you're correct that the mobile networks don't have any illusion of net neutrality (sadly), but that's totally irrelevant here. SMS and voice calls travel over what are, in effect, two distinct networks. SMS travels over the control network, which has very different characteristics than the data network (in any network). it's roughly the equivalent of your ISP charging you one rate for IP access and a higher rate for the ATM link which likely underlies your ADSL connection (which I often wish they'd offer).
the sibling comment about the difference between endpoint neutrality vs. service neutrality is very important, too. the current debate over net neutrality started with some comments by at&t's whitaker about wanting to charge different endpoints differently (he was upset google was using his pipes "for free", the moron). they've since reframed the issue to be about service neutrality, in an attempt to convolute the issue. enforcing service neutrality has a lot of very serious technical drawbacks, and it's not clear that it's desirable for consumers at all; enforcing endpoint neutrality is good for everyone except oversized price-gouging monopolistic telcos.
I used to think the US charging model was outright stupid, too, but I've since seen more of the advantages and disadvantages of each model. In the US, every major operator has an email gateway, since they're assured of getting paid by the receiver; in most of the world, only a handful of operators are willing to eat the costs (such that they are) like that. As long as somebody's paying, they don't look like they're giving things away. You're absolutely right that the caller-pays model does wonders for pre-paid service. The US, however, we doesn't have different charge rates for calling fixed vs. mobile, as most of the world does; that's a huge win in my book. The US is also doing well in the distribution of free "in network" calling. There are definite advantages to having the charging model more closely approximate the resource usage.
i think you underestimate what you describe as the "opportunity cost" of SMS. take a look at some of the papers on feasibility of performing a DDoS attack via SMS. the service is a horrid hack that never should've been let into the hands of consumers. shoving data into a control channel is a huge network engineering no-no.
also, you don't mention the up-front costs here. the marginal cost may be very low, but it requires a rather impressive investment to get to that point (it's expensive to make things cheap). none of that justifies the prices we see in the US for per-message rates (but the bundled plans aren't so bad).
your point on MMS, however, is right on the money. it's just small amounts of data sent over a fairly wide data network; opportunity cost becomes just a matter of congestion (nil at this point) and marginal cost stays close to zero. they should wipe out SMS and move text messages to MMS.
speaking as the former director of a few R&D organizations:
this depends entirely on the nature of your development shop. there's two factors to consider when evaluating whether or not to allow a diverse set of tools. there's an inverse correlation between the quantity and the quality of the people you need to hire to support a wide set of tools. if you're a small shop - two people, say - you're going to need to hire real superstars to support a half-dozen languages safely. this is less a cost issue - the cost will increase, but much less than linearly - than an availability issue. better people are in higher demand. conversely, if you're a very large shop, supporting a dozen languages isn't such a big deal, and finding suitable replacements for staff losses will be easy.
my management experience has been in development groups from 3-15 developers. we've always used a diverse set of tools, and it worked very well for us. things fell apart, however, when new executive management wanted to replace skilled, experienced developers with warm bodies in india. the focus for them was cost savings, to the exclusion of all else, so they got the cheapest indian developers they could find. needless to say they weren't up to the task of maintaining more than two languages (one compiled, one scripting).
relatedly, there's a correlation (not absolute, but very strong) between the quality of programmers and the range of tools they have command of. someone who knows C, C++, Java, Perl, m68k assembly, and Pascal is going to be more interesting than someone who only knows Perl, even if i'm 100% certain the project will never involve anything except Perl. knowing C makes you a better Perl programmer. this has played out in every hiring decision i've seen. at least 9 times out of ten, the guy who knows 3-4 languages will be a better choice than one who only knows 1-2. to a lesser extent, this is even true when the desired target language isn't in the toolset. i'd look at a candidate with awk, C, Limbo, and Java experience for a Perl job at least as strongly as one who only had Perl on his list.
on another topic entirely: was anyone else confused by the article's claim that perl was the "granddaddy of the open-source scripting languages"? awk - the primary inspiration for perl - was around in 1977, with the modern version in 1985 (gnu knock-offs in 1986 and 1989 respectively). history doesn't begin with our tool of choice.
and then go on to describe something very, very unlike what i'm suggesting. and, again, i'm not "suggesting" this as a "gee, wouldn't it be nice if..." sort of idea; i'm specifically referring to extant windowing systems that behave in this way. you think i want something the order of complexity of ICCCM? ew. again, you're totally missing what i'm saying. go look at plan 9 and rio; most applications have no explicit communication with the windowing system at all. those that want it have a very simple, well-defined interface and can create child windows that way. note that this is the first time you've talked about child windows at all. the discussion up until this point has been on clicking to raise (or not) a particular window. you're making some unfounded assumptions about how logically related windows have to be behave together. you need to stop arguing with me over positions i don't hold. i'm done until you can demonstrate you've understood the model i'm talking about; all of the windowing systems from the Research Unix/Plan 9 lineage are suitable examples. i encourage you to read up on them.
