stop being ashamed to be a citizen of the country that pioneered the modern democracy
I'm confused. Who in this discussion has expressed shame about being a citizen of the Netherlands?
Or, depending on how you want to define "pioneered the modern democracy" (and "country"), England, the Corsican Republic, or Finland?
True, the US does hold one or two records - it has the oldest surviving codified democratic (for a limited definition of "demos") constitution, for example. But claiming that we "pioneered the modern democracy" just shows a fairly dramatic lack of historical knowledge.
As for the flag-waving and other displays of patriotic fervor: I don't see any value to public masturbation over the virtues and successes, real or perceived, historical or present, of any organization I belong to. Education, yes. Analysis and discussion, certainly. Cheerleading, though - what the hell good does that do? It only discourages critical thought.
rtfd.org sends a certificate that's not valid for that domain. Not an auspicious start.
Of course, there's no reason whatsoever to use HTTPS for an uncontroversial information-only site. HTTPS-everywhere ideologues should at least get their certs right.
Here's my quick take:
- Object-capability is a good language feature; it doesn't give you everything that a true capability system (enforcing capabilities below the userland level) would, but it's still an improvement on the Von Neumann anything-is-anything free-for-all.
- Built-in cryptographic primitives are bad. 99% of developers should never touch cryptography. Cryptographic primitives are useless when they're composed into insecure protocols. Cryptography should be part of protocols, which should not be part of the core language, because they change quickly and use cases vary widely.
- Promise pipelining is probably good. It sounds good, in theory, but I haven't done any analysis of it. (Generally pipelines beat synchronization primitives, since there are so many ways to screw up the latter and they're hard for humans to reason about.)
- The adjective "metamodern" is horrible. I might excuse it in a particularly insightful bit of poststructuralist theory, but not in the description of a programming language.
- "Python-like syntax": I wouldn't call this a pro or a con, and I'm suspicious of holding it up as a language advantage. That's just asking to launch a syntax religious war. You have to choose some syntax, and either it will resemble a well-known language or it won't. Don't imagine that choice will win the approval of more than a fairly small minority of programmers, and it's unlikely to impress any significant number of computer scientists.
- As for the actual syntax, I feel it suffers from the "hey, let's use ASCII punctuation characters for random purposes derived from various other languages" problem, as so many other languages do. (This is my main complaint about most of the ML family, such as OCaml, and I find it an irritant in C, and even more so in C++.) Few people are working over a 300 bps connection these days; your syntax can be clearer and more verbose. I've looked at the samples, and I really find the syntax quite grating. If I really wanted three kinds of quoting, I'd write everything in Bourne shell.
- Contract features are good. More programmers should use them, and more languages should make that convenient. Interfaces and guards are hardly novel (most prominent OO languages have interfaces these days), but encouraging their use is a plus.
- Dynamic compiling is just eval. While this sort of thing appeals to my CS side, my software-development side regards it as highly suspect, for maintenance reasons as much as for security. (Yes, I realize the capability environment eliminates a large class of vulnerabilities.) I don't know that I'd say this is a feature I'd cut if it were my language, but if I were running a Monte-using development team, I'd make the bar very high for allowing its use. I'm not at all convinced by the "mobile code" use case; I think that's been a huge problem for ECMAScript in browsers.
Of course, the core issue is the same as with the vast majority of programming languages: opportunity costs and barriers to entry. I could easily use Monte for a personal project if I wanted. I couldn't use it for work without making a strong case to the rest of the development team and to management. I'd have to argue for it over the devils we know, and for it over all the other languages we don't currently use. I'd have to argue that it's worth the cost of having the development team learn it, and having future team members learn it, and having managers and high-level support people and others who have some stake in occasionally reading code learn it. I'd have to argue that the performance penalty is worth it; I might not care about that much (most of what I work on is I/O-bound and frequently waiting for work anyway), but it always eventually becomes
There is a lot an advertiser can tell about you from the car you drive, says Synaps.
There's a lot Synaps will be able to tell about public response when people start destroying their cameras. In most cases sneaking up from behind, probably, though there's ample evidence that many people are willing to commit crimes on camera.
Not that I would do such a thing, of course. But some folks are likely to take offense, and digital billboards are already hugely annoying. Add privacy invasion (real or perceived) to that and you're asking for trouble.
