To be precise, most people believe its impossible to write a non-trivial program without any bugs. Its a belief most programmers have no interest in refuting, because it eliminates accountability.

In reality, the truth is its impossible for most people to write non-trivial programs without bugs. Just like its impossible for most people to consistently land airplanes without crashing. In air travel, 99.9% of those people are weeded out of pilot programs. In programming, they move on to the next project.

Part of the problem is that the programming profession hasn't had its professional renaissance like the medical profession had in the twentieth century. We don't train programmers to be skilled, efficient, and above all conform to an agreed to set of professional standards. There's no such thing as programmer malpractice. Basically the software development industry is exactly where the medical profession would be if everyone owned a medical text from ancient Greece and treated themselves and their friends based on guesswork and late night infomercials.

Had the FCC's Net Neutrality rules been in force today? Verizon would probably be in violation of the no unreasonable discrimination rule (FCC 10-201, Section III.2). Specifically, given the facts as currrently presented, Verizon would be throttling a content provider based solely upon a customer's subscriber class without a valid network management requirement and without sufficient transparency. The FCC rules would have allowed Verizon to sell residential broadband that had less bandwidth than business class customers, but not explicitly notifying the customer that services such as Netflix or AWS hosted content would be bandwidth restricted to far less than the actual bandwidth being provided by the service would have likely been considered unreasonable.

Most Net Neutrality frameworks do not specifically prohibit service providers from charging customers more or less for different overall classes of service, but most including the 2010 FCC framework required them to disclose those differences in clear language to customers. I.e. "if you buy FIOS Verizon, you'll get huge bandwidth to your house but we're not going to allow you to use that bandwidth to stream Netflix to your house, we're going to throttle that traffic down to a lower level of quality than the bandwidth suggests."

So Verizon could theoretically do what they are doing, but only if they told their customers they were going to do it and allowed them to make a choice as to whether to pay for such a service. At the moment, Verizon does not appear to be doing so, no matter what their CS engineer told a single subscriber.

Actually, most of the millions upon millions of people who use Windows did not explicitly buy Windows. Sometimes the mob isn't even given a choice.

Its also worth noting that sometimes you can only push the mob so far. Its interesting that many hardware vendors that had previously gone along with Microsoft and changed their entire home PC lineup to be Windows 8 exclusively are now offering Windows 7 preloaded PCs again. There's only one reason to do that, and that's because those vendors believe forcing people to choose between Windows 8 or nothing is causing many people to choose nothing. I'm taking no small pleasure in lumping Windows 8 supporters with New Coke supporters and predicting the same ultimate destiny for both.

That only works if the contractor is stupid or young and inexperienced. The veteran strategy is to document all the bugs before fixing them. If possible, get them confirmed by the users as part of the documentation. If its part of your job to fix the bugs that documentation ought to be part of your job anyway, and you can't be blamed for bugs that existed before you got there. If its not part of your job, you have leverage: you can be a nice guy and fix them gratis, but with the understanding you didn't have to, or you can file the fixing of those bugs as change orders.

As long as you are not a dick about it, and genuinely appear to want to fix the project and make it work, the documentation trail you leave behind will protect you against being blamed for things not your fault and you'll probably be thanked for going the extra mile.

What does that gibberish even mean? If by that you mean the review does not make a judgment on how well each individual case study managed the design challenges of their data center designs, I don't see how that's the job of the reviewer. In fact, the whole point of the book seems to be that different skilled designers can approach the same problem in different ways, leading to different but still functional results. For a book reviewer to grade those decisions seems presumptuous at best and ridiculous at worst unless the reviewer possessed superior knowledge to the collective designers interviewed for the book.

What I would want to know before buying such a book is that it contained a diverse set of viewpoints from a diverse set of designers that ultimately generated successful designs. What the reviewer thought about how well "optimized" those designs were would matter not at all to me, unless I recognized the reviewer by name as an expert on data center design. In fact the notion that there exists a singular viewpoint or methodology by which you can judge "optimized" is apparently contradicted by the implicit thesis of the book. And as someone who delves into such design on a much smaller scale, I too have found that "conventional wisdom" is often extremely unwise, and the notion of "optimal design" is almost always very slippery.

I want to read how actual designers confronted actual design problems and resolved them with actual designs that then had to live with those designs implemented in real data centers. Everything else is random and generally worthless color commentary.