oh, and networked file systems in unix don't work as well as you'd like to think they do. ever tried to use a remote system's/dev/audio? good luck with those ioctl's. and god forbid you try to do something fancy like use/dev/audio from a Solaris/sparc machine on your Linux/386 box. plan 9 and inferno fix those with a distributed file system done properly - and doing away with ioctl.
you've misunderstood. my suggestion was never "use gconf". gconf was pointed out as one potential solution to the problem introduced by your suggestion of deferring certain window management decisions to the application (and no, i'm not excluding libraries). my suggestion is simply not to introduce the problem in the first place. design the window system with sensible decisions up front, so that no subsequent application code (including libraries) have to worry about it. gconf shouldn't exist (at least for these sorts of uses).
you're not reading carefully. my "suggestion" is that the windowing system should make sensible decisions about window management up front, relieving the applications (including their libraries) of the burden of having to address those details. this is consistent with the windowing systems from the Research Unix lineage, as well as a handful of X11 window managers (for example, most of the tiling ones).
no, it's not totally irrelevant. when designing a window system, it's very important to note which decisions you're making, which you're pushing to the user, and which you're pushing to the application. your suggestion defers this decision to each application. this will result in applications behaving differently, meaning this won't be consistent within a the window system. generally a bad thing. there are ways of resolving this: for example, have applications look at a common configuration, like gconf does - moving the deferral from the application to the user. but whatever the solution, by deferring the decision, you've added significant complexity to the system as a whole. this is the same kind of error X11 makes throughout: a misunderstanding of generality incurs a significant cost in complexity. further, you're now requiring the application know a significant amount about its windowing system to get sensible behavior. this is something that "modern" unix developers take for granted, but is by no means the way it has to be - and by no means the way it should be. environments like rio in plan 9 allow applications to run knowing nothing about the windowing system they're running in (or not!) and still get sensible behavior. again, making good decisions up front reduces complexity in all subsequent layers of the system.
also: no window management code should ever be in the operating system. that's just dumb. but who ever suggested that? you're response (to what?) was the first time i heard it.
sometimes, as long as you're sufficiently close to whatever configuration the developer guessed at. and when it doesn't work, you're in a rather impressive mess trying to fix it.
...doesn't require much pre-installed (it does have some requirements, as my attempts to run it on os-400 found)...
again, it doesn't require much pre-installed as long as you're sufficiently close to what the developer assumed. os-400 is an example. my own worst experiences with it are on plan 9. even when the actual code that does things works fine, autohell just adds heaps of other things that can go wrong. something which claims to aim at portability should never require gnu tools if posix will do.
...and is universal. Universally hated, but still universal.
same as the previous point. it's very not universal. it's exactly the wrong solution to portability.
i'm not in love with cmake either, by the way. make, itslef, is cross-platform. i can't help thinking the better path would've been something that reads plain makefiles and outputs workspaces/projects for the various tools cmake supports.
the discussion is over gui design and/or features. of course hardly anyone would take a SPARCstation 1 over a modern PC for their primary computer. you're intentionally obscuring the issue.
in another post in this thread, i mentioned mux, by Rob Pike, as the 20-year-old window manager i'd point to as preferable to MS Windows. no, that doesn't mean i'd like to replace my MacBook with a DMD 5620 connected to a VAX in my basement (although it's tempting). but to compare the gui, you can look at rio in plan 9, the philosophical successor to mux with very few user-facing changes, and compare that to MS Windows. there, again, as far as the gui goes, i would - no, i do - choose rio every time, even of OS X's Aqua. but throw hardware and application support into the mix, and you get a very different answer. i still think Aqua's gui model is inferior, but the application support alone pushes it well over the line into being my primary work environment.
in terms of the gui, i'd take mux, by Rob Pike for AT&T Research Unix v9, circa 1986, in a second. simple, clean, fast, consistent, and - most importantly! - transparent. and by this point, we're just a year or two away from 8½, which is even better, including obviating the need for a separate virtual desktop system by making the windowing system inherently support recursion. the unix people went way off track when they got their gui ideas from MIT rather than Bell Labs.
feature #1 is very dependent on the rest of your environment. there's some windowing systems where it's simply crucial to usability (Acme, wmii), and others where it's a real detriment. most "modern" windowing systems are designed in such a way that it makes very little difference. i used to use Win9x with FFM hacked in, and it was a mild detriment to the experience. in most X11 window managers, it's a significant win. the point is you can't take a single feature in isolation and blindly apply it to a windowing system.