Indeed. While I'm no fan of pervasive advertising, I'd rather have random objects in my environment broadcasting low-power FM transmissions on known channels (which I am then free to avoid) than, say, broadcasting audio, which is a hell of a lot more intrusive. Or those fucking animated billboards which have begun to pollute our roads.
This is an opt-in channel. Don't opt in, and you won't even notice it's there.
And it's important to call it, oddly, a boson because that sounds like Bose in Bose-Einstein condensate.
Why "oddly"? It "sounds like Bose" because bosons are named for Bose. Or was that your point, and it was just too oblique (or I'm too dense today) for me to get it?
AIUI, not all composite bosons can form Bose-Einstein condensates. More generally, while an atom (an electrically-neutral one, not an ion) containing an even number of fermions (electrons and neucleons in this case) is a composite boson, they don't always follow exactly the Bose statistics of an elementary boson. When the wave functions of two composite bosons interact strongly, for example, other factors become significant.
But for many purposes, composite bosons behave like elementary bosons, and if you want to whip up a Bose-Einstein condensate, you'll want to start with bosons.
Watched him hit the batter pack it with a hammer, drive nails through it and cut it up with scissors all while the battery kept producing power.
Does no one see the danger? We traditionally have stopped killer robots by shooting them in the battery pack. What will we do once they're equipped with these??!
I mean, besides talking them to death with logic puzzles (the "Kirk method"). I never go anywhere without a laminated card that reads "Can a type-4 reasoner ever believe it is consistent?". That seems to do the trick.
I imagine vanishingly few trips are 600miles without a significant break.
Perhaps imagining isn't the best way to determine that. And perhaps "vanishingly few" isn't good enough for some people.
I just drove 635 miles a couple of days ago, and 830 a couple of days before that. I drive those kinds of distances several times a year. The shorter trip would just be feasible with an EV and a fast-charging station located at the appropriate point (good luck with that), but the location of that charging station would still constrain my travel plans in a way that a gasoline engine does not. Much of this trip is through mostly-uninhabited wilderness ("ranches", technically, but don't expect to see any buildings or more than a handful of other people), but even so I think the longest stretch without a gas station is less than 100 miles.
EVs are fine for many purposes. They're not yet fine for all we put ICE vehicles to. Someday, perhaps - though autonomous vehicles will reduce the market opportunity for long-range manually-operated EVs - but not today.
Really it's consumers who are sitting on high-mileage cars - not 300 mpg, but significantly higher than what we typically drive now.
Cars these days are hilariously overpowered by the standards of a few decades ago. The '85 Honda Civic CRX got up to 42 mpg city / 51 highway, under the new US EPA rating system (at the time it was rated 54/67). That was with a fuel-injected engine, but it didn't have the advantage of features (some available then, some introduced later) such as higher compression ratios, variable valve lift and timing, aspirated injectors, methods for building lighter engine components, etc.
Mostly it was an engine with a very good power-to-weight ratio moving a very light car.
These days of course most customers prioritize other things over mileage, such as passenger and cargo capacity, features, and performance. And regulations have increased burdens on engines too. But mostly what we've seen since the mid-1980s is that consumers have gotten used to vehicles with more power, and they're willing to pay a mileage premium for it.
That's not necessarily wrong of them; neither is it wrong that consumers who do care about mileage gravitate toward economy-focused hybrids (as opposed to performance-focused ones) and electric vehicles. I myself am fond of the little high-efficiency hatchbacks of yore, but I recognize that it's not where the market is right now, at least in the US - any more than diesel or manual transmissions are.
Uber doesn't gain the benefits of scale using its current business model.
Indeed. Except for software development on the base app and in a few administrative areas, their marginal costs would pretty much have to be all their costs. It's unsustainable, and their only path to long-term success would seem to be to gain enough monopoly power or customer loyalty to raise prices enormously.
I'm reminded of the old SNL skit about Change Bank - a bank that only made change. "How do you make money?" "Volume!"
Maybe they'll get enough loyal customers in some markets that they can contract sharply, cut all the losing markets (or let them atrophy after they stop subsidizing prices), and become a smaller, focused version. In the meantime, they're waging a war of attrition against competitors like Lyft, using other people's money. That's actually a fairly rational strategy. I'm dubious Kalanick thinks that far ahead, though, and nothing I've seen from him makes me think he's that strategic.
