory discussion thread trailing at the bottom of the API documentation. I was brutally easy to get my C code right because I was coding against APIs which I rigorously understood. I never felt half as confident in my PHP code.
In several cases, the PHP documentation should have just said "this is exactly what you have to do to properly escape input into this function" and it didn't, not even close.
Because the documentation sucks, it's impossible to be rigorous coding any acceptable speed for your typical web application. No one wants to pay people to code web applications on slow and methodical embedded application time frames.
Ultimately what drives bug are high expectations and sloppy API documentation. Sloppy documentation of APIs is greatly enhanced if the API itself was stupidly designed. Good design leads to simple documentation.
One more example. In my embedded work, when I have to ensure that output from one module is valid as input to the next module, I only use positive filters: I match on known good patterns. Never filter by subtraction of suspected evil, you'll end up missing something. If your filter is too conservative, you'll find out about it, from a bug report, rather than an exploit. Do PHBs running web development shops accept that trade off? Or is the attitude to err on the side of uptime, security be damned?
From my notes: Joseph Traub, Mihaly Csikszentmihalyi, David Shaw, David Myers, Milford H. Wolpoff, John McCarthy, Philip Brockman, Howard Rheingold, Samuel Barondes and John Rennie all put forward the scientific method or some variation on scientific progress, alongside all the usual votes for the pill, the printing press, the a-bomb, digital electronics, and hay.
Around the time of that survey, a lot of people convinced themselves that this new model for managing a company (or a software product line) made sense, because that was how the world worked now. No, actually. Even in the dotcom stampede, bad management was bad management. This will hold true of science as well.
The scientific method is way too important in the history of modern civilization to have the IPCC make lite of this tradition in order to win a political grudge match.
Too bad Wolfram used up the title "A New Kind of Science". We could have saved it for IPCC committee reports with 2,500 eminent signatures.
I say this with full conviction that the central human activity of the 21'st century will be paying the piper for high living. I still harbour a dim hope that we'll pay off these days less rashly than we entered into them. Yes, I can see it now. This will all come to pass through a non-contentious political process involving scientific walled gardens, incestuous peer review, and sanctified data hoarding.
As a footnote, I read this the other night and was quite impressed with it. This will mostly appeal to slashdotters with low digit IDs and mild Aspergers, if such a creature exists.
Note that he takes a long view of science as I do. The key slide that just popped into mind is slide 49 with the text:
However, carve a chunk from it, say the object part, and we can talk about the motion of the hand CAUSING this light ray to change angle.
The precautionary principle is fundamentally interventionist. However, the focus of precaution is necessarily a human construct, which depends upon how the image is sliced. This claim is heavily supported in the presentation as a whole.
This insight courtesy of Judea Pearl, who is becoming known as one of the giants of AI. He's a major influence on the recent work of Daphne Koller. Under no circumstances check out the accomplishments of Daphne Koller if you're feeling low about your productivity in the recent week or decade. She's just polished off a nice 1,200 page tome http://www.amazon.com/Probabilistic-Graphical-Models-Principles-Computation/dp/0262013193/">Probabilistic Graphical Models: Principles and Techniques (Adaptive Computation and Machine Learning). I'd rush out to buy this, but I'm not sure my ego can handle the blow.
We agree with the CRU view that the authority for releasing unpublished raw data to third parties should stay with those who collected it.
In the ordinary scheme of things, where science proceeds at a slow and deliberate pace, and the stakes rarely exceed ego and pride and lifetime accomplishment, it would be fine to allow scientists to manage their walled garden as they are accustomed to doing.
In the extraordinary scheme of things where fluctuations so minor they are hard to measure are beating the drum on global policy, and recourse to sober reflection has been staked through the heart with language of imminent "tipping points", this is not good enough.
According to the respected tradition of science, the scientists will soon come to a sober and reliable consensus on AGW 1980-2010, where soon is somewhere between 2050 and 2100.
Cripes, Einstein wasn't awarded a Nobel prize until 1921, sixteen years after his annus mirabilis, roughly the equivalent of winning all four majors in the same year by ten strokes each. Tiger never had a year that good. Gretzky never had a year that good. At the pace that science traditionally moves, Gretzky would be a recent induction to the Hall of Fame, after a sober cooling off period, rather than handing his final game jersey and jockstrap to a burly Scotsman handcuffed to a velvet trunk. The tradition in science is to allow the sweat to dry before inviting posterity to drop in for a look see.
Now that the stakes are so high (apparently), Big Science has slapped 2,500 signatures on a fat report in total contravention of every word of wisdom in Brook's "Mythical Man Month": you can't accelerate solid results by piling on resources or amassing "looks good to me" chits by the basket load. It doesn't work in software, and it doesn't work in science.
If they think this is a viable route to sober consensus, they deserve a level of outside scrutiny that would make NASA blush.
Where did this idea come from that science can function as a miracle short-order chef, just because it has to? I hate to parrot Thomas Kuhn, but he did point out that error in science is weeded out of the system on a generational time scale. Personally, I don't see a huge difference between Harry Markopolous and Stephen McIntyre, Both have a long history of having a sharp eye for dubious claims.
If science thinks they can produce results on a Wall Street time frame, welcome to Wall Street scrutiny, er, I mean the scrutiny we now wish had been in place before it was too late.
Right now we have a natural order and an economic order that are working at cross principals. We can't apply the precautionary principle on both sides of the fence simultaneously. So the debate devolves to which of the two should we F with, and in what proportion? Clamping down on the current economic order out of fear that an effect is major and immediate rather than moderate and mid-horizon has real effects on global standards of living.
It's possible that the environmental scientists are right, which would make their urgency good politics, but still not good science, not until the day that fully disclosed data subjected to the best possible analysis fully supports their conclusions.
Maybe this issue is too important to wait for good science. Nevertheless, under no conditions am I buying into this revisionist agenda as to what good science looks like. If we have to turn to lame science, and that's the best we have, so be it. Meanwhile, I'd be quite pleased to dispense with all this faux identity maintenance by scientists who have clearly overstepped the tradition of sober reflection, no matter if for the best of reasons.
People always say this. So, where's the list of all Congressmen with net worth of more than a billion dollars?
Ah, reminds me of the old joke about the priest who skives his responsibilities to play a secret game of golf, then God punishes him with a hole-in-one, on the grounds "who can he tell?"
A congressman selling influence isn't going to be paying tax on the transaction. He might be rich as stink offshore, but who can he tell?
This is why society develops secret societies with secret handshakes, so that old men can gather together and speak in veiled and cryptic phrases about how to best invest funds that don't exist.
I'm not saying there are any billionaire congressmen out there, but I am saying that your standard of evidence has a Marcos-sized hole.
I didn't say there wasn't still a gap, I said the gap has been significantly reduced.
I haven't looked at the numbers closely, but I suspect the gap hasn't been reduced at all. ATI's new Evergreen is kicking out a trillion SP FLOPS and now has full IEEE double precision as well. After ignoring graphics for years, OpenCL has caught my fancy. I once hoped that IBM would kick out an upgraded version of Cell with fast IEEE double precision, but their unholy alliance with Sony proved to be quite the fiasco.
What has greatly changed is the relevance of the gap. When you're a student living in a 300 sq ft apartment, 600 sq ft is a screaming upgrade. Later in life, the upgrade from 2500 sq ft to 5000 sq ft has a narrower appeal. Good if your favorite game in life is playing indoor hide and seek with your grand-toddlers or you're a world class model train builder.
The people who are happy enough with the tower PC they already have are not going away. Like Jobs predicts, it's not a sector capable of supporting the high living Apple desires. Everyone's creative energy is going to be poured into finding the bisection point between mobile and cloud. Everyone with aspirations for high living is piling into this sector. It won't be cheap. HP is going to congress to ask for a $50/month levy for virtual ink on every user of cloud computing services.
On the other side of the coin, the boring old sector will fade from mind a lot faster than it will fade from reality. I predict large PCs will fade away about as quickly as large SUVs in America. If you look at the dealerships, you might come to one conclusion. But then look on the road around you. Sit in your Smart Car in heavy traffic some day and count the number of bumpers at eye level before declaring dinosaurs extinct.
After writing such a nice screed about Flash, Jobs won't have any difficulty understanding why Google might wish to undermine H.264 with V8.
The interesting thing about the computer industry these days is that control points have less to do with de facto product monopolies, and a lot more to do with physical embodiment. Running a huge data center, the pendulum swings toward the ability to maintain trade secrets, and I suspect there's some of that in the closed device mobile sector.