#2 is interesting. the same point above about not taking features in isolation holds, but there's more other factors here. for one thing, it confuses the heck out of novices, which is a big deal for mainstream OSs. there's also technical issues, like what do you do if the title bar is obscured? that's easy to do, and can lead to real hassle with crowded screens. my own experience was that as graphical apps became more and more prevalent, this became less desirable.
incidentally, to reinforce the point of not taking features in isolation: mux, the windowing system by Rob Pike, had a great solution to the problem in #2: there were no title bars at all, the relevant menu worked everywhere. changing the behavior of, say, OS X to not raise windows with focus without making more sweeping changes would be a disaster. i'm now curious for what the rational was behind removing this behavior when mux became 8½ on plan 9.
i've not yet used KDE4 (torrent just finished), but wanted to comment on another point:
It's like navigating the menus on my fucking cell phone. Those menus are clunky because they have to be, since screen real estate is at a premium. I can forgive that.
don't! i can't. there's no reason the menus on mobile phones have to be clunky. the obvious counter to this is the iPhone, but let's not even go that far, since that depends on input technology you don't have on other phones. even given the standard buttons, you can do much better than pretty much everyone does. there's a small handful of reasons why they suck.
operators shouldn't design the UI. okay, the rest are in no particular order, but at least in the US, this is the biggest problem. here (not sure about elsewhere), many of the operators will replace the manufacturer-provided UI with one of their own. VZW does this on nearly all their phones, but most operators do it on a large percentage of their line. the problem here is two-fold. first, the operators tend to know nothing about UI design (but more on that later), so it's simply poorly executed. the bigger problem is that the primary objective of doing so is a misguided understanding of "consistency" - they want all their phones to look and behave the same. well, that's all well and good, but they're dealing with devices with significantly different characteristics and capabilities. numbers of buttons differ, both on the face and on the side. display sizes differ. things like speakerphone capabilities differ. cpu capabilities differ. the manufacturers at least have a good understanding of the capabilities of the device and try to build the software around them.
the operators tend not even to do the level of consistency that they should be doing correct. my housemate and i both got new VZW phones from the same store on the same day. the options for text entry are functionally the same, but are called different things. the menu choices are the same, but in different places. and so on.
real UI engineers should design UIs. too often, the UI is based on some ideas the marketing department had, handed to the engineers who did the internals. there simply isn't any real design process, and there aren't any real UI engineers. i wouldn't ask a kernel hacker to write my slick web app, or vice versa, but that's almost exactly what's going on here. the operators are the worst at this, but the manufacturers (like much of the computer industry) tend to give UI design less attention than it deserves, as well.
there's virtually no competition (at least in the US). Apple's iPhone interface is pretty slick, but it's got flaws. the reason it seems as good as it does is because it's competing in a field where the options range from "entirely mediocre" to "causes headaches and hand cramps". this is largely because of the fact that people buy phones from their operator, not the manufacturer, and are discouraged or prohibited from doing otherwise. i'm hopeful that the new opening of the networks this year will improve this front.
neophobia. everybody looks at the other guy, tries to recycle the ideas cheaply, and (hopefully!) create something that doesn't suck quite as bad. there's very little in the way of actual new interfaces. Palm did this, but they've sorta self-destructed, and were only ever available on the high end. how would assigning menu items to keypad buttons work? radial menus? nested or not? my phone has a friggin' 247mhz ARM9 in it! yet the only real software advances i've seen this century are better multimedia players.
bah. i, for one, welcome our upcoming Android overlords.
I like the sibling "Type R" comment very much, but have no mod points just now, so I'm going to comment instead.
I got and enjoyed the "funny", but wanted to point out that there's not zero legitimacy to all body kit type work. Stickers are obvious no-ops, but spoilers and exhausts can be useful. It's just that most don't, because they're not designed for it. A spoiler that's actually designed for aerodynamic (rather than visual) effect will diminish lift (or in extreme cases, create drag, but those are huge), thus improving traction, useful for high-speed handling. I don't know of an easy way to visually distinguish between the junk and the goods, so the <10% goods get lumped in with the >90% junk. Exhaust systems replacements that go at least back to the catalytic converter (preferably all the way to the engine block) can improve airflow, thus improving engine HP. To detect junk, look for things which are just shiny chrome heads on the end of the pipe, or a muffler replacement; those're just designed to be shiny and loud. You actually see a reasonable percentage of these exhaust kits that have performance effect (although it's not uncommon to see a real performance-enhancing exhaust kit combined with a performance-limiting but shiny noise producing tail).
what are you talking about? what "computer skills" are kids not going to be able to learn on the XO? Microsoft Word? oh, the horror.
please. OLPC isn't about vocational training for office clerks, it's about fundamental educational opportunities. textbook replacement, communication, experimentation, classroom tools, and so on. for all those things, it's a nearly ideal platform. the fact that it doesn't run MS Office is just gravy.
first off, that link is really neat. thanks for that.