I've used many cabs over the past few decades, and rarely had any problems. Your generalization is complete bullshit, of course - have you tested every single cab company in a methodologically-sound manner - and rather suggests you're incapable of critical thinking, or too lazy to bother.
94% of all programs won't run properly without those rights.
Bullshit. I wish rubbish like this wouldn't keep getting modded Informative.
You cannot even install a simple program without elevated rights.
Many programs can be "installed" without elevation, by avoiding the MS installation model and secured parts of the filesystem tree. There's a huge range of Windows software that's packaged as a simple zipped executable. Microsoft even has some - most or all of the SysInternals collection, for example.
And to make matters worse, "elevated" means "full access, anywhere". There is no granularity, it's only "can't do jack shit" or "total control". You cannot open up the program files to install a normal program without also giving that program the ability to drop a low level driver into your system.
There's plenty of granularity. You just have to know how to manage it. Security policies and group policies, for a start. And programs can drop privileges they don't need.
Look, I'm perfectly happy to admit that Microsoft hugely bungled the permissions model from the original NT 3.1 release on up. The underlying thread-token-and-permission mechanism isn't bad, and has a lot more granularity than the classic UNIX one.[1,2] But with the initial release they made it essentially unusable. NT 4 made it usable but pretty much only for determined experts, and meanwhile they continued with the Win32-based line of completely insecure customer OSes and let stupidly insecure software flourish. It wasn't until Vista and UAC that they started to get things out of control, and then they had both a user base and a software base that were utterly unsuited for it.
But it does no one any favors to pollute the discussion with myths and half-truths.
[1] Not including the various attempts to introduce fine-grained privileges into UNIX, which go back to at least SVR4, and have in some cases had some success.
[2] Mind you, some of the privileges are still mind-bogglingly stupid. You need SeDebugPermission - which is local-admin-equivalent - to be alerted when another process exits, for example.
I have use standard accounts since Windows NT 4.0.
Same here. In fact, I think it was even possible to do this in NT 3.5, though if memory serves only console-mode applications could be elevated in that release, and even that required a third-party utility (unless you wanted to write one yourself).
I call it user laziness. I run both my work and personal Windows machines with UAC set to the strictest setting - prompt for credentials on the secure desktop - and I do quite a lot of work that requires occasional admin privilege (such as running builds that require local admin rights during the installation phase). It's not onerous.
People have been living with manual, explicit privilege elevation for decades: runas on older Windows releases (and add-ons for even earlier ones), su for UNIXy systems, operator terminals for mainframes, and so on. The modern era of minimal-effort click-a-button elevation is a trivial cost for significant protection. (UAC isn't a security boundary, but it blocks a lot of less-clever exploits). Anyone who can use a computer can quickly learn how to use it.
How would, say, a web-based image editing application "degrade gracefully and remain usable when scripting is disabled"?
Gosh, thinking is hard, isn't it?
For a start, it could display the image with text indicating why other functionality requires scripting. It could give the user the option to download the image (yes, present in the browser already; doesn't mean you can't improve the UX with an explicit link, which of course only requires HTML), edit it offline in the tool of their choice, and upload it again (which only requires an HTML form).
In any case, the existence of a small subset of "web applications" that require scripting to do anything useful does not relieve all the other fucking sites of the obligation to degrade gracefully. And even those that really do need scripting can do a much better job of explaining why.
Likewise for a web-based front end to a chat room.
Oh, you kids! We had those before Javascript was invented. Somehow we survived.
The user would have to keep clicking "check for new messages",
Horrors! God forbid anyone manually update messages. We used to do that back in the day, too, and the casualties were unimaginable.
Perhaps the vast majority of your users are unwilling? Fine, they can enable scripting. No reason not to let those who don't want scripting have the manual mechanism.
Oh, and then there's the HTTP Refresh header. Whoops - problem solved after all.
after which the server would have to retransmit even those messages that had already been transmitted to the user's browser.
Again, horrors! I can't download 10KB of text - I need that bandwidth for the 10MB of pointless images.
HTML iframe element. Fucking pages. (HTTP range requests would be nice, but existing browser UAs aren't smart enough to use them.) All of which can be served only to people who have scripting disabled (or, better, to people who have scripting disabled, or people who explicitly ask for the manual interface).