In this new world, the incumbent monopolists have an alternative to circling the waggons and supporting each other's misbegotten prominence. Steve's essay is an excellent rendering of the king is dead, long live the king.
Embarrassing to be Canadian. Well, at least the Pope isn't Canadian, although to hear what he had to say about the internet, he might be on board this paper thing.
What the Domtar dipshit didn't take into account is embodied energy. People tend to have wacky ideas about embodied energy, unless you bother to work the numbers.
In the modern world, where nearly every problem has a collaborative dimension (even rootkit developers exchange tips of the trade) the phrase "excess collaboration" reminds me of the line from Jack Nicholson's character in One Flew Over the Cuckoos Nest, "Who ever heard of a man getting too much poozle?"
[http://www.mtc.mb.ca/WorkArea/linkit.aspx?LinkIdentifier=id&ItemID=4786 McMurphy spars with the untroubled Dr Spivey]
It makes you wonder about the rampant NIH that has long existed in the software profession.
Maybe they should instead be teaching students to *never* work from scratch if they can source a more suitable starting point. Maybe they should deep six sitting in front of a droning professor. You can do that online these days from the close proximity of your residence bar fridge.
Maybe nearly 100% of computer class time should be having the students conduct group walk-throughs of proposed designs, algorithms, and implementation fragments. Maybe the CS professor should be assessing how the students are interacting rather than mindlessly reciting prepared solutions (if you can find enough CS professors with a facility for eye contact).
Maybe an obstacle-course education in emergency wheel-reinvention should be regarded as a commando level training option for elite troops only.
Maybe the curriculum should include a course in outsource management, where your project group is graded on the work quality of its very own team of docile and compliant Indian math geeks.
Maybe the education system as it now stands is part of the problem, rather than part of the solution.
I also agree this is a futile quest. While it might be possible for an enlightened company to profit from open source, Oracle is not that company, and is highly unlikely to become that company no matter what argument is advanced here.
"We believe that [20 GB/cm^2] is at least a factor of two better storage than flash memory will [] have [three years from now]"
I wouldn't be too thrilled with that proposition. This is an ambitious new technology introduction. It won't take much for that time frame to slip by a year or two, in which case your edge is shot and your price structure is unlikely to be competitive.
Suppose you get 5% initial market share where the difference in feature set is "just right" for some set of early adopters. Now with 5% of the revenue base, your business requirement is to scale faster than a mature competitor sitting on 95% of the revenue base, funded by infinitely deep pockets (Intel, to name one).
Even if this new technology is fundamentally sound and has scaling headroom to burn, it could easily be a decade before it pulls into the passing lane once and for all. Maybe there's a 10% chance all the arrows line up and it arrives on schedule three years from now cheaper, denser, and faster than existing flash. That prospect alone is enough to fund a serious market push, if the up front R&D is not outrageous.
OTOH, if it arrives late or expensive or slow in performance, this could play out more like the Thrilla in Manilla, with one or the other of the heavyweights collapsing on their stool at the end of the 14'th round.
Time to check out the spot price on Froogle Futures for April 2013. All I'm getting is "HP 404". What does that mean?
So according to this logic every C programmer must be lazy and stupid because you can not find a single, non-trivial program...
Well, the bad news is that you failed statistics, but the good news is the AI department called and wants you back. Since AI is actually making some progress these days, they're at risk of outstripping trivial. After 50 years it would be a shame to sneak over the bar by accident now that HAL is merely another half century down the road. With your help, history will record that HAL is the first ever non-trivial AI application.
In the late 1980s I was lead architect on a non-trivial software package written entirely in C. We took some prudent measures to eliminate pointer abuse, both in terms of architecture and coding style. The number of times our team make a pointer mistake in a release build was running roughly neck and neck with the number of times we reported a code generation error concerning the Microsoft or Watcom C compiler. I pretty much forbid the use of CodeView until it was a coin flip whether it was the code or the compiler.
Of course, we did have assert statements going off on a daily basis during heavy coding. We also made a lot of typing mistakes and sometimes we ran our build scripts with the wrong arguments. Doh! Yet after all that, most software projects still use keyboards and build scripts. Will we ever learn?
What's the rationale for Goo against the use of assert() statements? Having eliminated pointers, it's no longer possible to violate a post-condition? Or does Goo include a magic eight ball to decide which developer should first tackle a code failure, having broken their side of the interface contract?
That was the whole point of the assert() discipline, to create a social mechanism by which the person who screws up is most often the person who fetches the mop and bucket.
This is a response to my own post. Sometimes after uncorking a minor screed, I note to myself "that was more obnoxious than normal" and then my subconscious goes "ding!" and I get what's grinding me.
The secret of x86 longevity is to have been so coyote-ugly that it turns into pablum the brain of any x86-hater who tries to make a chip to rid the planet of the scourge once and for all.
For three decades right-thinking chip designers have *wanted* x86 to prove as bad in reality as ugliness ought to dictate.
Instead of having a balanced perspective on beauty, the x86-haters succumb to the rule of thumb that the less like x86, the better. And almost always, that lead to a mistake, because x86 was never in fact rotten to the gore. You need a big design team, and it bleeds heat, but all other respects, it proved salvageable over and over and over again.
On the empirical evidence, high standards of beauty in CPU design are overrated. Instead, we should have been employing high standards of pragmatic compromise.
If any design team had aimed merely for "a hell of lot less ugly", instead of becoming mired in some beauty-driven conceptual over-reaction, maybe x86 might have died already.
Maybe instruction sets aren't meant to be beautiful. Of course, viewed that way, this is an age-old debate.
If the 1.8GHz Xeon was based on the Netburst architecture, first you have to multiply by 2/3rds to correct for diet Pepsi clock cycles, then if your code base is scientific, you have to divide by two for the known x86 floating point catastrophe, and finally, if your scientific application is especially large register set friendly, there's another factor of 0.75. So on that particular code base, a 1.8GHz Netbust is about equal to a 400MHz Alpha (I only ever worked with the in-order edition). Netburst usually had some stinking fast benchmarks to show for itself if it happened to have exactly the right SSE instructions for the task at hand. And it gained a lot of relative performance on pure integer code. BTW, were you running Xeon in 64-bit mode? That could be another factor of 0.75.
A lot of people, myself included, think itanium should never have existed, and that the development effort should have been put into alpha instead - an architecture that already had a good software and user base
Yeah, you and a lot of clear headed people with insight into the visible half of the problem space. Not good enough.
Alpha was a nice little miracle, but it fundamentally cheated in its fabrication tactics. This is a long time ago, but as I recall, in order to get single-cycle 64-bit carry propagation, they added extra metal layers for look-ahead carry generation. For a chip intended Intel scale mass production, this kind of thing probably makes an Intel engineer's eyebrows pop off. That chip was tuned like a Ferrari. I'm sure the Alpha was designed to scale, but almost certainly not at a cost of production that generates the fat margins Intel is accustomed to.
Around the time Itanium was first announced, I spent a week poking into transport triggered architectures. There was some kind of TTA tool download, from HP I think, and I poked my nose into a lot of the rationale and sundry documentation.
TTA actually contains a lot of valid insight into the design problem. The problem is that Intel muffed the translation, through a combination of monopolistic sugar cravings, management hubris, and cart before the horse engineering objectives. I'm sure many of the Intel engineers would like to take a Mulligan on some of the original design decisions. There might have been a decent in there somewhere trying to get out. Itanium was never that chip.
I pretty much threw in the towel on Itanium becoming the next standard platform for scientific computing when I discovered that the instruction bundles contained three *independent* instructions. They went the wrong way right there. They could have defined the bundles to contain up to seven highly dependent instructions, something like complex number multiplication: four operands, seven operations, two results. It should have been possible to encode that in a single bundle. Either the whole bundle retires, or not at all.
Dependencies *internal* to a bundle are easy to make explicit with a clever instruction encoding format. You wouldn't need a lot of circuitry to track these local dependencies. What you gain is that you only have to perform four reads from the register file and two writes to the register file to complete up to, in this example, seven ALU operations. Ports on the register file is one of the primary bottlenecks in TTA theory.