Geez, it is only about 20% of the budget.
you say that like it's a small number. first of all 20% of the money the government takes from me goes to the military? that's a lot! even by percentage. it's at least double what it should be. and by dollar value, that's just about half a trillion dollars. every year! that's just insane. we could send every college student to school free for less than that. every student. or pay down our overwhelming, ever-growing, interest-bearing debt.
I don't think that is too much money to keep a standing military with trained personnel, weapon systems (maintenance and development), etc.
your argument here is flawed; you're attacking the wrong point. sure, maybe what we pay for our military is a good value for the military we get in return. it's certainly the best in the world, and that costs. fine. but is that really what we want? your argument says "if you want that military, you've got to pay for it", but the opening premise is wrong. i, and most who argue for military budget cuts, fully recognize (often enthusiastically) that cutting the military budget will result in a smaller armed forces. but that's not a bad thing. i don't disagree that there's plenty of other fat to trim, and i'd even agree with most of your examples. but that's no reason to give the single largest chunk of my income tax hit a free walk.
you're welcome to continue to think that, i guess, even though it's disgusting and abhorrent. it certainly isn't the way democratic government is meant to run, though. it's not what was advertised - a fair trial - and represents a remarkably bad way to start a government. you don't want a dressed-up lynching setting the tone for your legal system. Hussein's trial needed to be beyond reproach to get any sort of credibility; instead, it barely flirted with being passable.
on the "average people" thing: the point isn't that insurgents behave like average people, but that average people can easily become insurgents when put under the wrong conditions; conditions like an invasion and subsequent occupation by a foreign power. and the comparison to the nazis is unfounded. 1930s Germany had a pre-existing military machine, which became Nazi by fiat and definition; it's akin to saying our military is Republican now, but was Democratic under Clinton. it tells you nothing about the people who signed up.
...I don't believe that the average terrorist in Iraq is an average Iraqi.
you're looking at it backwards from set theory. the statement isn't that the average Iraqi is an insurgent, but that the insurgents are, for the most part, average Iraqis. similarly, the average American isn't in the army, but the average US soldier comes from "average america", as opposed to being mercenaries, psycho killers, or foreign nationals.
The average Iraqi is now more afraid of the terrorists then the US military.
oh, i think it's fair to say the average Iraqi is plenty afraid of both the insurgents and the US military. yay, we're no longer the worst killers in the country. is that supposed to make things better?
America has a long history of fighting pirates and terrorists.
sure... when our money's on the line. the marines were on the shores of Tripoli because the protection money we'd been paying to the Barbary pirates got too expensive. that's right, our "long history" includes significant instances of paying off the same pirates and terrorists. Iran-Contra, anyone? How 'bout the fact that the Iraqi military, while killing masses of its own people, was being supported by the US?
...he's been upfront about all of this since 9/11.
that's just a lie. Bush went to great pains to convince America that Iraq was somehow a threat to us, while he knew (or reasonably should have known) that it was obviously false. the WMD line, the original justification for the war, was a lie. the "liberating the Iraqi people" line, the backpedal after the WMD lie was exposed, was a lie. he's been upfront about nothing.
you mention the taliban and afghanistan. see, if that were what it was about, great - somebody comes after us, we're going to knock 'em down. but we left afghanistan before the job was done to go after someone who was no threat to us whatsoever. we let bin Laden go when we had him cornered. and as a result, the ex-Taliban now control a significant portion of Afghanistan again, with al Queda having free run of those areas. the Taliban was a repressive, authoritarian regime doing serious harm to the Afghan people even before they funded and supported al Queda in their attack on the United States. they were as much a terrorist state as you can get, and they deserved to go down. sadly, Bush decided Iraq was a more lucrative target.
and Saddam Hussein was tried by a kangaroo court and lynched by a mob. the entire thing makes me sick.
sure, but the point is they're not fixed groups; there's changeover. if all we had to do was kill the terrorists who were there when Bush declared an end to major combat operations, he would've been right, and most of our troops would've been home by now. given that this crossover happens, it's important to look at how and why. otherwise we'll end up fighting most of the 20-something million Iraqis eventually.
