You stupid USians don't know what's going to hit you when Microsoft finds out that you illegally broke away from the EU 240 years ago!
You stupid USians don't know what's going to hit you when Microsoft finds out that you illegally broke away from the EU 240 years ago!
Bollocks on their predication rate. Real forecasters report skill. By contrast, actual progress on predicting the North Atlantic Oscillation, perhaps an achievable goal, would be huge.
Both of these issues are covered in Judith Curry on Climate Change, a podcast from 2013 which, as it happens, I consumed yesterday.
Concerning the rush to embarrass themselves by reporting their weather prediction rate, it's because of the taxonomic land grab.
Host: I wonder how you feel about how your particular field has changed as you've grown up in it and been out for 25 years.
... Do you feel that we are making progress in the scientific world on this particular topic? Or are we in trouble?
Guest: I think we're in big trouble. When I left graduate school, nobody called themselves a climate scientist. They were an atmospheric dynamicist or a geochemist or a physical oceanographer or things like that. And we were all focused on increasing fundamental understanding. And that was the focus. It was the breakthrough in understanding, changing the way people think, was what mattered. And somebody who published too many papers was probably looked at with suspicion--they are doing the quick and easy stuff; they are not really digging in. It was potentially superficial.
The other thing that was looked down upon, say in the 1980s, was doing something that was too applied, working to deal with regional problems or something like that. That was viewed as soft core; it was what the people did who couldn't really make fundamental contributions to understanding, so they moved on to some of these applied topics, which were useful in some way to regional decision-makers.
I would say in 2000--it was a gradual transition, but I think circa 2000 there was a switch to people finding it beneficial to self-label them as a climate scientist. There was a lot of money, research dollars in this area; there was a lot of influence to be had, in terms of sitting on panels and boards and committees and being interviewed by journalists and being invited to testify in front of Congress. And so the value and the influence of the scientist sort of switched into that dimension where your measure of influence was not so much how you increased our fundamental understanding of how the oceans worked, but it was really to what boards and committees you sat on, your press, and your influence in policy, being invited to testify in front of Congress, and whatever. So I've seen that switch.
The problem is, the concern that I have for the health of our field, is that there's still a lot of fundamental things that we don't understand. The climate models aren't good enough. We need to go back to basics, increase our understanding about the non-linear dynamics of all these ocean oscillations and complexity of the system and things like that.
There are a lot of fundamental things that are getting short shrift, that the sex appeal in our field right now and a lot of funding is to do what I call mock 'climate model taxonomy', where people are analyzing the output of climate models and finding something interesting, alarming, or using them to infer that we won't be able to grow grapes in California in 2100 or something like this. This is the stuff that gets published in Nature and Science and PNAS. People get a press release.
Note that the word "useful" as I chose to hear it, is entirely confined to the domain of career advancement and the writing of committee-room position papers.
Two things about Russ.
One is that he doesn't connect as much as he should. He's (since) done other podcasts which talk about how the regional nature of congressional representation makes politics in America intensely regional. This is why the phrase "grapes in California" is so revealing. Only when your claims are sufficiently regionalized do you become grist for the mill, where the constant circulation of dollars sets up its own giant, oscillatory loops.
The other thing is that Russ loves to hide behind "we can't know". "If we can't know, leave things alone" eminently suits the Koch brothers ("alone", by definition, means business as usual). Russ goes mainly that far in padding their empire. The Kochs probably consider Russ as a tiny public-service inoculation against Grand Plan Reformist Flu.
On the flip side, the intro to Hardcore History contains the gravelly line from Edward R. Murrow "we are not descended from fearful men.ï" True that, except when fearfulness plays to an activist industrialist world order. Russ is, sadly, within the domain of human agency, a fearful man.
Russ's answer, therefore, to decision making under extreme uncertainty is to fall back on an ideological crap shoot. Just put the invisible hand on the steering wheel, responding to local, distributed information on a global scale. What could go wrong? Lots of things. Would it be more or less than what could go wrong it we actively attempted to steer by overselling science. Hard to say.