What you lose is that these bundles have a very long flight time before final retirement. Using P6 latencies, it's about ten clock cycles for the complex multiplication mul/add tree in this example (not assuming a fused mul-add). This means you have to keep a lot of the complexity of the P6 on the ROB side (retirement order buffer). But that also functions as a shock absorber for non-determinism, and takes a huge burden off the shoulders of the compiler writers. This was apparent to me long before the dust settled on the failure of the Itanium compiler initiative.
In my intuitively preferred approach, instructions within bundles would be tightly bound and s
People, take a quick bunny hop through List of cognitive biases and ask yourself how many of these constitute cognitive feeder arteries for Godwin's law.
Every so often a topic comes up where everyone simultaneously decides to let their amygdalas off the leash in the same dog park, who all quickly pair up for a circular open-jaw square dance. After the dust settles, what do you have? A giant patch of lawn to circumvent until the next heavy rainfall.
I suppose the game is a bit more earnest for the born complainer, who does have to somehow realign the pole of supreme evil with every successive regime change. Those of us with the attention span to get through the first page of Anna Karenina understand that evil has a multitude of poles, any one of which can erupt into the supreme pimple in the rapidly shifting context of real life.
if anybody shakes MS loose from that, my bet would be Google rather than Apple
There are at least half a dozen major players tugging on different fingers of the Beast of Redmond. Sony is doing their ineffective best tugging on the pinky finger with their once-powerful PS3 franchise. The unholy alliance Snoracle has a firm but limp-wristed grasp on the middle finger on office suite revenue streams. Linux/Apache/Firefox inflicted a hairline fracture on a wristbone. Google extracted a fingernail from the ring finger when it became the ultimate talent drain. That had to hurt. And now they're proceeding to bend back the index finger by sucking up the air supply in online search. Learn from the best. A horsefly named Gnome Evolution landed on the thumb and carted off the largest divot of flesh it could manage, which considering all the other wounds, is of no real consequence whatsoever, unless horseflies are a vector for Ebola, and so far it appears that they aren't. All things considered, I think that Microsoft can hang there by their relatively undamaged, enterprising thumbs for another thirty years or so.
The biggest risk with Apple is that they manage to leverage their carefully cultivated charisma (if it survives their having become a big enough company to matter in these discussions) to make DRM palatable to the masses.
Sony is far more evil in the DRM department (witness the recent "other OS" rescindment fiasco) but they suffer from a bad case of cartoon evil: whatever their grasping ambition, it's soon equally matched by their incompetence. They managed--on the back of a half billion dollar war chest--to leverage their dominant Play Station franchise into a slow and lukewarm victory in a dying physical media platform.
This rivals anything accomplished by the Hudson Bay Company (oldest corporation in North America) which once laid claim to half the natural resources in Canada, but decided the crown jewel was retailing dress shirts. If Warren Buffett had gained control of HBC in the late 1700s, America might now be the 11'th Canadian province, or an economic protectorate, like Puerto Rico. (If BG gained control of the HBC in the late 1700s, Russia would now be the world's great democracy and white knight of freedom.)
Wish the Sony/HBC disease were true of Apple, but it isn't.
I could continue grave digging in this vein for another day or two, but hey, it's Easter, and whatever your opinion on the back story, there was an important lesson in there about the rush to judgement.
I can't play 'tube on this box. It's likely segment 11 or 12 where Feynman gets into a multitasking competition with statistician John Tukey, and discovers that they can multitask different things, if no-one around here is patient enough to sit through the whole thing.
Just what did he think he was doing wasting all this time figuring this stuff out 40 years ago before anyone cared?
Broken URL syntax in the previous post brought to you by my favorite Fireflog plug-in Make Link. Which is why Boomtango has the slow horse in this race.
Are you related to Bill Clinton, and learned to parse words to death?
From [http://www.nytimes.com/2010/03/26/opinion/26krugman.html Paul Krugman - Going to Extreme]
For today's G.O.P. is, fully and finally, the party of Ronald Reagan -- not Reagan the pragmatic politician, who could and did strike deals with Democrats, but Reagan the antigovernment fanatic, who warned that Medicare would destroy American freedom. It's a party that sees modest efforts to improve Americans' economic and health security not merely as unwise, but as monstrous. It's a party in which paranoid fantasies about the other side -- Obama is a socialist, Democrats have totalitarian ambitions -- are mainstream. And, as a result, it's a party that fundamentally doesn't accept anyone else's right to govern.
Bill is far from the first charismatic leader whose tallywhacker has gone walk about. What was underlying the whole charade was a dispute over Clinton's right to govern. He was elected by a majority decision, for better or worse, and the republicans failed to respect this.
What exactly is the correct way to parse a question that never should have been asked in the first place? The correct answer was "what does that have to do with the price of tea in China?"
"Mr Limbaugh, are you addicted to Hillbilly Heroin?" What's he supposed to say? Sure, he's a two-faced lout, just like Clinton, but it's still an ad hominem attack, and undeserving of a straight answer.
Does that work? I guess it depends on your attitude toward natural pain relief over artificial pain relief.
On a technical note, this is the very first time Boomtango helped me to locate something I've recently read on the internet faster than I could have done it myself. And for this, all day my Firefox is an extra beat sluggish. For now, I'm sticking with it on potential.
Since you don't know what this amount is you have to use a proxy. Oracle uses features, the number of cores and ram as their proxy.
This is a valid recitation of economic orthodoxy which unfortunately leaves off right at the moment where thinking begins. Ships are lost on this basis.
The problem is that the choice of proxy has downstream consequence which can destroy a lot of value, or in the worst case, almost the entire value of the sales proposition.
In the case of licensing by core, it introduces a huge non-linear term in right-sizing infrastructure. I can't stand this stuff myself, but some of my closest friends have made an excellent living showing up to solve the Oracle license fee non-linear optimization problem, which contains large elements of uncertainty and non-determinism because the outcome depends on unknown future events.
Anyone with the least insight into systems theory knows that non-linearities are like a sexual disease. They have a noted tendency to give on giving. Properly understood, the effort involved in damping out these non-linearities can easily exceed the value proposition of adopting Oracle solutions in the first place.
Fortunately for Oracle, there's a huge real world shortfall in the quantity "properly understood". Microsoft, among others, makes a mint from truncated TCO studies. The assumption underlying every TCO study I've ever seen is that the higher order non-linearities can be safely neglected. If that were true, why is anyone relying on a vendor-funded TCO? In the case where the higher order terms can be safely neglected, it's usually easy enough for the customer to work their own TCO on the back of napkin, and get an immediate answer everyone immediately believes.
The cases where this breaks down are the sales propositions absolutely freighted with non-linearities, to the point where no one trusts their own numbers, surprise, surprise.
The first rats off a sinking ship are the best swimmers. The first wave of people to abandon Oracle are those outfits with a larger than average insight into "properly understood".
From [http://www.riskglossary.com/link/barings_debacle.htm Barings Debacle]
In November 1993, BSL was merged into BB&Co. in anticipation of a subsequent initiative to form a Barings Investment Bank (BIB). The merger was not easy because the two firms had markedly different cultures. It was a distraction right in the middle of Leeson's tenure at BSL. ... Barings was just starting to form a risk management function. Risk controllers were appointed in London, Tokyo and Hong Kong during 1994, but not in Singapore. ... As part of the 1993 reorganization, Barings had adopted a "matrix" approach to management of its offices.... Employees complained that lines of reporting were not always clear.... Another issue was that Leeson was an accomplished liar.
Every aspect of this is situation normal at most medium or large companies. You best executive attention is devoted to various political fires. Many organizations are just too busy with other pressing demands to step back and engage in the kind of clear thinking it takes to put out the Oracle fire. So what if the Oracle pricing model induces non-linearities? We're planning to auction block that division anyway.
However, in the companies where choosing the right database and the right database architecture is the dominant fire, non-linearities associated with proxy pricing models can escalate into a serious business concern. Some of those people will go talk to Oracle and try to cut a special deal. Many of them won't. An attrition sets in.
Look what happened to Microsoft when Google became the hot job opportunity. That sucking sound is your technical clout packing family photos into their briefcases.
A major coming of age event in commoditization of a technology is crossing the threshold where the price proxy shifts from being a
That's a dialect of Canadian known as Hansardese. Hansardese was actually a staple of the early machine translation efforts, since you could be quite certain that Hansardese of fuddle duddle in French was an immaculate parallel translation. Not sure it was such a good idea to prime machine translation on empty calories, though.
These multinationals have got a nice thing going. "But mommy says we have to ratify!" Works every time.