Sometimes, what a non-technical manager brings to the table is a lack of understanding. The biggest source of failure in software development, among other large IT projects, is consistent and almost institutionalized miscommunication between the producers of the project and its consumers. Most developers are not user-centric when it comes to thinking about the requirements of a software project, and most users are not sufficiently competent to request what they want with technical precision. So you have end users asking for things using jargon they do not really understand, and programmers writing things they think the users want without knowing for certain. These projects *only* succeed if there are people in the middle constantly vetting what's going on and making sure that both sides don't just say they agree on what should happen, but actually understand enough to provide informed consent. And that's hard when both sides don't want to look stupid and thus tend not to ask questions or demand clarification.

The guy who can admit he doesn't understand what the developers are saying but demands they keep explaining it until he does, the guy who is willing to tell the customers that he doesn't know what they are asking for and sits down with them to drop the jargon and have them explain it in their own language, is *enormously* valuable. He is actually the only chance the entire team has to be successful, unless they happen to consistently hire intrinsically lucky people.

Sometimes your own technical staff just happens to have people who can serve this purpose, and don't need non-technical go-betweeens. But such people are far rarer than competent developers themselves. Sometimes, what you need is someone around who reminds you that your customers are as technically ignorant as he is, and if you can't explain it to him your customers can't possibly have understood you either, and conversely makes your customers feel comfortable enough to explain things without believing they should try to exhibit more technical knowledge than they actually have.

Its entirely possible there are people who take it as an axiom that all sufficiently complex anything is automatically intelligent. However, I don't think that could remotely be considered a generally accepted axiom in the field of AI. That's not the same thing as the much more generally accepted belief that intelligence is an emergent property of complex systems. *Some* complex systems have that property. A lot of them do not, as we currently define intelligence. Furthermore, someone that has been dead for a few seconds has nearly the same complexity as they did when they were alive, but generally speaking we wouldn't assert they were still intelligent. Its clear that the only way to make that assertion true would be to spend a lot of time defining "sufficiently complex" in very convoluted and ad hoc ways.

One way I've heard people attempt this trick is to argue that intelligence itself is complex, and therefore any system that possesses the kind of complexity that defines intelligence is automatically itself intelligent. But peel away the semantics, and all that says is all intelligent systems are intelligent, which is a circular definition.

The distinguishing characteristic of Science, and what unambiguously distinguishes it from Religion, is that Science is about testing. We trust quantum mechanics because we use it, and it works. You're reading this on a computer that has parts that rely on quantum mechanics being correct, that only works because the predictions of quantum mechanics matched reality.

Also, *ideas* are rarely ridiculed by the scientific community at large. Faster than light travel, for example, has been the subject of serious academic study for as long as relativity itself has existed. What gets ridiculed are the attitudes often found in conjunction with such ideas, like "actually if you repeat something enough, eventually the majority will believe it... which is the basis of science and religion" for example. The notion that there's a big scientific conspiracy to quash novel ideas, when the truth is that while individual scientists are as corrupt and foolish as the general population, the big scientific conspiracy is to quash silly thinking, and erroneous ideas about how all great scientific ideas are "revolutionary" and "overturn" prior knowledge. Science mostly evolves: even the most "revolutionary" ideas of the last century like relativity and quantum mechanics are themselves evolutionary ideas and built upon prior work, and were themselves built upon by many other scientists.

I don't blindingly trust everything Science says, and no working scientists does either. The purpose of working scientists is specifically to advance the field of Science, which specifically *requires* questioning. But yes, I am far less critical about a physics textbook than I am a religious text. And the reason is because the principles in the physics textbook have been tested, repeatedly, uncounted numbers of times. The physics in that textbook consistently and accurately describes how my car accelerates and breaks, how the sun dies my clothes, how electrons in my computer's CPU flow, how GPS satellites work. Time dilation effects of general relativity and quantum electronic effects work precisely as physics theory state they should to make the GPS in my cellphone work. Every time anyone uses turn by turn navigation, they are confirming that relativistic time dilation is real: if GPS didn't account for it, GPS would be generate errors of miles per day.

Those principles are testable, and tested, every day. And tested in new ways every day, not just the same ways. When you find a religious text that makes predictions about how the world works that is as well tested and confirmed as the average college freshman physics textbook, I will read it with equal attention.

It would in fact fit the bill of confirming my statement that there is no such axiom in the field of AI.

First: the quote above stated:

Your article states:

Note the article refers to brains, not "any system."

Second, the article also states:

The article states that some people believe consciousness is an inevitable emergent behavior of complex brains, but other people disagree. The viewpoint is not generally accepted, and its not an axiom either way. Axioms are statements that are (as far as anyone knows) intrinsically unprovable but accepted to be true without proof.

Scientific theories aren't proven, they are confirmed through amassing sufficient supporting evidence. Mathematical conjectures are proven and provable. But there will exist no time in which the theory of relativity is "proven" except colloquially, because Science doesn't prove theories.