There's no shortage of examples of atrocious steering. There's no shortage of examples of atrocious non-steering. Please enjoy your stay on the N=1 blue marble. In a nutshell, name your spin.
It's not "gun controllers bringing it up", it's manufacturers working on them. What do you have against manufacturers developing new products?
I have absolutely nothing against manufacturers developing new gun safety products and offering them on the market. The concern with these "smart" guns is that they'll be mandated by law. This has already happened in New Jersey. The 2002 Childproof Handgun Law says that three years after "smart" guns are available for sale in the US, all guns for sale in New Jersey must be "smart". The law doesn't require that the guns be in any way reliable or have obtained any significant market share, just that they've been available for sale. So if these actually make it to market people in NJ who want reliable guns are screwed. And if any other states, or Congress, passes a similar law, then all of us are screwed.
Actually, I'd have no problem with smart guns if they were really reliable. And there's a really simple reliability screening test we can use: offer them to military and law enforcement personnel. Cops in particular should see a lot of value in smart guns because cops occasionally get shot with their own guns. However, they also need their guns to be extremely reliable, and big departments and the FBI have the institutional resources and motivation to seriously test them. So, once the technology reaches a level where police are not only willing to use smart guns but actively want them then it's fine to mandate them for civilians.
Of course, thanks to the NJ law, civilians are going to fight like hell to keep these things off the shelves, which means that the years of refinement needed to make them reliable is never going to happen. Not in the US, anyway.
I won't dream of a single (or multiple) damn quantum thing until I see an equation that describes a real-world superposition scaling limit, species type "immovable object".
I believed in Moore's law because it was on a collision course with the atom, right from day one. Even as a child, I didn't believe in a Laplacian universe, in the sense that the accumulated knowledge required to compute the deterministic outcome could exist in one place—a place smaller than the universe itself—for any value of "smaller" my small mind was capable of entertaining.
I've been reading articles about quantum computing seemingly for decades now, and not a single article has pointed out any practical scaling limit. For all these dunderheads seem to know, we could cajole the entire universe into a state of Laplacian superposition, if only we didn't suck at stacking these tiny little Lego blocks.
No scaling boundary equation widely promulgated = no credibility widely disseminated = very little fantasy action for people who don't believe in giant green men with anger management issues.
Or do you seriously think an Apple Intel CPU is more reliable than a Dell Intel CPU?
Never heard of Xeon? At least half of a high-end chip's reliability comes from the post-manufacture test procedure and binning standard.
At Apple's scale, they can negotiate any production standard with Intel that they wish to have. This isn't even uncommon, as companies like Google and Facebook are already negotiating custom Xeons for the datacenter, which certainly involves tweaking some internal chip firmware (e.g. changing cache allocation policies or thermal envelopes), all the way up to possibly adding specialized instructions and/or execution units.
Finally, far more problems arise from the mainboard and assembly quality than the underlying chip quality, but at the end of the day it all adds up.
Welcome to Supply Chain 501. It's not your father's Supply Chain 101.
That said, Apple (the company) is a cult-like Black Box of the highest order. When it serves their agenda, they make good products. When their agenda shifts with the winds of fashion—so long, sweet Mini—caveat emptor.
Once upon a time, a very nice product, too bad about the "greatness" removal tool presiding from the glass office.
There's no proof that it has anything to do with Wikileaks, but in a world of IoT devices with no thought toward security, anyone who cares to do so can mount DDOS with the power of a national entity.
What's the point of doing what Assange and Wikileaks have been doing without any moral position? He isn't helping his own case.
No, of course it is not legal to set a trap to intentionally hurt someone, even if you expect that the trap could only be activated by the person committing property theft or vandalism. Otherwise, you'd see shotguns built into burglar alarms.
Fire alarm stations sometimes shoot a blue dye which is difficult to remove or one which only shows under UV. Never stand in front of one when pulling the lever! But they are not supposed to hurt you.
And of course these booby traps generally are not as reliable as the so-called "inventor" thinks and tend to hurt the innocent.
Google no longer supports non-security questions for account recovery.