In some of these open source projects, version 1.0 is like the first time the odometer in your car rolls over. Or like a couple who finally decide to get married after 15 years of living in sin. I wonder if this big decision involved a trip to Vegas.
Version 1.0 isn't that different from getting marriage. Some enter into it on the basis of hope and enthusiasm with neither experience nor skill, while others circle each other like planets in a decaying orbit.
A long run in the zero point nineties is like the people who are technically married, but have not yet escaped their parents' basements, lacking either the spirit or means of independence.
Then comes the bold and tremulous day when they finally cut the apron strings, while everyone stands around in state of genuine micro-perplexity going "I had no idea".
I read the other day that the dung beetle has been discovered to be one of the world's strongest organisms by body mass. I've never seen a single dung beetle toting a Champagne magnum. It's clear they can't get the cork out. Or maybe no one has figured out how to make the bubbles small enough to fit in the bottle.
One of the best things I've read on this subject was an interview with top level muck muck at NASA (I think it was _the_ guy) when NASA was bent on fixing their safety culture, following a little bug that cascaded into millions of small pieces. In a safety audit they found a small number of broken wires in some electrical harnesses that weren't supposed to have this defect in this quantity. The engineers did what I might term a sensitivity analysis and determined that with all the other layers of redundancy, there was essentially zero chance this would lead to a major incident.
The big muck muck decided to postpone the launch regardless, because, in his words, "of what the wires were trying to tell me". The wires were trying to tell him that their statistical models concerning production quality were incorrect one way or another. If you quality process is broken, this is something you want to know before you let the big dawg ring in the new year.
The second question you need to ask is this: is the root cause behind the bug forgivable? Or was the programmer responsible (or just as likely, the culture created by the management team) unable to free itself from a wet paper bag? Does the bug indicate a state of crisis in personal topology assessment? Is this the first visible black buboe in a software release candidate rapidly progressing toward death rattle?
Forgivable bugs are bugs created when programming to a poorly documented API, which leaves out important cases, or gives you false of conflicted information. Especially if "I don't trust this edge condition" is found in the program comments and some sane defensive coding strategy was employed to mitigate this.
The economic angle is a dangerous meme, because it lets incompetent management off the hook by allowing them to triage incompetence, and in particular, to ignore process failures, and the failure of team culture, both of which point back to management itself. We all know how well that worked out for NASA. This wasn't fixed until lives were lost.
Given a choice, I prefer to treat small bugs as canaries in the coal mine.
Another point. Obscurity is a fragile defense. Oops, the senior climate scientist just launched a month-long computation at E3 with the wrong initial parameter because of an obscure user bug three levels deep in an advanced sub-menu.
I'm more forgiving if the feature was only added because an important, yet lame customer demanded a feature that should never have been added in the first place, because they were too lazy to realize they are trying to use the software in the wrong way (but we've always done it that way). It's darn hard in practice to perfect the wrong basic approach.
In many other contexts, the obscurity of the function that blows up is directly proportional to the importance and gravitas of your first customer who trips over it. Does this person take the economic view (small bugs are expensive to fix, so I'm better off using a buggy piece of crap) or does he do what your management doesn't: evaluate the competence of the organization who produced the bug, from a psychology of extreme prejudice towards incompetence? I know I'm usually PEPTIC.
A triage culture lacking a robust root cause determination (e.g. five whys) is deep into the management zone "run, baby, run".
Does this economic wisdom boil down to anything more intelligent than the maxim that "if you can smell the fuel leak all the way back to mission command DON'T press the big button"? Yes, there's a natural priority in the order of things to first rooting out egregious incompetence. Then the serious work by serious people begins.
There's no better quality culture than one were every minor lapse of competence sails directly and efficiently back to its rightful owner. In the five whys culture, every defect generates a list of rightful owners, all the way up the chain. Now you're really getting somewhere. Along with that, it's nice if everyone buckles down to fix their corner of the problem rather that groping around for plausible yet ultimately failed excuse perspectives.
Some people might not be content with OO.org, true. But I would bet that is a small minority. OO.org is compatible enough with MS products that you have to do some relatively arcane things before you will notice much difference in the end product.
I have to say you don't get out much. My GF does agricultural policy and regularly works on collaborative documents where she gets a Word document that OO doesn't import properly, or tries to send something back after she edits it, and some version of Office doesn't handle it well. By regularly I mean every week. The problem is further exacerbated by her list of regulars not all having the same version of Office.
Not one of them is competent to properly structure a complex document using any of these programs. They just do whatever seems to work in the moment. These are smart, energetic, capable people who are more interested in furthering their cause than their word processing skills (for that matter, they would--every one of them--prefer to spend an hour forking horseshit onto their organic gardens than learn another Microsoft Office command).
If this thing amounted to middle ground on quality of interchange (as purported), I'd probably spring for it just to put an end to this charade.
OTOH, OO has improved over the years, so if we can tough it out for another year or two, the problem might just go away all on its own.
Slashdot Mirror
Anyway, other major BSD-like projects don't suffer from private forks, and there's no reason to think LLVM will either.
That's kind of funny. I have a dark suspicion that Apple created LLVM *precisely* so that they could maintain a private fork concerning some premium technologies (OpenCL, GCD, H.264 all come to mind). Generally, however, maintaining a private fork is a PITA so it happens far less than GPL bigots promulgate. (Non bigots don't promulgate. That's how you tell.)
Why do you think Intel bought Kuck and Associates? To make it optimize their own code more better, or optimize someone else's code less better? KCC before it was Intelified used to really kick ass, *especially* on AMD.
Unlike GCC, outsiders can successfully provide patches and bug reports to LLVM. The process isn't so incredibly oppressive that people just don't bother.
Well, the last time I reported a edge case error in template support for GCC, it was fixed within *two days* of receiving my report. I guess the response time depends on the nature of what you report. But you have a point. A comparable edge case bug that I reported on the R users mailing list was fixed in 15 minutes by one of the core developers, so GCC does kind of suck. There was a bug I used to follow on Ubuntu that was still in triage three years after filing and had hundreds of people following, as resolution slipped from one release to the next. It's hard to speak in generalities about these things.
Finally, to all those people who think that only the generated code speed matters, grab a brain! The rate limiting factor on software quality and complexity is the tooling ecology. For example, suppose some day Eclipse CDT adopts the LLVM C++ parser and error message components. If this advantage were to enable me to write more powerful algorithms, *even if* the generated code under LLVM was 10% slower than some other compiler, my superior solution would still kick ass.
If the tooling advantage of LLVM is *already* making it more productive to add complex optimization passes to LLVM, somehow I doubt that 10% penalty continues to exist much longer.
The most important feature in a compiler is making the software developer smarter and more efficient.
How did that retort get moderated insightful? It's far more clueless than the post he's responding to, which as least has its heart in the right place. Every second podcast at Econtalk has a long seventh inning stretch on a Hayekian view of capitalism cut _exactly_ from this mold.
If you're taking the grand view of what capitalism requires, small government is not on the list. Twenty years ago it used to be said that Russians understood capitalism better than Americans, because they could actually define it, and list the institutions it entails (in a negative light).
These days no one actively debates the grand view of capitalism. The active debate is about capitalism as a mainspring of wealth creation and the role of government to A) abet or B) hinder the golden goose. In the blue trunks: free market fundamentalism. In the red trunks: liberal society and justice for all.
Its a dearly held tenant of the invisible-hand contingent that markets are able to solve allocation problems though the pricing system that a centralized system could never properly manage, because the required information can't be collected at a central point, unless one waves a magic wand to approximate the utility function of people not present to speak for themselves. That kind of sucks.
It was Stiglitz who showed that the magical ability of markets to solve allocation problems through the price mechanism breaks down under conditions of asymmetrical information. *If* you have price transparency (and a few other things) markets can do an excellent job where government can't.
What you end up with is a system where the vigorous new enterprise favours price transparency (which permits greater economic mobility) while the incumbent corporations do everything in their power to debase price transparency (telecoms industry, media industry, to name just a few).
I don't trust the views of anyone who doesn't think that information transparency leads to a more effective and vigorous market economy. But then I believe that wealth should be earned rather than squatted upon. I know, it's a radical idea.
I was reading some commentary on the media business, including How to Save the News which is interesting, but didn't impress me. One of the articles mentioned Bertrand competition, which suggests that in the absence of product differentiation, the product will end up selling at marginal production cost. (I'm not an economist, so sue me if I didn't get that phrase quite right.)