I know. So difficult! Why, it's almost like real programming.
I can't really criticize Weinstein, since no doubt there's some nonzero number of idiot admins who are using the Perspective API to filter comments, or are considering doing so.
But the Jigsaw blog post releasing the thing says that 1) it's in alpha, 2) it has both poor precision (too many false positives) and poor recall (misses many "forms", as they like to put it), and 3) it's not to be used in production. What's more, they tell you how it works - logistic regression on a supervised-learning model built from corpora evaluated by a wide range of mostly-non-expert human judges - and anyone with a decent background in NLP or ML would be able to tell you that, no, that model is not going to be very good.
(Logistic regression does have the advantage here of being a real-valued classifier, rather than a discrete one like a collection of SVMs or similar; and it has the advantage over some other regressions of reducing the influence of outliers. But it's still a single-dimensional model, with no semantic-structure analysis, and probably little in the way of syntactic-structure analysis.)
As for "algorithm-driven censorship" &c, that bandwagon is already well down the road. People who actually study digital media and online writing have been discussing all sorts of aspects about it for decades. Digital rhetoric has been an established field since at least the turn of the century, and computational rhetoric since at least 1991 (Makuta-Giluk's dissertation). Doesn't mean we don't need more people looking at it, of course, but it would be nice if folks brushed up on the existing field before launching new projects.
If all refined sugar consumption is lumped together, then there is a strong correlation between population sugar consumption and population prevalence of T2 diabetes. Additionally, changes in population t2 diabetes prevalence follow and correlate with changes in population sugar consumption.
Right. The causal relationship between HFCS and T2, insofar as there is one, is probably economic: by making it cheaper to increase the amount of sugar in processed foods, HFCS encourages consumers to increase the sugar in their diet. Maybe.
Where people in the US might once have been buying the occasional 6.5-ounce bottle of Coca-Cola as a treat, for example, the declining cost of the stuff meant that in 1955 Coke could introduce the 10- and 12-ounce "King Size" and 26-ounce "Family Size" bottles. And average portion sizes continued to grow - also helped, of course, by the move to plastic bottles and other factors. So now the typical convenience-store Coke drinker purchases a 20-ounce plastic bottle and consumes three times the sugar of his 1954 equivalent.
Of course they're paying more now. In 1954 that Coke likely cost a nickle or possible six or seven cents (Coke was in the process of discontinuing their 5-cent fixed price at the time). That's about $0.55 today, whereas that 20oz bottle of cola now probably goes for around three times that. So price per gram of sugar, adjusted for inflation, for Coca-Cola seems to be about the same now as it was in '54.
On the other hand, that's only one product, and consumer prices don't necessarily reflect industry costs. Coke has a strong brand which affords it significant protection from market forces. Also, HFCS really became a player in the 1970s, and declining sugar prices then might have accustomed consumers to eating more sugar in general.
If most people change the default to no JS, what steps should a developer of a web application take to convince prospective users that the web application is legitimate?
A good start is designing them so they degrade gracefully and remain usable when scripting is disabled.
This is excellent news. Because bees die shortly after fouling, this greatly reduces the opportunity for controversial calls by the referees. At last we can have truly reliable football matches.
Biological creatures, OTOH, tend to behave non-deterministically, that is their behaviors, given identical inputs, tend to produce varying sets of output behaviors
Care to produce any evidence to back that up?
Which is more likely: that the system under investigation is non-deterministic, or that your model of it doesn't capture the entire state, or all of the inputs, or both?
I'm aware that various arguments for the non-determinism of either the CNS or the mind (depending on the of abstraction the arguer prefers) have been advanced, for example by Penrose. I don't find Penrose's position compelling (or even mildly persuasive), but at least it's more sophisticated than handwaving appeals to a magical "Life Principle".
Slashdot Mirror
stop being ashamed to be a citizen of the country that pioneered the modern democracy
I'm confused. Who in this discussion has expressed shame about being a citizen of the Netherlands?
Or, depending on how you want to define "pioneered the modern democracy" (and "country"), England, the Corsican Republic, or Finland?
True, the US does hold one or two records - it has the oldest surviving codified democratic (for a limited definition of "demos") constitution, for example. But claiming that we "pioneered the modern democracy" just shows a fairly dramatic lack of historical knowledge.
