The number of failures are too low to mean much
Title says it all. With failure rates like this, you'd want the drive counts to be well over ten thousand.
95 publicly visible posts • joined 7 Nov 2007
"True democracy should (arguably?) give everyone an equal voice"
Maybe that's why democracies don't scale. At all. And notably come to bad ends. And the US was explicitly designed not to be a democracy.
"Arguably"? :) Yeah, would we want a 5 year old to have equal voice in policy?
But you've laid out some basic objections to the US's First Amendment. Anti-First Amendment is certainly the most natural, intuitive, and easy side to take. Same could be said for others of the Bill of Rights. Or, why it's hard to run an American style government. Doesn't come natural. Lotta training and tools and culture needed.
Anyway, it's unfortunate that the words "loss of democracy" have been down-cycled to mean "loss of the next election".
As others have noted, "purpose of a company is to MAKE MONEY" is false.
Due to the nature of money and also the existence of "gambler's ruin", companies *must* make money. But is your *purpose* to breath, as you must? Or to eat? Or to sleep? Or is a rat-in-a-maze's purpose to stay inside the walls, as it must, or to get to the cheese?
First, Kieren, great article. You laid out a lot of stuff clearly in limited space.
Beyond that, why hasn't C# never come up? Sun made it clear that they wanted control of not just the implementation of Java, but of the wider language - the Java "API". So Microsoft, craftily augmenting the "API" with code-in-comments and such-like kludges, decided to write their own Java, in house.
Why didn't Google do the same? Of course, we all know why. It's expensive and time consuming to design an "API".
Google's lawyer said many times that Oracle was trying to make copyright law in to patent law. Well, yes. The border between the two will always be fuzzy. And, yes, this court case is trying to nail down that border.
Just at a glance, it seems to me that Sun/Oracle made it plain that they consider their "API" to be proprietary and valuable in its own right. That is, the "API" is part of the implementation. Public, sure. But copyrighted and not reusable by others.
Other people and organizations who have built "API"s and implementations have been clear that they are copyrighting the implementation, but that the API is meant to be public and freely clone-able.
Martin King, the guy behind Tegic's T9, had an eye tracker prototype back in the '90's. Intended for people who could move an eye and little else. If I recall, it involved some LEDs and sensor diode(s) around the frame of glasses. Dirt cheap to make. The idea was to pick up the general eye direction (and blinking?) and use T9-ish logic to drive the output text.
Comments are what can really stretch out a line of "code". Most comments are *so* much nicer when they are (column aligned) to the right of the code they talk about.
People put multiple windows of code on-screen at the same time?!? Egad. If I ever did that, I'd just hang the other code on another screen. Problem solved. It's 2020, people. If you're a programmer you can spring for $2000+ in screens, no problem. Cover your frigging walls with a 75" TVs in 4k mode or something. Jeez. You don't need to run CGA emulators to save ASR-33 paper rolls and punch cards when you're on a multi-core, killer-GPU, 11-teen Gig super-computer.
Tossing my CGA board and expanding to a 134 character wide VGA blasted me with a feeling - exactly the same feeling I had when moving from an assembler requiring names no more than 6 characters to one that allowed a massive 8 characters per name. Wonderful feeling of liberation and freedom.
Linus is 20 years late. Good for him!
The description is oddly specific about the combination of attributes required to be in the software for that software to be export-restricted. On the other hand, each attribute, itself, is general enough to match one-day utility programs anyone might write to process images of any sort.
The press release is certainly shorter and more coherent than this article. The 3rd and last paragraph:
"Having confirmed Ms. Carter’s two articles, I have fulfilled those citizenship obligations of which my Rabbi reminded me. I will speak no more on the subject. Instead, having lived in places lacking Rule of Law and having witnessed the consequences of its absence, I plan on sitting back and watching the United States Department of Justice re-establish Rule of Law in our country."
The Reg's Kieren McCarthy is apparently upset by this.
Ah, a village - where everyone knows everyone else's secrets. It's the future of humanity.
It's interesting that "bias" is trotted out against facial recognition systems. Aaand, what happens when these systems are fed enough images to beat humans at the "They all look alike" game? Which of the noisemakers on either side of this issue will change their tune?
That's a very good point.
Currently, probably most people think a 3rd party DNA upload is sketchy behavior. That thought seems supported in part because DNA is still magical stuff to us. Most people don't feel comfortable playing with magical stuff that may be dangerous.
From another view, is it sketchy behavior to upload a picture of another person? In many ways, a picture is more personal and carries more information than the DNA information these sites use. How long do you figure it will be before someone starts building "AI" systems to figure out, given a large group of pictures, who is the child or parent of who?
Welcome to the global village.
@Wellyboot : Well, "silly" is in the eye of the beholder. How about entertaining? As evidenced by the number of people who have plunked down $50 to $100 to be entertained.