The Atlantic article goes on an on without mentioning the core point: why do people volunteer themselves to have their purchasing preferences manipulated by visual images in the first place? If ad revenue represents 80% of a newspaper's income, how does the effect the nature of the story reported? Is it to inform the reader, or to create a warm context for associated display ads? The theory of advertising impressions is that you get the viewer into a receptive emotional state, and then burn your image into the viewers amygdala while under the influence of the warm glow. Hence all the Superbowl ads, which are beamed at men awash in vicarious sexual potency. Not such a good model for funding an insightful report on genocide in Somalia.
I'm all for a world with far greater price transparency. It would weed out many of the people who wish to live fat lifestyles without ever creating much of value. Opportunities for value creation have never been better. Personally, I wouldn't mind seeing more of the carpet baggers bagging carpets until they change their ways.
I think a marketplace which maximizes informed choice on *both* sides of every transactions could work small economic miracles. Big business believes in such a market until they don't. Big business believes in small government until they require a big bailout. This is just wealthy peopl
In fact, knowing the size of the flow is kind of crucial to stopping it.
The biggest point you make should be the most obvious: We aren't doing what we have already have proven to work, boom and capture.
I thought the premise is that first you have to know the size of the problem, before you can determine if the previously successful solutions are applicable or not.
If "what worked before" is irreproachably scale invariant, then BP is right in their conjecture that there's no scope for data to drive a more effective response. Wasn't part of the problem with Katrina more water than pumps?
What's completely stupid about the BP position on this is that there's no contention between assessing the scale of the problem and mounting the most effective response. People who aren't working on the clean up can work on assessing the data.
What's the downside to enlarging participation? The only aspect of the response that would be stretched thin by this approach is spin management by the BP PR flacks, who would view having a disaster *and* open data to contend with as too much like Germany fighting a war on two fronts.
For what purpose? The entire PR effort could be replaced with a simple pro-capitalist capitulation: "We screwed up, big time, and unlike the banks, we're solvent enough to pay restitution." As a sop to their shareholders they could also add "and we're going to litigate the hell out of our partners in crime after the tar settles". Capitalism is not such a bad system when the gears are allowed to mesh.
I've done a lot of embedded development, some of it in the security sector, which is not tolerant of sloppiness.
One of the products had a web interface implemented in PHP, which I partly coded. I don't use PHP regularly, so I was entirely at the mercy of the PHP online documentation. This was a few years back, but my god was it a struggle to glean from the documentation the bounds of validity of the different PHP function calls. In many cases I had to resort to untangling a long and often contradictory discussion thread trailing at the bottom of the API documentation. I was brutally easy to get my C code right because I was coding against APIs which I rigorously understood. I never felt half as confident in my PHP code.
In several cases, the PHP documentation should have just said "this is exactly what you have to do to properly escape input into this function" and it didn't, not even close.
Because the documentation sucks, it's impossible to be rigorous coding any acceptable speed for your typical web application. No one wants to pay people to code web applications on slow and methodical embedded application time frames.
Ultimately what drives bug are high expectations and sloppy API documentation. Sloppy documentation of APIs is greatly enhanced if the API itself was stupidly designed. Good design leads to simple documentation.
One more example. In my embedded work, when I have to ensure that output from one module is valid as input to the next module, I only use positive filters: I match on known good patterns. Never filter by subtraction of suspected evil, you'll end up missing something. If your filter is too conservative, you'll find out about it, from a bug report, rather than an exploit. Do PHBs running web development shops accept that trade off? Or is the attitude to err on the side of uptime, security be damned?
Mostly it's culture, not language.
Probabilistic Graphical Models: Principles and Techniques (Adaptive Computation and Machine Learning)
There, I feel better now.
Here's another postscript. Edge Foundation way back in 1999 ran a question with some historical depth.
EDGE: What Is The Most Important Invention?
From my notes: Joseph Traub, Mihaly Csikszentmihalyi, David Shaw, David Myers, Milford H. Wolpoff, John McCarthy, Philip Brockman, Howard Rheingold, Samuel Barondes and John Rennie all put forward the scientific method or some variation on scientific progress, alongside all the usual votes for the pill, the printing press, the a-bomb, digital electronics, and hay.
Around the time of that survey, a lot of people convinced themselves that this new model for managing a company (or a software product line) made sense, because that was how the world worked now. No, actually. Even in the dotcom stampede, bad management was bad management. This will hold true of science as well.
The scientific method is way too important in the history of modern civilization to have the IPCC make lite of this tradition in order to win a political grudge match.
Too bad Wolfram used up the title "A New Kind of Science". We could have saved it for IPCC committee reports with 2,500 eminent signatures.
I say this with full conviction that the central human activity of the 21'st century will be paying the piper for high living. I still harbour a dim hope that we'll pay off these days less rashly than we entered into them. Yes, I can see it now. This will all come to pass through a non-contentious political process involving scientific walled gardens, incestuous peer review, and sanctified data hoarding.
As a footnote, I read this the other night and was quite impressed with it. This will mostly appeal to slashdotters with low digit IDs and mild Aspergers, if such a creature exists.
The Art and Science of Cause and Effect
Note that he takes a long view of science as I do. The key slide that just popped into mind is slide 49 with the text:
However, carve a chunk from it, say the object part, and we can talk about the motion of the hand CAUSING this light ray to change angle.
The precautionary principle is fundamentally interventionist. However, the focus of precaution is necessarily a human construct, which depends upon how the image is sliced. This claim is heavily supported in the presentation as a whole.
This insight courtesy of Judea Pearl, who is becoming known as one of the giants of AI. He's a major influence on the recent work of Daphne Koller. Under no circumstances check out the accomplishments of Daphne Koller if you're feeling low about your productivity in the recent week or decade. She's just polished off a nice 1,200 page tome http://www.amazon.com/Probabilistic-Graphical-Models-Principles-Computation/dp/0262013193/">Probabilistic Graphical Models: Principles and Techniques (Adaptive Computation and Machine Learning). I'd rush out to buy this, but I'm not sure my ego can handle the blow.
We agree with the CRU view that the authority for releasing unpublished raw data to third parties should stay with those who collected it.
In the ordinary scheme of things, where science proceeds at a slow and deliberate pace, and the stakes rarely exceed ego and pride and lifetime accomplishment, it would be fine to allow scientists to manage their walled garden as they are accustomed to doing.
In the extraordinary scheme of things where fluctuations so minor they are hard to measure are beating the drum on global policy, and recourse to sober reflection has been staked through the heart with language of imminent "tipping points", this is not good enough.
According to the respected tradition of science, the scientists will soon come to a sober and reliable consensus on AGW 1980-2010, where soon is somewhere between 2050 and 2100.
Cripes, Einstein wasn't awarded a Nobel prize until 1921, sixteen years after his annus mirabilis, roughly the equivalent of winning all four majors in the same year by ten strokes each. Tiger never had a year that good. Gretzky never had a year that good. At the pace that science traditionally moves, Gretzky would be a recent induction to the Hall of Fame, after a sober cooling off period, rather than handing his final game jersey and jockstrap to a burly Scotsman handcuffed to a velvet trunk. The tradition in science is to allow the sweat to dry before inviting posterity to drop in for a look see.
Now that the stakes are so high (apparently), Big Science has slapped 2,500 signatures on a fat report in total contravention of every word of wisdom in Brook's "Mythical Man Month": you can't accelerate solid results by piling on resources or amassing "looks good to me" chits by the basket load. It doesn't work in software, and it doesn't work in science.
If they think this is a viable route to sober consensus, they deserve a level of outside scrutiny that would make NASA blush.
Where did this idea come from that science can function as a miracle short-order chef, just because it has to? I hate to parrot Thomas Kuhn, but he did point out that error in science is weeded out of the system on a generational time scale. Personally, I don't see a huge difference between Harry Markopolous and Stephen McIntyre, Both have a long history of having a sharp eye for dubious claims.
If science thinks they can produce results on a Wall Street time frame, welcome to Wall Street scrutiny, er, I mean the scrutiny we now wish had been in place before it was too late.
Right now we have a natural order and an economic order that are working at cross principals. We can't apply the precautionary principle on both sides of the fence simultaneously. So the debate devolves to which of the two should we F with, and in what proportion? Clamping down on the current economic order out of fear that an effect is major and immediate rather than moderate and mid-horizon has real effects on global standards of living.