the elimination of people from the military and government for political reasons is hardly new, though. probably the most significant 20th century example was Stalin's purge of the Soviet military. that resulted in a well-funded, well-equipped, powerful military... that was totally incompetent. the resulting whooping that the Soviets took from the vastly outnumbered Finns during the Winter War gave Hitler the guts to try and invade Russia. sadly, i suspect we're in the middle of the first 21st century example of the same mess, with us as Russia and Iraq as Finland.
the role of Germany has not yet been cast, but Iran's audition is going well.
the fundamentals of the idea are good, but i think the XO isn't the right vehicle (for this; i'm a big supporter generally and encourage everyone to check out the Give One Get One program).
your idea is based on two points: first, do something constructive instead of destructive, and second, encourage communication. both are excellent. the military is actually doing some of the former, but not nearly enough. we did a lot of damage to infrastructure in Iraq during the invasion, and the standards there were never all that high. the army's sole purpose at this point (really since the "Mission Accomplished" nonsense) should be building infrastructure (water, sewage, housing, power, transportation, and telecommunications) and defense of those assets. if the brass needs to feel like they're doing something more "active", let the marines (who're more "surgical" in their actions, and a lot smaller) go after only confirmed foreign military units acting in Iraq; better yet, though, stick 'em on border patrol and training. the key point here is let the Iraqi military and police forces go after Iraqi terrorists; the US needs to stop killing Iraqis if we're going to stop generating new terrorists. the second point is trickier. the poster below who noted the huge cultural issues is right: direct communication isn't likely to do much in many cases. i don't have a good answer for fixing the American side of the equation, but the Iraqi side is much easier. Iraq is still a resource-rich country; fix the infrastructure so that business can work normally and normal trade will shortly resume. it's not quite as direct as letter writing, but international trade with close neighbors does a good job of helping people get along better. this has been true for at least a millennium (contrast the attitudes of eastern and western Christians towards their Muslim neighbors around the time of the crusades, for example), and has only increased with the industrial and then technological revolutions.
the idea has flaws (the cultural barrier noted by your sibling is very significant), but is not totally off-base. your criticism of terrorists vs. average people is a far more troubling over-simplification, on two counts. first, terrorists need a support network. aside from the active combatants, they need to get food and supplies from others. they need people to hide them, or at the very least turn a blind eye. convert the masses, and the operational support the terrorists rely on goes away. the insurrection will be very short-lived at that point. second, the terrorists are average people, at least in bulk. sure, there's a few ex-Iraqi-military types, and the rumors of Iranian and Syrian military are probably not entirely false, but they certainly aren't the majority of the combatants. the occupation and our current (and recent) actions turn average people into terrorists, or at the very least give them a good shove in that direction.
a day or two after Bush talked about how the Iraqi Olympic football (soccer) team was such a shining example of the liberation of Iraq i read an interview with one of the members saying if he wasn't at the Olympics, he'd be home fighting in Fallujah (this is when that was the current hotspot) - against the Americans. so much for being greeted as liberators. and this is hardly an entirely foreign concept: i think Bush is probably our worst president ever and is actively undermining the foundations upon which the Republic is built, but if the Chinese decided we needed a regime change, invaded and occupied us, i don't have to squint very hard to see myself out in the streets fighting them off. because it's not their country, and they should get the hell out. occupation and military action breeds resentment and hostility, universally.
we got our couch. but now i'm waiting for another housemate to wake up so we can move mattresses from the basement to the attic. so, bored again.
yup, read 'em both, and i still think you're wrong. i think you're wrong on the quantity of violence in the old testament; there's whole genocides in there! cities to salt, slaughter of the first born, fire and brimstone falling from the sky, and a flood that wipes out nearly the whole human population. and no orders of violence? for a good time, check out what Deuteronomy orders us to do to disobedient children. that's a real high point. no, it's not controversial to point out that different religions are different. but you're ignoring history. i agree, Jesus' portrayal in Christian texts is much better (by any secular or objective standards) than Mohammed's portrayal in Islamic texts. but you need to keep in mind that every major Christian denomination claims the old testament as part of their Bible. that means Christians don't get to just point to Jesus and say "see? all good!". we also have to own up to David, a military conquerer with... "significant moral blemishes." we have to own up to Abraham's marriage practices. Noah's drunkenness and the resulting incest (and these were the best folks God could find!). i could believe that, on some percentage basis, the Koran is more violent than the Old Testament (i don't believe that at current, based on my reading, but i've only read the Koran once, don't own one to compare, and probably wouldn't bother with the lengthy textual comparison if i did (i'm not that bored). the point is i could be convinced). my point is that the differences are, at best, one of degree, and not a huge degree, either.