As for the flag-waving and other displays of patriotic fervor: I don't see any value to public masturbation over the virtues and successes, real or perceived, historical or present, of any organization I belong to. Education, yes. Analysis and discussion, certainly. Cheerleading, though - what the hell good does that do? It only discourages critical thought.
You thinking that is a bad thing just makes me think that you've been at work too long today and should get your head checked.
Interesting. Your obsession with sabri's quality of life makes me think you're incapable of critical thinking.
Not everyone is you. I know that's a hard idea to grasp, but do try, eh?
rtfd.org sends a certificate that's not valid for that domain. Not an auspicious start.
Of course, there's no reason whatsoever to use HTTPS for an uncontroversial information-only site. HTTPS-everywhere ideologues should at least get their certs right.
Here's my quick take:
- Object-capability is a good language feature; it doesn't give you everything that a true capability system (enforcing capabilities below the userland level) would, but it's still an improvement on the Von Neumann anything-is-anything free-for-all.
- Built-in cryptographic primitives are bad. 99% of developers should never touch cryptography. Cryptographic primitives are useless when they're composed into insecure protocols. Cryptography should be part of protocols, which should not be part of the core language, because they change quickly and use cases vary widely.
- Promise pipelining is probably good. It sounds good, in theory, but I haven't done any analysis of it. (Generally pipelines beat synchronization primitives, since there are so many ways to screw up the latter and they're hard for humans to reason about.)
- The adjective "metamodern" is horrible. I might excuse it in a particularly insightful bit of poststructuralist theory, but not in the description of a programming language.
- "Python-like syntax": I wouldn't call this a pro or a con, and I'm suspicious of holding it up as a language advantage. That's just asking to launch a syntax religious war. You have to choose some syntax, and either it will resemble a well-known language or it won't. Don't imagine that choice will win the approval of more than a fairly small minority of programmers, and it's unlikely to impress any significant number of computer scientists.
- As for the actual syntax, I feel it suffers from the "hey, let's use ASCII punctuation characters for random purposes derived from various other languages" problem, as so many other languages do. (This is my main complaint about most of the ML family, such as OCaml, and I find it an irritant in C, and even more so in C++.) Few people are working over a 300 bps connection these days; your syntax can be clearer and more verbose. I've looked at the samples, and I really find the syntax quite grating. If I really wanted three kinds of quoting, I'd write everything in Bourne shell.
- Contract features are good. More programmers should use them, and more languages should make that convenient. Interfaces and guards are hardly novel (most prominent OO languages have interfaces these days), but encouraging their use is a plus.
- Dynamic compiling is just eval. While this sort of thing appeals to my CS side, my software-development side regards it as highly suspect, for maintenance reasons as much as for security. (Yes, I realize the capability environment eliminates a large class of vulnerabilities.) I don't know that I'd say this is a feature I'd cut if it were my language, but if I were running a Monte-using development team, I'd make the bar very high for allowing its use. I'm not at all convinced by the "mobile code" use case; I think that's been a huge problem for ECMAScript in browsers.
Of course, the core issue is the same as with the vast majority of programming languages: opportunity costs and barriers to entry. I could easily use Monte for a personal project if I wanted. I couldn't use it for work without making a strong case to the rest of the development team and to management. I'd have to argue for it over the devils we know, and for it over all the other languages we don't currently use. I'd have to argue that it's worth the cost of having the development team learn it, and having future team members learn it, and having managers and high-level support people and others who have some stake in occasionally reading code learn it. I'd have to argue that the performance penalty is worth it; I might not care about that much (most of what I work on is I/O-bound and frequently waiting for work anyway), but it always eventually becomes
There is a lot an advertiser can tell about you from the car you drive, says Synaps.
There's a lot Synaps will be able to tell about public response when people start destroying their cameras. In most cases sneaking up from behind, probably, though there's ample evidence that many people are willing to commit crimes on camera.
Not that I would do such a thing, of course. But some folks are likely to take offense, and digital billboards are already hugely annoying. Add privacy invasion (real or perceived) to that and you're asking for trouble.
Indeed. While I'm no fan of pervasive advertising, I'd rather have random objects in my environment broadcasting low-power FM transmissions on known channels (which I am then free to avoid) than, say, broadcasting audio, which is a hell of a lot more intrusive. Or those fucking animated billboards which have begun to pollute our roads.