It's possible that the environmental scientists are right, which would make their urgency good politics, but still not good science, not until the day that fully disclosed data subjected to the best possible analysis fully supports their conclusions.
Maybe this issue is too important to wait for good science. Nevertheless, under no conditions am I buying into this revisionist agenda as to what good science looks like. If we have to turn to lame science, and that's the best we have, so be it. Meanwhile, I'd be quite pleased to dispense with all this faux identity maintenance by scientists who have clearly overstepped the tradition of sober reflection, no matter if for the best of reasons.
People always say this. So, where's the list of all Congressmen with net worth of more than a billion dollars?
Ah, reminds me of the old joke about the priest who skives his responsibilities to play a secret game of golf, then God punishes him with a hole-in-one, on the grounds "who can he tell?"
A congressman selling influence isn't going to be paying tax on the transaction. He might be rich as stink offshore, but who can he tell?
This is why society develops secret societies with secret handshakes, so that old men can gather together and speak in veiled and cryptic phrases about how to best invest funds that don't exist.
I'm not saying there are any billionaire congressmen out there, but I am saying that your standard of evidence has a Marcos-sized hole.
I didn't say there wasn't still a gap, I said the gap has been significantly reduced.
I haven't looked at the numbers closely, but I suspect the gap hasn't been reduced at all. ATI's new Evergreen is kicking out a trillion SP FLOPS and now has full IEEE double precision as well. After ignoring graphics for years, OpenCL has caught my fancy. I once hoped that IBM would kick out an upgraded version of Cell with fast IEEE double precision, but their unholy alliance with Sony proved to be quite the fiasco.
What has greatly changed is the relevance of the gap. When you're a student living in a 300 sq ft apartment, 600 sq ft is a screaming upgrade. Later in life, the upgrade from 2500 sq ft to 5000 sq ft has a narrower appeal. Good if your favorite game in life is playing indoor hide and seek with your grand-toddlers or you're a world class model train builder.
The people who are happy enough with the tower PC they already have are not going away. Like Jobs predicts, it's not a sector capable of supporting the high living Apple desires. Everyone's creative energy is going to be poured into finding the bisection point between mobile and cloud. Everyone with aspirations for high living is piling into this sector. It won't be cheap. HP is going to congress to ask for a $50/month levy for virtual ink on every user of cloud computing services.
On the other side of the coin, the boring old sector will fade from mind a lot faster than it will fade from reality. I predict large PCs will fade away about as quickly as large SUVs in America. If you look at the dealerships, you might come to one conclusion. But then look on the road around you. Sit in your Smart Car in heavy traffic some day and count the number of bumpers at eye level before declaring dinosaurs extinct.
After writing such a nice screed about Flash, Jobs won't have any difficulty understanding why Google might wish to undermine H.264 with V8.
The interesting thing about the computer industry these days is that control points have less to do with de facto product monopolies, and a lot more to do with physical embodiment. Running a huge data center, the pendulum swings toward the ability to maintain trade secrets, and I suspect there's some of that in the closed device mobile sector.
In this new world, the incumbent monopolists have an alternative to circling the waggons and supporting each other's misbegotten prominence. Steve's essay is an excellent rendering of the king is dead, long live the king.
Embarrassing to be Canadian. Well, at least the Pope isn't Canadian, although to hear what he had to say about the internet, he might be on board this paper thing.
What the Domtar dipshit didn't take into account is embodied energy. People tend to have wacky ideas about embodied energy, unless you bother to work the numbers.
Catherine Mohr builds green
Next time I see a solar powered logging truck, I'll think "damn, John Williams was so right".
What's that funny metal pipe, daddy? That's called a muffler, Sally. Back in the day, most logging trucks had one.
Here are some nice photos about how the logging industry used to look before petroleum was banned.
TJ's Woodshop - Logging Photos
In the modern world, where nearly every problem has a collaborative dimension (even rootkit developers exchange tips of the trade) the phrase "excess collaboration" reminds me of the line from Jack Nicholson's character in One Flew Over the Cuckoos Nest, "Who ever heard of a man getting too much poozle?"
[http://www.mtc.mb.ca/WorkArea/linkit.aspx?LinkIdentifier=id&ItemID=4786 McMurphy spars with the untroubled Dr Spivey]
It makes you wonder about the rampant NIH that has long existed in the software profession.
Maybe they should instead be teaching students to *never* work from scratch if they can source a more suitable starting point. Maybe they should deep six sitting in front of a droning professor. You can do that online these days from the close proximity of your residence bar fridge.
Maybe nearly 100% of computer class time should be having the students conduct group walk-throughs of proposed designs, algorithms, and implementation fragments. Maybe the CS professor should be assessing how the students are interacting rather than mindlessly reciting prepared solutions (if you can find enough CS professors with a facility for eye contact).
Maybe an obstacle-course education in emergency wheel-reinvention should be regarded as a commando level training option for elite troops only.
Maybe the curriculum should include a course in outsource management, where your project group is graded on the work quality of its very own team of docile and compliant Indian math geeks.
Maybe the education system as it now stands is part of the problem, rather than part of the solution.
I also agree this is a futile quest. While it might be possible for an enlightened company to profit from open source, Oracle is not that company, and is highly unlikely to become that company no matter what argument is advanced here.
From TFA:
"We believe that [20 GB/cm^2] is at least a factor of two better storage than flash memory will [] have [three years from now]"
I wouldn't be too thrilled with that proposition. This is an ambitious new technology introduction. It won't take much for that time frame to slip by a year or two, in which case your edge is shot and your price structure is unlikely to be competitive.
Suppose you get 5% initial market share where the difference in feature set is "just right" for some set of early adopters. Now with 5% of the revenue base, your business requirement is to scale faster than a mature competitor sitting on 95% of the revenue base, funded by infinitely deep pockets (Intel, to name one).
Even if this new technology is fundamentally sound and has scaling headroom to burn, it could easily be a decade before it pulls into the passing lane once and for all. Maybe there's a 10% chance all the arrows line up and it arrives on schedule three years from now cheaper, denser, and faster than existing flash. That prospect alone is enough to fund a serious market push, if the up front R&D is not outrageous.
OTOH, if it arrives late or expensive or slow in performance, this could play out more like the Thrilla in Manilla, with one or the other of the heavyweights collapsing on their stool at the end of the 14'th round.
Time to check out the spot price on Froogle Futures for April 2013. All I'm getting is "HP 404". What does that mean?
So according to this logic every C programmer must be lazy and stupid because you can not find a single, non-trivial program ...
Well, the bad news is that you failed statistics, but the good news is the AI department called and wants you back. Since AI is actually making some progress these days, they're at risk of outstripping trivial. After 50 years it would be a shame to sneak over the bar by accident now that HAL is merely another half century down the road. With your help, history will record that HAL is the first ever non-trivial AI application.
In the late 1980s I was lead architect on a non-trivial software package written entirely in C. We took some prudent measures to eliminate pointer abuse, both in terms of architecture and coding style. The number of times our team make a pointer mistake in a release build was running roughly neck and neck with the number of times we reported a code generation error concerning the Microsoft or Watcom C compiler. I pretty much forbid the use of CodeView until it was a coin flip whether it was the code or the compiler.
Of course, we did have assert statements going off on a daily basis during heavy coding. We also made a lot of typing mistakes and sometimes we ran our build scripts with the wrong arguments. Doh! Yet after all that, most software projects still use keyboards and build scripts. Will we ever learn?
What's the rationale for Goo against the use of assert() statements? Having eliminated pointers, it's no longer possible to violate a post-condition? Or does Goo include a magic eight ball to decide which developer should first tackle a code failure, having broken their side of the interface contract?
That was the whole point of the assert() discipline, to create a social mechanism by which the person who screws up is most often the person who fetches the mop and bucket.
This is a response to my own post. Sometimes after uncorking a minor screed, I note to myself "that was more obnoxious than normal" and then my subconscious goes "ding!" and I get what's grinding me.
The secret of x86 longevity is to have been so coyote-ugly that it turns into pablum the brain of any x86-hater who tries to make a chip to rid the planet of the scourge once and for all.
For three decades right-thinking chip designers have *wanted* x86 to prove as bad in reality as ugliness ought to dictate.
Instead of having a balanced perspective on beauty, the x86-haters succumb to the rule of thumb that the less like x86, the better. And almost always, that lead to a mistake, because x86 was never in fact rotten to the gore. You need a big design team, and it bleeds heat, but all other respects, it proved salvageable over and over and over again.