i would like a citation for the correlation, but not because i disbelieve it. i agree it's there. but the logical step you're missing is going from that correlation to some particular action. the same argument works for pornography: i think there's plenty of cases where it leads viewers to objectification of women and associated evils. but in neither case does that necessarily lead to suggesting any particular action. where is the boundary between personal responsibility and government mandates or interference? i agree that parents need to be involved here, but i'm unclear how you get from that to the v-chip (on video games? what?).
i've not seen any reputable poll indicating congress's ratings to be worse than bush's; please cite. but that's entirely beyond the point. you're argument was that any comment about bush would be censored. that has not, as yet, happened. you're even free to make unfounded, libelous comments about Obama being a liar without significant fear of incarceration. the point is neither of our posts has been removed or be mod-bombed. my timer's still running.
abortion was an option well more than 200 years ago. methods are described, written in a medical text, at least as early as Soranus, the Greek physician who died in 128AD and wrote the book on Gynaecology (quite literally). it shows up in several other cultures, too, Devereux wrote, in "A typological study of abortion in 350 primitive, ancient, and pre-industrial societies" that "There is every indication that abortion is an absolutely universal phenomenon, and that it is impossible even to construct an imaginary social system in whiich no woman would ever feel at least compelled to abort." and in what jurisdiction are you talking about this double homicide thing? it certainly wasn't universal, nor the biblical prescription. but, again, you're not arguing the point you initially opened with. you've done nothing to prove that your definition of terms is universal, which is a pre-requisite to the rest of your argument on this point. please do that first. you might start by explaining the Biblical passages which very clearly state that the human form is given a soul much, much later than conception: particularly, the first breath. see, among others, Gen 2:7, Job 33:4, and Ezk 37:1-6.
Slashdot Mirror
you're correct that the mobile networks don't have any illusion of net neutrality (sadly), but that's totally irrelevant here. SMS and voice calls travel over what are, in effect, two distinct networks. SMS travels over the control network, which has very different characteristics than the data network (in any network). it's roughly the equivalent of your ISP charging you one rate for IP access and a higher rate for the ATM link which likely underlies your ADSL connection (which I often wish they'd offer).
the sibling comment about the difference between endpoint neutrality vs. service neutrality is very important, too. the current debate over net neutrality started with some comments by at&t's whitaker about wanting to charge different endpoints differently (he was upset google was using his pipes "for free", the moron). they've since reframed the issue to be about service neutrality, in an attempt to convolute the issue. enforcing service neutrality has a lot of very serious technical drawbacks, and it's not clear that it's desirable for consumers at all; enforcing endpoint neutrality is good for everyone except oversized price-gouging monopolistic telcos.
I used to think the US charging model was outright stupid, too, but I've since seen more of the advantages and disadvantages of each model. In the US, every major operator has an email gateway, since they're assured of getting paid by the receiver; in most of the world, only a handful of operators are willing to eat the costs (such that they are) like that. As long as somebody's paying, they don't look like they're giving things away. You're absolutely right that the caller-pays model does wonders for pre-paid service. The US, however, we doesn't have different charge rates for calling fixed vs. mobile, as most of the world does; that's a huge win in my book. The US is also doing well in the distribution of free "in network" calling. There are definite advantages to having the charging model more closely approximate the resource usage.
i think you underestimate what you describe as the "opportunity cost" of SMS. take a look at some of the papers on feasibility of performing a DDoS attack via SMS. the service is a horrid hack that never should've been let into the hands of consumers. shoving data into a control channel is a huge network engineering no-no.
also, you don't mention the up-front costs here. the marginal cost may be very low, but it requires a rather impressive investment to get to that point (it's expensive to make things cheap). none of that justifies the prices we see in the US for per-message rates (but the bundled plans aren't so bad).
your point on MMS, however, is right on the money. it's just small amounts of data sent over a fairly wide data network; opportunity cost becomes just a matter of congestion (nil at this point) and marginal cost stays close to zero. they should wipe out SMS and move text messages to MMS.
speaking as the former director of a few R&D organizations:
this depends entirely on the nature of your development shop. there's two factors to consider when evaluating whether or not to allow a diverse set of tools. there's an inverse correlation between the quantity and the quality of the people you need to hire to support a wide set of tools. if you're a small shop - two people, say - you're going to need to hire real superstars to support a half-dozen languages safely. this is less a cost issue - the cost will increase, but much less than linearly - than an availability issue. better people are in higher demand. conversely, if you're a very large shop, supporting a dozen languages isn't such a big deal, and finding suitable replacements for staff losses will be easy.