This is an opt-in channel. Don't opt in, and you won't even notice it's there.
And it's important to call it, oddly, a boson because that sounds like Bose in Bose-Einstein condensate.
Why "oddly"? It "sounds like Bose" because bosons are named for Bose. Or was that your point, and it was just too oblique (or I'm too dense today) for me to get it?
AIUI, not all composite bosons can form Bose-Einstein condensates. More generally, while an atom (an electrically-neutral one, not an ion) containing an even number of fermions (electrons and neucleons in this case) is a composite boson, they don't always follow exactly the Bose statistics of an elementary boson. When the wave functions of two composite bosons interact strongly, for example, other factors become significant.
But for many purposes, composite bosons behave like elementary bosons, and if you want to whip up a Bose-Einstein condensate, you'll want to start with bosons.
Watched him hit the batter pack it with a hammer, drive nails through it and cut it up with scissors all while the battery kept producing power.
Does no one see the danger? We traditionally have stopped killer robots by shooting them in the battery pack. What will we do once they're equipped with these??!
I mean, besides talking them to death with logic puzzles (the "Kirk method"). I never go anywhere without a laminated card that reads "Can a type-4 reasoner ever believe it is consistent?". That seems to do the trick.
I imagine vanishingly few trips are 600miles without a significant break.
Perhaps imagining isn't the best way to determine that. And perhaps "vanishingly few" isn't good enough for some people.
I just drove 635 miles a couple of days ago, and 830 a couple of days before that. I drive those kinds of distances several times a year. The shorter trip would just be feasible with an EV and a fast-charging station located at the appropriate point (good luck with that), but the location of that charging station would still constrain my travel plans in a way that a gasoline engine does not. Much of this trip is through mostly-uninhabited wilderness ("ranches", technically, but don't expect to see any buildings or more than a handful of other people), but even so I think the longest stretch without a gas station is less than 100 miles.
EVs are fine for many purposes. They're not yet fine for all we put ICE vehicles to. Someday, perhaps - though autonomous vehicles will reduce the market opportunity for long-range manually-operated EVs - but not today.
Really it's consumers who are sitting on high-mileage cars - not 300 mpg, but significantly higher than what we typically drive now.
Cars these days are hilariously overpowered by the standards of a few decades ago. The '85 Honda Civic CRX got up to 42 mpg city / 51 highway, under the new US EPA rating system (at the time it was rated 54/67). That was with a fuel-injected engine, but it didn't have the advantage of features (some available then, some introduced later) such as higher compression ratios, variable valve lift and timing, aspirated injectors, methods for building lighter engine components, etc.
Mostly it was an engine with a very good power-to-weight ratio moving a very light car.
These days of course most customers prioritize other things over mileage, such as passenger and cargo capacity, features, and performance. And regulations have increased burdens on engines too. But mostly what we've seen since the mid-1980s is that consumers have gotten used to vehicles with more power, and they're willing to pay a mileage premium for it.
That's not necessarily wrong of them; neither is it wrong that consumers who do care about mileage gravitate toward economy-focused hybrids (as opposed to performance-focused ones) and electric vehicles. I myself am fond of the little high-efficiency hatchbacks of yore, but I recognize that it's not where the market is right now, at least in the US - any more than diesel or manual transmissions are.
Don't listen to this guy! Buy my book, which teaches you how to grow these crystals in your brain, so you'll never be without them.
Face it - if you hadn't already damaged your quantum consciousness, you'd have purchased the book already. How can you afford not to?
100 millibits per second? What's so great about that?
Uber doesn't gain the benefits of scale using its current business model.
Indeed. Except for software development on the base app and in a few administrative areas, their marginal costs would pretty much have to be all their costs. It's unsustainable, and their only path to long-term success would seem to be to gain enough monopoly power or customer loyalty to raise prices enormously.
I'm reminded of the old SNL skit about Change Bank - a bank that only made change. "How do you make money?" "Volume!"
Maybe they'll get enough loyal customers in some markets that they can contract sharply, cut all the losing markets (or let them atrophy after they stop subsidizing prices), and become a smaller, focused version. In the meantime, they're waging a war of attrition against competitors like Lyft, using other people's money. That's actually a fairly rational strategy. I'm dubious Kalanick thinks that far ahead, though, and nothing I've seen from him makes me think he's that strategic.