On the empirical evidence, high standards of beauty in CPU design are overrated. Instead, we should have been employing high standards of pragmatic compromise.
If any design team had aimed merely for "a hell of lot less ugly", instead of becoming mired in some beauty-driven conceptual over-reaction, maybe x86 might have died already.
Maybe instruction sets aren't meant to be beautiful. Of course, viewed that way, this is an age-old debate.
The Rise of ``Worse is Better''
Empirically, x86 won.
The lingering question is this: is less worse less better, or was there a way out, and all the beauty mongers failed to find it?
If the 1.8GHz Xeon was based on the Netburst architecture, first you have to multiply by 2/3rds to correct for diet Pepsi clock cycles, then if your code base is scientific, you have to divide by two for the known x86 floating point catastrophe, and finally, if your scientific application is especially large register set friendly, there's another factor of 0.75. So on that particular code base, a 1.8GHz Netbust is about equal to a 400MHz Alpha (I only ever worked with the in-order edition). Netburst usually had some stinking fast benchmarks to show for itself if it happened to have exactly the right SSE instructions for the task at hand. And it gained a lot of relative performance on pure integer code. BTW, were you running Xeon in 64-bit mode? That could be another factor of 0.75.
A lot of people, myself included, think itanium should never have existed, and that the development effort should have been put into alpha instead - an architecture that already had a good software and user base
Yeah, you and a lot of clear headed people with insight into the visible half of the problem space. Not good enough.
Alpha was a nice little miracle, but it fundamentally cheated in its fabrication tactics. This is a long time ago, but as I recall, in order to get single-cycle 64-bit carry propagation, they added extra metal layers for look-ahead carry generation. For a chip intended Intel scale mass production, this kind of thing probably makes an Intel engineer's eyebrows pop off. That chip was tuned like a Ferrari. I'm sure the Alpha was designed to scale, but almost certainly not at a cost of production that generates the fat margins Intel is accustomed to.
Around the time Itanium was first announced, I spent a week poking into transport triggered architectures. There was some kind of TTA tool download, from HP I think, and I poked my nose into a lot of the rationale and sundry documentation.
TTA actually contains a lot of valid insight into the design problem. The problem is that Intel muffed the translation, through a combination of monopolistic sugar cravings, management hubris, and cart before the horse engineering objectives. I'm sure many of the Intel engineers would like to take a Mulligan on some of the original design decisions. There might have been a decent in there somewhere trying to get out. Itanium was never that chip.
I pretty much threw in the towel on Itanium becoming the next standard platform for scientific computing when I discovered that the instruction bundles contained three *independent* instructions. They went the wrong way right there. They could have defined the bundles to contain up to seven highly dependent instructions, something like complex number multiplication: four operands, seven operations, two results. It should have been possible to encode that in a single bundle. Either the whole bundle retires, or not at all.
Dependencies *internal* to a bundle are easy to make explicit with a clever instruction encoding format. You wouldn't need a lot of circuitry to track these local dependencies. What you gain is that you only have to perform four reads from the register file and two writes to the register file to complete up to, in this example, seven ALU operations. Ports on the register file is one of the primary bottlenecks in TTA theory.
What you lose is that these bundles have a very long flight time before final retirement. Using P6 latencies, it's about ten clock cycles for the complex multiplication mul/add tree in this example (not assuming a fused mul-add). This means you have to keep a lot of the complexity of the P6 on the ROB side (retirement order buffer). But that also functions as a shock absorber for non-determinism, and takes a huge burden off the shoulders of the compiler writers. This was apparent to me long before the dust settled on the failure of the Itanium compiler initiative.
In my intuitively preferred approach, instructions within bundles would be tightly bound and s
People, take a quick bunny hop through List of cognitive biases and ask yourself how many of these constitute cognitive feeder arteries for Godwin's law.
Every so often a topic comes up where everyone simultaneously decides to let their amygdalas off the leash in the same dog park, who all quickly pair up for a circular open-jaw square dance. After the dust settles, what do you have? A giant patch of lawn to circumvent until the next heavy rainfall.
I suppose the game is a bit more earnest for the born complainer, who does have to somehow realign the pole of supreme evil with every successive regime change. Those of us with the attention span to get through the first page of Anna Karenina understand that evil has a multitude of poles, any one of which can erupt into the supreme pimple in the rapidly shifting context of real life.
if anybody shakes MS loose from that, my bet would be Google rather than Apple
Apparently, slashdotters don't read Swift, either.
There are at least half a dozen major players tugging on different fingers of the Beast of Redmond. Sony is doing their ineffective best tugging on the pinky finger with their once-powerful PS3 franchise. The unholy alliance Snoracle has a firm but limp-wristed grasp on the middle finger on office suite revenue streams. Linux/Apache/Firefox inflicted a hairline fracture on a wristbone. Google extracted a fingernail from the ring finger when it became the ultimate talent drain. That had to hurt. And now they're proceeding to bend back the index finger by sucking up the air supply in online search. Learn from the best. A horsefly named Gnome Evolution landed on the thumb and carted off the largest divot of flesh it could manage, which considering all the other wounds, is of no real consequence whatsoever, unless horseflies are a vector for Ebola, and so far it appears that they aren't. All things considered, I think that Microsoft can hang there by their relatively undamaged, enterprising thumbs for another thirty years or so.
The biggest risk with Apple is that they manage to leverage their carefully cultivated charisma (if it survives their having become a big enough company to matter in these discussions) to make DRM palatable to the masses.
Sony is far more evil in the DRM department (witness the recent "other OS" rescindment fiasco) but they suffer from a bad case of cartoon evil: whatever their grasping ambition, it's soon equally matched by their incompetence. They managed--on the back of a half billion dollar war chest--to leverage their dominant Play Station franchise into a slow and lukewarm victory in a dying physical media platform.
This rivals anything accomplished by the Hudson Bay Company (oldest corporation in North America) which once laid claim to half the natural resources in Canada, but decided the crown jewel was retailing dress shirts. If Warren Buffett had gained control of HBC in the late 1700s, America might now be the 11'th Canadian province, or an economic protectorate, like Puerto Rico. (If BG gained control of the HBC in the late 1700s, Russia would now be the world's great democracy and white knight of freedom.)
Wish the Sony/HBC disease were true of Apple, but it isn't.
I could continue grave digging in this vein for another day or two, but hey, it's Easter, and whatever your opinion on the back story, there was an important lesson in there about the rush to judgement.
"Physics is Fun to Imagine" interview with BBC
I can't play 'tube on this box. It's likely segment 11 or 12 where Feynman gets into a multitasking competition with statistician John Tukey, and discovers that they can multitask different things, if no-one around here is patient enough to sit through the whole thing.
Just what did he think he was doing wasting all this time figuring this stuff out 40 years ago before anyone cared?
Broken URL syntax in the previous post brought to you by my favorite Fireflog plug-in Make Link. Which is why Boomtango has the slow horse in this race.
This is OT as hell.
Are you related to Bill Clinton, and learned to parse words to death?
From [http://www.nytimes.com/2010/03/26/opinion/26krugman.html Paul Krugman - Going to Extreme]
For today's G.O.P. is, fully and finally, the party of Ronald Reagan -- not Reagan the pragmatic politician, who could and did strike deals with Democrats, but Reagan the antigovernment fanatic, who warned that Medicare would destroy American freedom. It's a party that sees modest efforts to improve Americans' economic and health security not merely as unwise, but as monstrous. It's a party in which paranoid fantasies about the other side -- Obama is a socialist, Democrats have totalitarian ambitions -- are mainstream. And, as a result, it's a party that fundamentally doesn't accept anyone else's right to govern.
Bill is far from the first charismatic leader whose tallywhacker has gone walk about. What was underlying the whole charade was a dispute over Clinton's right to govern. He was elected by a majority decision, for better or worse, and the republicans failed to respect this.
What exactly is the correct way to parse a question that never should have been asked in the first place? The correct answer was "what does that have to do with the price of tea in China?"
"Mr Limbaugh, are you addicted to Hillbilly Heroin?" What's he supposed to say? Sure, he's a two-faced lout, just like Clinton, but it's still an ad hominem attack, and undeserving of a straight answer.
forged prescription:addiction::lucky blow job:adultery
Does that work? I guess it depends on your attitude toward natural pain relief over artificial pain relief.