my management experience has been in development groups from 3-15 developers. we've always used a diverse set of tools, and it worked very well for us. things fell apart, however, when new executive management wanted to replace skilled, experienced developers with warm bodies in india. the focus for them was cost savings, to the exclusion of all else, so they got the cheapest indian developers they could find. needless to say they weren't up to the task of maintaining more than two languages (one compiled, one scripting).
relatedly, there's a correlation (not absolute, but very strong) between the quality of programmers and the range of tools they have command of. someone who knows C, C++, Java, Perl, m68k assembly, and Pascal is going to be more interesting than someone who only knows Perl, even if i'm 100% certain the project will never involve anything except Perl. knowing C makes you a better Perl programmer. this has played out in every hiring decision i've seen. at least 9 times out of ten, the guy who knows 3-4 languages will be a better choice than one who only knows 1-2. to a lesser extent, this is even true when the desired target language isn't in the toolset. i'd look at a candidate with awk, C, Limbo, and Java experience for a Perl job at least as strongly as one who only had Perl on his list.
on another topic entirely: was anyone else confused by the article's claim that perl was the "granddaddy of the open-source scripting languages"? awk - the primary inspiration for perl - was around in 1977, with the modern version in 1985 (gnu knock-offs in 1986 and 1989 respectively). history doesn't begin with our tool of choice.
note that this is the first time you've talked about child windows at all. the discussion up until this point has been on clicking to raise (or not) a particular window. you're making some unfounded assumptions about how logically related windows have to be behave together.
you need to stop arguing with me over positions i don't hold. i'm done until you can demonstrate you've understood the model i'm talking about; all of the windowing systems from the Research Unix/Plan 9 lineage are suitable examples. i encourage you to read up on them.
oh, and networked file systems in unix don't work as well as you'd like to think they do. ever tried to use a remote system's
you've misunderstood. my suggestion was never "use gconf". gconf was pointed out as one potential solution to the problem introduced by your suggestion of deferring certain window management decisions to the application (and no, i'm not excluding libraries). my suggestion is simply not to introduce the problem in the first place. design the window system with sensible decisions up front, so that no subsequent application code (including libraries) have to worry about it. gconf shouldn't exist (at least for these sorts of uses).
you're not reading carefully. my "suggestion" is that the windowing system should make sensible decisions about window management up front, relieving the applications (including their libraries) of the burden of having to address those details. this is consistent with the windowing systems from the Research Unix lineage, as well as a handful of X11 window managers (for example, most of the tiling ones).
um, wow. switch to decaf?
no, it's not totally irrelevant. when designing a window system, it's very important to note which decisions you're making, which you're pushing to the user, and which you're pushing to the application. your suggestion defers this decision to each application. this will result in applications behaving differently, meaning this won't be consistent within a the window system. generally a bad thing.
there are ways of resolving this: for example, have applications look at a common configuration, like gconf does - moving the deferral from the application to the user. but whatever the solution, by deferring the decision, you've added significant complexity to the system as a whole. this is the same kind of error X11 makes throughout: a misunderstanding of generality incurs a significant cost in complexity.
further, you're now requiring the application know a significant amount about its windowing system to get sensible behavior. this is something that "modern" unix developers take for granted, but is by no means the way it has to be - and by no means the way it should be. environments like rio in plan 9 allow applications to run knowing nothing about the windowing system they're running in (or not!) and still get sensible behavior. again, making good decisions up front reduces complexity in all subsequent layers of the system.
also: no window management code should ever be in the operating system. that's just dumb. but who ever suggested that? you're response (to what?) was the first time i heard it.
stop yelling, think more, write clearly.
yes, exactly right. thanks for the correction.
i'm not in love with cmake either, by the way. make, itslef, is cross-platform. i can't help thinking the better path would've been something that reads plain makefiles and outputs workspaces/projects for the various tools cmake supports.
the discussion is over gui design and/or features. of course hardly anyone would take a SPARCstation 1 over a modern PC for their primary computer. you're intentionally obscuring the issue.
in another post in this thread, i mentioned mux, by Rob Pike, as the 20-year-old window manager i'd point to as preferable to MS Windows. no, that doesn't mean i'd like to replace my MacBook with a DMD 5620 connected to a VAX in my basement (although it's tempting). but to compare the gui, you can look at rio in plan 9, the philosophical successor to mux with very few user-facing changes, and compare that to MS Windows. there, again, as far as the gui goes, i would - no, i do - choose rio every time, even of OS X's Aqua. but throw hardware and application support into the mix, and you get a very different answer. i still think Aqua's gui model is inferior, but the application support alone pushes it well over the line into being my primary work environment.