"My brief anecdotal experience is a compelling argument that will apply equally to everyone."
See how many words you could have saved?
Try any cab company. All much worse.
I've used many cabs over the past few decades, and rarely had any problems. Your generalization is complete bullshit, of course - have you tested every single cab company in a methodologically-sound manner - and rather suggests you're incapable of critical thinking, or too lazy to bother.
94% of all programs won't run properly without those rights.
Bullshit. I wish rubbish like this wouldn't keep getting modded Informative.
You cannot even install a simple program without elevated rights.
Many programs can be "installed" without elevation, by avoiding the MS installation model and secured parts of the filesystem tree. There's a huge range of Windows software that's packaged as a simple zipped executable. Microsoft even has some - most or all of the SysInternals collection, for example.
And to make matters worse, "elevated" means "full access, anywhere". There is no granularity, it's only "can't do jack shit" or "total control". You cannot open up the program files to install a normal program without also giving that program the ability to drop a low level driver into your system.
There's plenty of granularity. You just have to know how to manage it. Security policies and group policies, for a start. And programs can drop privileges they don't need.
Look, I'm perfectly happy to admit that Microsoft hugely bungled the permissions model from the original NT 3.1 release on up. The underlying thread-token-and-permission mechanism isn't bad, and has a lot more granularity than the classic UNIX one.[1,2] But with the initial release they made it essentially unusable. NT 4 made it usable but pretty much only for determined experts, and meanwhile they continued with the Win32-based line of completely insecure customer OSes and let stupidly insecure software flourish. It wasn't until Vista and UAC that they started to get things out of control, and then they had both a user base and a software base that were utterly unsuited for it.
But it does no one any favors to pollute the discussion with myths and half-truths.
[1] Not including the various attempts to introduce fine-grained privileges into UNIX, which go back to at least SVR4, and have in some cases had some success.
[2] Mind you, some of the privileges are still mind-bogglingly stupid. You need SeDebugPermission - which is local-admin-equivalent - to be alerted when another process exits, for example.
I have use standard accounts since Windows NT 4.0.
Same here. In fact, I think it was even possible to do this in NT 3.5, though if memory serves only console-mode applications could be elevated in that release, and even that required a third-party utility (unless you wanted to write one yourself).
I call it user laziness. I run both my work and personal Windows machines with UAC set to the strictest setting - prompt for credentials on the secure desktop - and I do quite a lot of work that requires occasional admin privilege (such as running builds that require local admin rights during the installation phase). It's not onerous.
People have been living with manual, explicit privilege elevation for decades: runas on older Windows releases (and add-ons for even earlier ones), su for UNIXy systems, operator terminals for mainframes, and so on. The modern era of minimal-effort click-a-button elevation is a trivial cost for significant protection. (UAC isn't a security boundary, but it blocks a lot of less-clever exploits). Anyone who can use a computer can quickly learn how to use it.
There's really no excuse.
How would, say, a web-based image editing application "degrade gracefully and remain usable when scripting is disabled"?
Gosh, thinking is hard, isn't it?
For a start, it could display the image with text indicating why other functionality requires scripting. It could give the user the option to download the image (yes, present in the browser already; doesn't mean you can't improve the UX with an explicit link, which of course only requires HTML), edit it offline in the tool of their choice, and upload it again (which only requires an HTML form).
In any case, the existence of a small subset of "web applications" that require scripting to do anything useful does not relieve all the other fucking sites of the obligation to degrade gracefully. And even those that really do need scripting can do a much better job of explaining why.
Likewise for a web-based front end to a chat room.
Oh, you kids! We had those before Javascript was invented. Somehow we survived.
The user would have to keep clicking "check for new messages",
Horrors! God forbid anyone manually update messages. We used to do that back in the day, too, and the casualties were unimaginable.
Perhaps the vast majority of your users are unwilling? Fine, they can enable scripting. No reason not to let those who don't want scripting have the manual mechanism.
Oh, and then there's the HTTP Refresh header. Whoops - problem solved after all.
after which the server would have to retransmit even those messages that had already been transmitted to the user's browser.
Again, horrors! I can't download 10KB of text - I need that bandwidth for the 10MB of pointless images.