On a technical note, this is the very first time Boomtango helped me to locate something I've recently read on the internet faster than I could have done it myself. And for this, all day my Firefox is an extra beat sluggish. For now, I'm sticking with it on potential.
Since you don't know what this amount is you have to use a proxy. Oracle uses features, the number of cores and ram as their proxy.
This is a valid recitation of economic orthodoxy which unfortunately leaves off right at the moment where thinking begins. Ships are lost on this basis.
The problem is that the choice of proxy has downstream consequence which can destroy a lot of value, or in the worst case, almost the entire value of the sales proposition.
In the case of licensing by core, it introduces a huge non-linear term in right-sizing infrastructure. I can't stand this stuff myself, but some of my closest friends have made an excellent living showing up to solve the Oracle license fee non-linear optimization problem, which contains large elements of uncertainty and non-determinism because the outcome depends on unknown future events.
Anyone with the least insight into systems theory knows that non-linearities are like a sexual disease. They have a noted tendency to give on giving. Properly understood, the effort involved in damping out these non-linearities can easily exceed the value proposition of adopting Oracle solutions in the first place.
Fortunately for Oracle, there's a huge real world shortfall in the quantity "properly understood". Microsoft, among others, makes a mint from truncated TCO studies. The assumption underlying every TCO study I've ever seen is that the higher order non-linearities can be safely neglected. If that were true, why is anyone relying on a vendor-funded TCO? In the case where the higher order terms can be safely neglected, it's usually easy enough for the customer to work their own TCO on the back of napkin, and get an immediate answer everyone immediately believes.
The cases where this breaks down are the sales propositions absolutely freighted with non-linearities, to the point where no one trusts their own numbers, surprise, surprise.
The first rats off a sinking ship are the best swimmers. The first wave of people to abandon Oracle are those outfits with a larger than average insight into "properly understood".
From [http://www.riskglossary.com/link/barings_debacle.htm Barings Debacle]
In November 1993, BSL was merged into BB&Co. in anticipation of a subsequent initiative to form a Barings Investment Bank (BIB). The merger was not easy because the two firms had markedly different cultures. It was a distraction right in the middle of Leeson's tenure at BSL. ... Employees complained that lines of reporting were not always clear. ... Another issue was that Leeson was an accomplished liar.
...
Barings was just starting to form a risk management function. Risk controllers were appointed in London, Tokyo and Hong Kong during 1994, but not in Singapore.
...
As part of the 1993 reorganization, Barings had adopted a "matrix" approach to management of its offices.
Every aspect of this is situation normal at most medium or large companies. You best executive attention is devoted to various political fires. Many organizations are just too busy with other pressing demands to step back and engage in the kind of clear thinking it takes to put out the Oracle fire. So what if the Oracle pricing model induces non-linearities? We're planning to auction block that division anyway.
However, in the companies where choosing the right database and the right database architecture is the dominant fire, non-linearities associated with proxy pricing models can escalate into a serious business concern. Some of those people will go talk to Oracle and try to cut a special deal. Many of them won't. An attrition sets in.
Look what happened to Microsoft when Google became the hot job opportunity. That sucking sound is your technical clout packing family photos into their briefcases.
A major coming of age event in commoditization of a technology is crossing the threshold where the price proxy shifts from being a
That's a dialect of Canadian known as Hansardese. Hansardese was actually a staple of the early machine translation efforts, since you could be quite certain that Hansardese of fuddle duddle in French was an immaculate parallel translation. Not sure it was such a good idea to prime machine translation on empty calories, though.
These multinationals have got a nice thing going. "But mommy says we have to ratify!" Works every time.
In some of these open source projects, version 1.0 is like the first time the odometer in your car rolls over. Or like a couple who finally decide to get married after 15 years of living in sin. I wonder if this big decision involved a trip to Vegas.
Version 1.0 isn't that different from getting marriage. Some enter into it on the basis of hope and enthusiasm with neither experience nor skill, while others circle each other like planets in a decaying orbit.
A long run in the zero point nineties is like the people who are technically married, but have not yet escaped their parents' basements, lacking either the spirit or means of independence.
Then comes the bold and tremulous day when they finally cut the apron strings, while everyone stands around in state of genuine micro-perplexity going "I had no idea".
I read the other day that the dung beetle has been discovered to be one of the world's strongest organisms by body mass. I've never seen a single dung beetle toting a Champagne magnum. It's clear they can't get the cork out. Or maybe no one has figured out how to make the bubbles small enough to fit in the bottle.
One of the best things I've read on this subject was an interview with top level muck muck at NASA (I think it was _the_ guy) when NASA was bent on fixing their safety culture, following a little bug that cascaded into millions of small pieces. In a safety audit they found a small number of broken wires in some electrical harnesses that weren't supposed to have this defect in this quantity. The engineers did what I might term a sensitivity analysis and determined that with all the other layers of redundancy, there was essentially zero chance this would lead to a major incident.
The big muck muck decided to postpone the launch regardless, because, in his words, "of what the wires were trying to tell me". The wires were trying to tell him that their statistical models concerning production quality were incorrect one way or another. If you quality process is broken, this is something you want to know before you let the big dawg ring in the new year.
The second question you need to ask is this: is the root cause behind the bug forgivable? Or was the programmer responsible (or just as likely, the culture created by the management team) unable to free itself from a wet paper bag? Does the bug indicate a state of crisis in personal topology assessment? Is this the first visible black buboe in a software release candidate rapidly progressing toward death rattle?
Forgivable bugs are bugs created when programming to a poorly documented API, which leaves out important cases, or gives you false of conflicted information. Especially if "I don't trust this edge condition" is found in the program comments and some sane defensive coding strategy was employed to mitigate this.
The economic angle is a dangerous meme, because it lets incompetent management off the hook by allowing them to triage incompetence, and in particular, to ignore process failures, and the failure of team culture, both of which point back to management itself. We all know how well that worked out for NASA. This wasn't fixed until lives were lost.
Given a choice, I prefer to treat small bugs as canaries in the coal mine.
Another point. Obscurity is a fragile defense. Oops, the senior climate scientist just launched a month-long computation at E3 with the wrong initial parameter because of an obscure user bug three levels deep in an advanced sub-menu.
I'm more forgiving if the feature was only added because an important, yet lame customer demanded a feature that should never have been added in the first place, because they were too lazy to realize they are trying to use the software in the wrong way (but we've always done it that way). It's darn hard in practice to perfect the wrong basic approach.
In many other contexts, the obscurity of the function that blows up is directly proportional to the importance and gravitas of your first customer who trips over it. Does this person take the economic view (small bugs are expensive to fix, so I'm better off using a buggy piece of crap) or does he do what your management doesn't: evaluate the competence of the organization who produced the bug, from a psychology of extreme prejudice towards incompetence? I know I'm usually PEPTIC.
A triage culture lacking a robust root cause determination (e.g. five whys) is deep into the management zone "run, baby, run".
Does this economic wisdom boil down to anything more intelligent than the maxim that "if you can smell the fuel leak all the way back to mission command DON'T press the big button"? Yes, there's a natural priority in the order of things to first rooting out egregious incompetence. Then the serious work by serious people begins.
There's no better quality culture than one were every minor lapse of competence sails directly and efficiently back to its rightful owner. In the five whys culture, every defect generates a list of rightful owners, all the way up the chain. Now you're really getting somewhere. Along with that, it's nice if everyone buckles down to fix their corner of the problem rather that groping around for plausible yet ultimately failed excuse perspectives.
Some people might not be content with OO.org, true. But I would bet that is a small minority. OO.org is compatible enough with MS products that you have to do some relatively arcane things before you will notice much difference in the end product.
I have to say you don't get out much. My GF does agricultural policy and regularly works on collaborative documents where she gets a Word document that OO doesn't import properly, or tries to send something back after she edits it, and some version of Office doesn't handle it well. By regularly I mean every week. The problem is further exacerbated by her list of regulars not all having the same version of Office.
Not one of them is competent to properly structure a complex document using any of these programs. They just do whatever seems to work in the moment. These are smart, energetic, capable people who are more interested in furthering their cause than their word processing skills (for that matter, they would--every one of them--prefer to spend an hour forking horseshit onto their organic gardens than learn another Microsoft Office command).
If this thing amounted to middle ground on quality of interchange (as purported), I'd probably spring for it just to put an end to this charade.
OTOH, OO has improved over the years, so if we can tough it out for another year or two, the problem might just go away all on its own.