General Relativity makes predictions, and those predictions have been demonstrated to be true to the best extent we can measure. That means General Relativity is "true" to a Scientific certainty. One day GR might be superceded just as GR superceded Newtonian gravity, but Einstein did not prove Newton wrong. Newton was basically right: objects continued to obey Newtonian gravity after Einstein published his work on General Relativity to the best extent Newton himself could have ever confirmed. Einstein demonstrated that Newton was approximately right, but not quite right in all cases, and Relativity is much more accurate. But we still teach Newtonian physics, because 400 years later its still basically right.

Speaking about evolution specifically, the theory of natural selection states that all species arise through natural variations in generations that reward certain traits which are passed on to future generations, eventually causing different populations to distinguish themselves in ways we refer to as different species. The actual *mechanisms* of evolution are not in question: they aren't theoretical because they've been observed to function on smaller time scales and in certain situations. All of human agriculture and animal domestication demonstrates the mechanisms in action over tens, hundreds, and thousands of years, for example. That evolution is happening is not in legitimate dispute. The only legitimate dispute is whether it can account for all speciation. Believing evolution did not create all species is denying the overwhelming Scientific evidence, but denying evolution itself isn't happening at all is denying direct observational facts.

There is no such axiom in the field of Artificial Intelligence.

Snowden does not have legal protection as a whistleblower. The laws protecting whistleblowers define whistleblowers very specifically as people who report activity to appropriate channels. The laws very explicitly exempt from protection any disclosure that is in violation of the law or that discloses information designated as protected national security information by executive order. The specific relevant directive states:

The OIG is the Office of the Inspector General and the OSC is the Office of Special Counsel.

To me, quibbling over labels seems unproductive. Is Snowden a traitor? If we're talking about the law, and the Constitution so many Snowden supporters claim to support, the answer is probably yes: he committed acts which almost certainly aided enemies of the United States, and he did so with the full knowledge that the government of the United States through the Constitutionally delegated authority of the executive branch deemed such knowledge to be sensitive and potentially damaging to the US. And there exists no provision in the Constitution for a citizen to override the judgment of the Constitutionally designated officers of the Executive branch.

If you believe in the Constitution, the Constitution will eventually resolve that issue via its designated authorities in the Executive and the Judiciary. To me, the important question is what to do about the disclosures, and the focus should be on the third Constitutional authority, the Congress. Ultimately, its up to Congress to decide what the Executive can and cannot do, and its ultimately in the hands of Congress to decide if the programs Snowden disclosed should continue into the future and if so in what form. I think Snowden declared victory a little bit too early, because while a lot of discussion is going on, a significant fraction of it revolves not around the programs themselves but Snowden's conduct. And every minute people discuss whether Snowden should have disclosed the NSAs programs is a minute not spent trying to come to a consensus about what we should do about those disclosures. And from what I can tell, there isn't really a genuine consensus about how to modify or curtail the NSA's surveillance programs among Americans generally.

In the absence of a very strong consensus to the contrary, these programs will continue regardless of how superficially unpopular they may appear to be. Which would make Snowden's disclosures ultimately futile. That would be unfortunate.

Its not necessary to presume such conspiracies if you read the judicial opinions of the two judges in question. They are both actually very reasonable in their determinations, even if they are contradictory. Judge Richard Leon originally ruled that the metadata collection program was "almost certainly unconstitutional" based upon his interpretation of a modernistic view of the Fourth Amendment right to privacy. He seems to argue that while the individual collection of data does not violate the Fourth Amendment, the overall scope of the collection combined with what modern technological analysis can do with that information constitutes a new kind of meta-information for which people can have an expectation of privacy. And its that not-entirely-black-and-white notion of "expectation" of privacy that is key. Judge William Pauley in overruling Leon states that by legal precedent the expectation of privacy is broken when people voluntarily disclose information to a third party, such as the phone company. Because the phone company collects the same data the NSA does (in fact the NSA gets it from the phone company, not the customer), there is no reasonable expectation of privacy for telephone customers, because they know and voluntarily allow that information to be collected by third parties (and not just their own phone company, but all other phone companies their calls might traverse).

The problem is that they are both right. Judge Pauley is legally correct (as far as I can judge) in his opinion that the law as it exists now provides for no specific expectation of privacy for bags of data when the individual pieces themselves have no expectation of privacy. Judge Leon is also correct that modern technology has created the means to look at information in ways unforseen by the legal system when the laws were drafted or adjudicated upon, and those new areas deserve separate carefully legislated protections.