There certainly are objective DNA differences around Europe. And, if your location-DNA info is from recent enough, you're likely to have cultural idiosyncrasies from your DNA location. That can be interesting to ponder.
As others have mentioned, a lot of the interest in DNA ancestry comes from "immigrant nations". Where did those old folks come from? For instance, ancestry can be particularly interesting to Negros in the Western Hemisphere whose families go back to slave ships. What other way than DNA is there to get a line on where your ancestors lived in Africa? Sure, we're not talking about particularly useful information (at this time), but that doesn't stop a person from being quite curious. Those various African source locations were a loooong way apart geographically and DNA-ly. Curiosity! Which one(s) did your folks come from? Makes history and geography come alive.
There are several misled comments here.
The DNA tests don't confuse fraternal and identical twins. Therefore, forget whether you think these two women are identical. They are. Or are being purposely misleading.
The DNA outfits report around 600,000+ "SNPs" - pairs of molecules (one from each parent) that are of 4 types, labelled "A", "G", "C", and "T". The DNA companies choose their 600k+ SNPs from our 3.5 billion SNPs. The ones they choose vary in A, G, C, and T value between different humans. Almost all of our 3.5 billion SNPs are the same for all of us.
Each company chooses a different set of SNPs to report. 23andMe has reported at least two different sets of SNPs, depending on when you did their test. I've found Ancestry,com and 23andMe tests have from 100K to 300K of the same SNPs between any two test/versions (out of the 600k+ each test reports).
Two tests for the same person vary by only a handful of SNPs. That is, there is noise, but it's not very loud. Different companies' tests report different SNP values, too. Again, very quiet noise.
The serial killer thing in the article used these DNA kit results. Crime labs use something quite different.
"Ancestry" or "ethnicity" is calculated using fancy arithmetic. Different companies do, indeed, report different "ancestries". As would be expected. Issues, among others:
1) Different SNPs between companies and company-versions.
2) Ancestor? When? These guys generally are shooting for a few hundred years ago in Europe, not thousands. But outside Europe? Read on.
3) How much (if any) data backs up "ancestry"? You can't find what you don't know. These outfits have SNP combo examples from south-south-eastern Duchy of Euroland, but have bupkus for much of the world. Only recently have they started differentiating Siberia from Chile, for instance. And Siberia and Chile diverged over 10 thousand years ago!
4) Ancestry calculations depend on knowing which AGCT of the SNP pairs go to which parent. That's rather a trick to know when you don't have DNA data for either parent. The article doesn't say whether these women got their mother or father to do a test so such information could be nailed down.
The FDA doesn't care about "ancestry" reports. They "FDA'd" 23andMe a few years ago because of health reports. 23andMe has since jumped through the FDA hoops and now benefits from the usual lack-of-competition regulation causes. Don't hold your breath waiting for useful health information from these DNA tests.
I'm Scandinavian to two of the DNA outfits, Brit to one of them, and changed from Scand to Brit at a fourth outfit a year ago. Smaller percentages change often. These guys are busy tweaking their code and data.
23andMe, for one, allows you to choose how reliable you want their ancestry information to be. The article doesn't say anything about the settings these two women used. If you want 23andMe's most reliable report, don't expect a lot of specific ancestry information. "Northern Europe", folks. Or "African". Sort of at the level our eyes see without close examination.
Hope this helps.
Remember, before you flame your solution to the problem, IoT manufacturers have a life span, too.
The correct solution is regulation to the extent that nothing new is manufactured. Or, since the effect of regulations is to limit competition, regulate manufacturers down to 1. That way, nothing could possibly go wrong.
@smudge Worry about shipping around the world was one reason why I tested with a small batch for each of the two big runs I did. My first ~10,000 slide run were slides from a recently deceased parent.
So far as I know all slides went through Bangalore. Took a month or so. The movies where done in Fremont, probably. The pics, I'm not sure. They took a while, like the slides.
The pics came back in better shape than I sent them. Pictures from around 1900 can be difficult to send and I'm not the world's best packer. I shipped in boxes, each with around 5000 slides in them, plus some pics, etc.
I do know when one of the 8mm movies came out blank (no surprise, I'd already seen that, but lazily put it in the big box anyway.) they sent back an overhead video of the unpacking and handling procedure along with apologies. What I figured from that video was:
1) They probably video all handling.
2) The video showed a professional-level production setup.
Yeah, I was impressed. Surprisingly, they did recover the other videos, in very bad shape as they were.
Also, I'd done quite a bit of experimentation myself using two or three methods and equipment, etc. And, some of my own slides I'd had done by a pro photography place long ago at something like a buck or two a slide. ScanCafe's results were notably superior to all methods I tried.