FTFY. Security questions are a joke. The answers are almost always easy for an attacker with a little bit of information about you to find, and a lot of the time the legitimate user can't remember them. Moreover, those two traits are strongly correlated: the harder it is for an attacker to find the answers, the more likely it is that the user won't be able to find them either.
Everyone should stop using them.
Google doesn't actually want your phone number for security. Google wants your phone number so that they can link the account in their database to other information that contains your phone number.
The number is to make account recovery possible in the event you've forgotten your password. The assumption is that attackers won't have access to your phone. That assumption is violated if your telco will transfer your number to the attacker's phone, of course.
If you prefer not to give your phone number to Google, don't. Just turn on two-factor auth using a non phone number-based auth method, either the Authenticator app or (better yet) a security key, or both. Then download and print out some backup 2FA codes and keep them somewhere safe. Google won't have your phone number and you won't be vulnerable to mistakes by dumb telco customer service reps.
So Elon's panties are all up in a bunch over supposedly overblown negative media coverage?
How about all the overblown positive media coverage he's been lapping up for years while running an unprofitable business that caters to wealthy customers getting taxpayer funded bonuses to buy his cars?
Isn't it fair that all of that media coverage should be withdrawn too?
They obviously know, but are legally forbidden from commenting.
I think people often forget that corporations are about the furthest thing possible from monolithic. It's entirely possible for one organization within a corporation to receive a request that is within its own ability and authority and to handle it without bothering to tell anyone else, or with only brief consultations with legal, who may not have kept any records. Given government secrecy requests/demands, that possibility grows even more likely. Further, corporations aren't static. They're constantly reorganized and even without reorgs people move around a lot, and even leave the company. There are some records of what people and organizations do, but they're usually scattered and almost never comprehensive.
It's entirely possible that they did something like this, that the system was installed and later removed, and that the only people who know about it have left the company or aren't speaking up because they were told at the time that they could never speak about it, and that the organization that was responsible for doing it and/or undoing it no longer even exists. It's possible that Yahoo's leadership's only option for finding out whether it happened is to scan old email to see if anyone discussed it via email (which may not have happened; see "government secrecy requests/demands") or to look in system configuration changleogs to find out if the system was ever deployed (and it may have been hidden under an innocuous-sounding name)... or to ask the government if the request was ever made.
Of course, my supposition here depends on a culture of cooperation with the government. I don't know if that existed at Yahoo. I think most of the major tech corporations at this point have a strong bias towards NON-cooperation, which would cause any request like this to go immediately to legal who would immediately notify the relevant C-level execs. But I have worked for corporations where the scenario I describe is totally plausible.
Yes it is but people have been trying to do that for 40 years and have not gotten it right yet so...
Wrong. Plenty of code correctness has been deployed in service of this goal.
Unfortunately, there are endemic economic and political reasons why we constantly choose the protocols and implementations that are bigger, hairier, and less continent.
All you need is a culture of kicking non-conforming implementations to the curb, and then the rigorous implementations have a chance to emerge from the weeds. Do we have such a culture? No—most of the time—no, we do not. Such a culture would cramp Megacorp style, and interfere with timeless value-adds, such as embrace and extend, closed ecosystem, DRM jungle, NIST-sanctioned algorithmic weevils, definition by implementation, documentation by implementation, etc. etc.
Far, far away in dull and dusty places like the Erlang OTP or Bernstein's qmail or Knuth's TeX—or perhaps even the Google protocol buffers for at least one lucky and unusually blessed language binding from the somewhat recent past—you just might find a rigorously coded parser or two.
For the most part, however, I agree. We'll probably never have rigorous parsers in a dominant culture of "screw everyone else", Wild West dysenteroperability.
Of course NASA passed on decades of hard-won experience. They're not psychopaths.
It went something like this:
Hire only the best and the brightest, keep the group challenged and engaged for decade upon decade, with frequent launch opportunities pushing the boundary of the possible at each and every iteration.
N.B.: Sorry, there's no silver bullet.
Well, it looks like practically every default logger for Apache/Nginx/etc. can be considered NSA spyware according to this edict.
That means you too, Slashdot.