HTML iframe element. Fucking pages. (HTTP range requests would be nice, but existing browser UAs aren't smart enough to use them.) All of which can be served only to people who have scripting disabled (or, better, to people who have scripting disabled, or people who explicitly ask for the manual interface).
I know. So difficult! Why, it's almost like real programming.
I can't really criticize Weinstein, since no doubt there's some nonzero number of idiot admins who are using the Perspective API to filter comments, or are considering doing so.
But the Jigsaw blog post releasing the thing says that 1) it's in alpha, 2) it has both poor precision (too many false positives) and poor recall (misses many "forms", as they like to put it), and 3) it's not to be used in production. What's more, they tell you how it works - logistic regression on a supervised-learning model built from corpora evaluated by a wide range of mostly-non-expert human judges - and anyone with a decent background in NLP or ML would be able to tell you that, no, that model is not going to be very good.
(Logistic regression does have the advantage here of being a real-valued classifier, rather than a discrete one like a collection of SVMs or similar; and it has the advantage over some other regressions of reducing the influence of outliers. But it's still a single-dimensional model, with no semantic-structure analysis, and probably little in the way of syntactic-structure analysis.)
As for "algorithm-driven censorship" &c, that bandwagon is already well down the road. People who actually study digital media and online writing have been discussing all sorts of aspects about it for decades. Digital rhetoric has been an established field since at least the turn of the century, and computational rhetoric since at least 1991 (Makuta-Giluk's dissertation). Doesn't mean we don't need more people looking at it, of course, but it would be nice if folks brushed up on the existing field before launching new projects.
If all refined sugar consumption is lumped together, then there is a strong correlation between population sugar consumption and population prevalence of T2 diabetes. Additionally, changes in population t2 diabetes prevalence follow and correlate with changes in population sugar consumption.
Right. The causal relationship between HFCS and T2, insofar as there is one, is probably economic: by making it cheaper to increase the amount of sugar in processed foods, HFCS encourages consumers to increase the sugar in their diet. Maybe.
Where people in the US might once have been buying the occasional 6.5-ounce bottle of Coca-Cola as a treat, for example, the declining cost of the stuff meant that in 1955 Coke could introduce the 10- and 12-ounce "King Size" and 26-ounce "Family Size" bottles. And average portion sizes continued to grow - also helped, of course, by the move to plastic bottles and other factors. So now the typical convenience-store Coke drinker purchases a 20-ounce plastic bottle and consumes three times the sugar of his 1954 equivalent.
Of course they're paying more now. In 1954 that Coke likely cost a nickle or possible six or seven cents (Coke was in the process of discontinuing their 5-cent fixed price at the time). That's about $0.55 today, whereas that 20oz bottle of cola now probably goes for around three times that. So price per gram of sugar, adjusted for inflation, for Coca-Cola seems to be about the same now as it was in '54.
On the other hand, that's only one product, and consumer prices don't necessarily reflect industry costs. Coke has a strong brand which affords it significant protection from market forces. Also, HFCS really became a player in the 1970s, and declining sugar prices then might have accustomed consumers to eating more sugar in general.
The daily recommendations for caloric intake is 2000 calories. This is for average person who probably doesn't exist.
Pfft. The average nonexistent person can get by just fine on 0 calories day. It's not like they'll starve into existence.
If most people change the default to no JS, what steps should a developer of a web application take to convince prospective users that the web application is legitimate?
A good start is designing them so they degrade gracefully and remain usable when scripting is disabled.
This is excellent news. Because bees die shortly after fouling, this greatly reduces the opportunity for controversial calls by the referees. At last we can have truly reliable football matches.
This is definitely one of the more hilariously misguided understandings of scientific epistemology that I've seen lately. Thanks!
Biological creatures, OTOH, tend to behave non-deterministically, that is their behaviors, given identical inputs, tend to produce varying sets of output behaviors
Care to produce any evidence to back that up?
Which is more likely: that the system under investigation is non-deterministic, or that your model of it doesn't capture the entire state, or all of the inputs, or both?
I'm aware that various arguments for the non-determinism of either the CNS or the mind (depending on the of abstraction the arguer prefers) have been advanced, for example by Penrose. I don't find Penrose's position compelling (or even mildly persuasive), but at least it's more sophisticated than handwaving appeals to a magical "Life Principle".