@smudge I've digitized 10000 to 15000 slides and also some pictures and a handful of 8MM movies from the 1930's. All through ScanCafe. They have regular sales with considerable discounts. I have no other connection to them except being a happy customer who tells friends and relations to go that route. (Note: I'm the type of person who gives 4 stars if a product is completely satisfactory. 5 stars mean, "Wow!")
I ran a few dozen or so slides and pics through ScanCafe first to evaluate.
Hope this helps.
The unique feature a watch provides is the time and date. Instantly. Fusslessly.
Checking a phone for the time or date is a hassle. Try in in the middle of a game. Even during a time out. Your phone's off-court in a bag. Try it climbing a gonzo hill. Try it driving when the car's clock is light-washed. And, if you're doing anything more exuberant than sitting at a desk, your phone might be securely tucked in somewhere not easy to slip out.
Battery? A $15 Casio's battery lasts longer than the pins that keep the band on. And longer than the band, itself. And such a watch can be on you 24x7. No fuss. No muss.
Killer app other than the time? Fashion. Apple, Rolex, and other jewelry outfits have that covered. Boring.
Alan, you seem to have more information than the rest of us. How did you know the Uber car didn't slow down for the pedestrian?
Also, it was news to me that Arizona was like the Eastern US with regards to pedestrian right-of-way. I always assumed they were like the other Western states, which have very different customs and laws from the East. But then, from my limited experience, I would have thought that Europe in general and the UK in particular were more like our East coast: Walker beware. Certainly the attitude of a London cabbie toward my walking behavior on my first visit to that area told me to assume cars there were out to get me. I was young and grew up in the West. It was a learning experience.
The dashes fix an ambiguity problem. 12031102? Nov 2, 1203 or Dec 3, 1102 or March 12, 1102. It's very, very rare to find YYYY-DD-MM. So if you're interpreting human generated dates, the dashes pretty much force YYYY-MM-DD. And YYYY-M-D is easily interpreted as a bonus.
As others have noted, file names really, really want to be in numeric, YYYY-MM-DD order. Take note when you supply your users the file name when they download from you!
Aside from Serbia during the Yugoslavian break up, have sanctions ever been effective?
They have certainly been ineffective. Castro/Cuba comes to mind. NK comes to mind. Iran comes to mind. Iraq comes to mind. The Soviet Union comes to mind.
I've always wondered whether sanctions were something championed by closet Marxists in the US government. Incompetents, that is. Never ascribe to malice ...
Oh, sanctions *do* give the sanction-ee an excuse - someone to blame. The OP's list of excuses for Venezuela's condition may have missed a minor detail or two, after all.
Sign of a civilization in decline?
And, from a group (Reg readers) who purport to be technologically adept?
Here's a use case I put some work in to in 1990: Language learning. Consider how nice it would be if, when you're learning a language, you heard your own voice, as you hear yourself, speaking with a native accent in the new language.
Fellow reader, stop watching Hollywood post-apocalyptic zombie junk and channel your inner entrepreneur. You can come up with good uses for this tech.
First question would be, "How many users think they have been charged for WIFI?"
Second question would be, "How many users think they are using WIFI, when, in fact, they were using cell data?"
Subtract answer #2 from #1 before paying lawyers to class-action themselves to a paycheck.
From the Tesla report:
Additionally, because Tesla is the only participant in the program that has a fleet of hundreds of thousands of customer-owned vehicles that test autonomous technology in “shadow-mode” during their normal operation ..., Tesla is able to use billions of miles of real-world driving data to develop its autonomous technology. In “shadow mode,” features run in the background without actuating vehicle controls in order to provide data on how the features would perform in real world and real time conditions. This data allows Tesla to safely compare self-driving features not only to our existing Autopilot advanced driver assistance system, but also to how drivers actually drive in a wide variety of road conditions and situations.
Put another way, Tesla is Big-Brothering their cars and can conduct a Delphi Poll on what a good driver does in very, very many circumstances.
Read through these reports. In particular, Waymo and Cruise. They are logging the most miles and their trend is clear. The latest reported months have a lot more miles and a lot fewer disengagements.
Cruise notes why they drive in Frisco instead of other places: It's a harder environment than suburbia or highways, so they learn faster.
Remember, these guys *want* disengagements. Each disengagement can be gone over like an airliner crash. Replayed millions of times, varying the parameters. When you run out of disengagements, you have a problem learning, don't you?
You get what you pay for. Paying people not to work seems an odd thing to do.
A UBI may be a more honest and transparent alternative to current welfare systems. But it does make governments and taxpayers appear to be bad parents. Hey, kid, "The world does own you a living. Money does grow on trees. There is a free lunch."
Yep, I remember reading an article in Datamation, the main magazine for data processing types back in the late '50's or so.