The problem is legislators are technologically illiterate in large part, and haven't filled in the gaps that need to be filled in to provide the protections people almost certainly want and expect, but that the current legal system considers "unreasonable" in the technical sense.

And there's the additional problem that we surrender our privacy every day to modern technology, not fully appreciating that there's no such thing as "privacy between me, my wife, and Google." When we allow technology companies to run parts of our lives, we're opening the door to government intrusion into those parts of our lives, short of legislation explicitly regulating what the government can do in those areas.

Its interesting to note a parallel to another recent case. Google recently won a ruling that stated that their scanning of millions of copyrighted books did not constitute copyright theft because they did not provide unrestricted access to those scans. Rather they argued they used them to create something new, their book searching system. The courts agreed, saying that Google book search represented an entirely new thing that Google was entitled to create under fair use, given that there was no way for Google users to download the books or reconstruct them through the use of searching. I wonder if a really smart lawyer could use that as a legal precedent and state that while individual telephone records are not protected, the database the NSA creates with it is an entirely new thing of the NSA's own creation and ironically contains information that individual Americans have a legitimate expectation of privacy for. In essence, by declaring that the NSA has created something new that would be worthy of copyright protection, they've also created something new people would not want disclosed under Fourth Amendment rights.

I can't channel enough Johnnie Cochran to know if that's workable or ridiculous.

Please read posts you reply to. I specifically state, in the quote above, that best evidence suggests the medieval warming period was global event, but not a large one - its effects were amplified in certain areas such as northern Europe.

Moreover the very paper you link to completely negates your next point:

You can't quote the paper above, which provides strong evidence that mid-oceanic warming is fifteen times faster than at any time in the last 10,000 years, to support the notion that the medieval warming period was a global event (which it seems to be, but not a dramatic one) and completely ignore its conclusion that oceanic warming is occurring faster than at any point in the last 10,000 years completely overriding the higher variations in regional surface temperatures and say the jury is still out. That's intellectual dishonesty.

How can you conclude the medieval warming period was a global event, when we only have accurate data for the last 35 years? One of the signs of dishonest skepticism is when it uses a different standard by which to judge contradictory evidence and supporting evidence. If we don't have enough evidence to conclude man-made global warming is occurring, we don't have enough evidence to conclude there was any sort of climatic event during the medieval warm period. It should just be a curious set of uninteresting anecdotes.

It doesn't, but that's because on a global scale there were no big swings in overall global temperature during the medieval warm period. The medieval warm period appears to be, based on the best climate evidence, a period of relatively small warming consistent with a small fluctuation in global temperature that was magnified locally in certain areas such as northern Europe. On scales of centuries, there are small changes in global climate related to fluctuations in Earth's inclination to the Sun, the Sun's 22 year cycle, and other factors. Those tend to cause small changes to overall global temperatures that can have dramatic localized changes in local climate in different parts of the world.

Combine a slightly warmer Sun and a period of relatively low volcanism (which usually tends to produce cooling), and you have a very slight global warming. Combine that with a more powerful Gulf Stream and you have a significantly more efficient transport of heat to places like Northern Europe which then experience a much stronger localized warming. Those factors are within the margin for error in explaining the global and local temperature differences during the period referred to as the Medieval Warming Period. It was globally warmer, but not by much. It was a lot warmer in Northern Europe and Greenland (back when the name was descriptive), but cooler in equatorial regions.

Its worth noting that even the medieval warming period took place over centuries. There are very few things that can change global temperatures on a timescale of decades. The only two I'm aware of are the Sun and the content of the global atmosphere. The warming that is occurring in the last hundred years is happening faster than any factor can explain except changing the content of the atmosphere. The Sun isn't currently changing by enough fast enough. All the other factors have a lot of inertia built into them: they take centuries or millenia to express themselves normally. But you can change the content of the atmosphere very fast, and that has a relatively rapid effect on changing the equilibrium of the temperature of Earth's surface. Volcanism can do it, although you would need to sustain volcanism for a decades to do it. If anything other than human beings was burning billions of tons of carbon on the surface of the Earth, there would be no controversy at all that that process was going to impact global temperatures. The fact its happening in engines and generators and not in volcanoes and forest fires is the only reason any real controversy exists at all. Try to find the large controversy over the notion that declining CO2 during the Carboniferous due to mega rainforest growth caused global cooling. Its pretty difficult, because prehistoric rainforests do not have a political lobby.

The fact that few argue sucking all that CO2 out of the atmosphere caused cooling, but there's controversy over the notion that putting it back would cause warming, is really all you need to know about the state of debate about modern climatology.