The article touted this great new thing, COBOL. COBOL would change the world. COBOL meant future managers could write their own programs rather than relying on pesky programmers!
The article's prediction was correct.
As time went on, "programmers" became indistinguishable from the "managers" of the article writer's imagination. And they wrote programs in COBOL.
Too, what programmer out there has not been in the business of writing a replacement for themselves at some time?
GitHub is, like the Internet, a huge advance in productivity. We're all richer because of them. Good deal.
I love Python.
But it doesn't run in the browser, despite some tries.
And it doesn't run under Android, despite some tries.
Concurrency can be a problem in Python, but that's true of pretty much all languages. The spiffs of Python 3 are tangential to it being usable in a browser and under Android. So they are irrelevant for Python's future prospects.
Anyway, Python3 is simply another language than Python2. It's "easy" to translate Python2 in to various languages, including Python3. But doing so is work. Grunt work. Overhead. Friction.
We can't assume the purpose of the ML these guys used was to find the tiger. What if the purpose was to find the anomalous Audi? This is a general problem philosophers have been working on since forever.
Think: optical illusions. Is it possible to build a machine that's without such illusions?
Violence has gone down during and after industrialization. Yes, violence continued well in to the 20th century in "industrialized" areas. Violence even exists today. Violence and other bad things have been limited as the world has learning-curved earlier industrial revolution processes.
""Millions, millions and millions."" Did you mean that to imply three equal thirds of a total? It does.
Let's adjust this:
At this point (half way through the industrial revolution, worldwide), 1 billion out of 7 are living no better off than Americans 120 years ago - post most of the "industrial revolution" as many think of it. The other 6 are in relatively good shape and getting better at an incredible clip. Expect that top 80+% to be at or above your material level in a generation or so. That sounds to me like a very pretty model.
"...poverty in the U.S." 160 years ago people starved to death in places that are, today, very, very rich. (I'm thinking of Ireland, in particular.) Where are the millions of starving, bloated bellies in the U.S.?
Humans are diverse, if only in age. So 10%, or 5%, or 50%, or whatever percent is half of our own material percentage, or whatever is double our percentage if we prefer to think of ourselves as poor - will always be "poor". That's math. Crying in our beer won't help.
As others have noted, there's nothing new about "robots". Automation has been a big economic factor in many places for a couple hundred years.
So it might be interesting to apply the techniques this study used to earlier time periods. To validate the study's techniques and assumptions.
As onlookers we can do likewise with our mental models of the effects of automation.
"When I see someone say there needs to be a retraining program they often ignore the fact that many do not have any real skills for the new jobs."
Huh? Retraining programs are *based* on the assumption that "many do not have any real skills for the new jobs".
Another base assumption of retraining programs is that people can be trained. Apparently, you think training is "likely to be a complete failure." Perhaps you're right. But, others may disagree.
Half that gig. The "parameters" would probably be 32-bit floats, but there are ways to cut that down, too.
The "parameters" are generally multipliers stashed in a GPU's memory and divvied up in parallel to however many computation units the GPU has (in the 1 to 3 thousand range nowadays for a single, good GPU card).
Have a ball scanning the "patent citations" in this thing. They start in 2003. Apparently email was new then.
And what a cluster of jokes!
They seem to be like the infamous Compton's patent, one of the first software patents. Compton seemed to take a typical system description document, such as big organizations make before beginning any project, and put it in patent-ese form and language. Viola! A novel idea. Which, hey, maybe it was novel to the talent at Compton's, apparently unfamiliar with computers. Or unfamiliar with an "electronic computing device", which back in the '70's or so was what you sprinkled your description of the wheel with to make it novel and patentable.
Wanna speculate on how may patent applications there are right now sprinkling their software descriptions with "machine learning" or some such blather to make blatantly obvious ideas able to support the livelihood of patent lawyers?
Sprinkling patent applications with accepted buzzwords is one of a plurality of novel methodologies my (patent pending) automated patent generation system employs. My novel buzz-sprinkle method, utilizing machine learning technology, will be for sale as soon as it passes review. As will my novel neural-net based buzzword identification and extraction method used to support the buzzword sprinkle ... blah, blah, blah.
Which is why one tries to keep the mind alive when doing this sort of thing.
Personal experience says that such extrapolations can predict with uncanny accuracy. The uncanny part comes from:
1) The curves are exponential. The mind does not do exponential well, normally.
2) Though others see apples and oranges, the prediction is about fruit - fruit-ness that others somehow miss.
The paper seems to be behind a pay-wall, so it's not clear whether the evidence of generalization (e.g. the bees pushing a closer, differently colored ball than what they've seen pushed) is truly generalization or whether it's that the bees just don't notice the "closeness" or "color" of the ball.
Put another way, how do you distinguish cluelessness from deep thought?