in terms of the gui, i'd take mux, by Rob Pike for AT&T Research Unix v9, circa 1986, in a second. simple, clean, fast, consistent, and - most importantly! - transparent. and by this point, we're just a year or two away from 8½, which is even better, including obviating the need for a separate virtual desktop system by making the windowing system inherently support recursion. the unix people went way off track when they got their gui ideas from MIT rather than Bell Labs.
feature #1 is very dependent on the rest of your environment. there's some windowing systems where it's simply crucial to usability (Acme, wmii), and others where it's a real detriment. most "modern" windowing systems are designed in such a way that it makes very little difference. i used to use Win9x with FFM hacked in, and it was a mild detriment to the experience. in most X11 window managers, it's a significant win. the point is you can't take a single feature in isolation and blindly apply it to a windowing system.
#2 is interesting. the same point above about not taking features in isolation holds, but there's more other factors here. for one thing, it confuses the heck out of novices, which is a big deal for mainstream OSs. there's also technical issues, like what do you do if the title bar is obscured? that's easy to do, and can lead to real hassle with crowded screens. my own experience was that as graphical apps became more and more prevalent, this became less desirable.
incidentally, to reinforce the point of not taking features in isolation: mux, the windowing system by Rob Pike, had a great solution to the problem in #2: there were no title bars at all, the relevant menu worked everywhere. changing the behavior of, say, OS X to not raise windows with focus without making more sweeping changes would be a disaster. i'm now curious for what the rational was behind removing this behavior when mux became 8½ on plan 9.
operators shouldn't design the UI. okay, the rest are in no particular order, but at least in the US, this is the biggest problem. here (not sure about elsewhere), many of the operators will replace the manufacturer-provided UI with one of their own. VZW does this on nearly all their phones, but most operators do it on a large percentage of their line. the problem here is two-fold. first, the operators tend to know nothing about UI design (but more on that later), so it's simply poorly executed. the bigger problem is that the primary objective of doing so is a misguided understanding of "consistency" - they want all their phones to look and behave the same. well, that's all well and good, but they're dealing with devices with significantly different characteristics and capabilities. numbers of buttons differ, both on the face and on the side. display sizes differ. things like speakerphone capabilities differ. cpu capabilities differ. the manufacturers at least have a good understanding of the capabilities of the device and try to build the software around them.
the operators tend not even to do the level of consistency that they should be doing correct. my housemate and i both got new VZW phones from the same store on the same day. the options for text entry are functionally the same, but are called different things. the menu choices are the same, but in different places. and so on.
real UI engineers should design UIs. too often, the UI is based on some ideas the marketing department had, handed to the engineers who did the internals. there simply isn't any real design process, and there aren't any real UI engineers. i wouldn't ask a kernel hacker to write my slick web app, or vice versa, but that's almost exactly what's going on here. the operators are the worst at this, but the manufacturers (like much of the computer industry) tend to give UI design less attention than it deserves, as well.
there's virtually no competition (at least in the US). Apple's iPhone interface is pretty slick, but it's got flaws. the reason it seems as good as it does is because it's competing in a field where the options range from "entirely mediocre" to "causes headaches and hand cramps". this is largely because of the fact that people buy phones from their operator, not the manufacturer, and are discouraged or prohibited from doing otherwise. i'm hopeful that the new opening of the networks this year will improve this front.
neophobia. everybody looks at the other guy, tries to recycle the ideas cheaply, and (hopefully!) create something that doesn't suck quite as bad. there's very little in the way of actual new interfaces. Palm did this, but they've sorta self-destructed, and were only ever available on the high end. how would assigning menu items to keypad buttons work? radial menus? nested or not? my phone has a friggin' 247mhz ARM9 in it! yet the only real software advances i've seen this century are better multimedia players.
bah. i, for one, welcome our upcoming Android overlords.
I like the sibling "Type R" comment very much, but have no mod points just now, so I'm going to comment instead.
I got and enjoyed the "funny", but wanted to point out that there's not zero legitimacy to all body kit type work. Stickers are obvious no-ops, but spoilers and exhausts can be useful. It's just that most don't, because they're not designed for it. A spoiler that's actually designed for aerodynamic (rather than visual) effect will diminish lift (or in extreme cases, create drag, but those are huge), thus improving traction, useful for high-speed handling. I don't know of an easy way to visually distinguish between the junk and the goods, so the <10% goods get lumped in with the >90% junk. Exhaust systems replacements that go at least back to the catalytic converter (preferably all the way to the engine block) can improve airflow, thus improving engine HP. To detect junk, look for things which are just shiny chrome heads on the end of the pipe, or a muffler replacement; those're just designed to be shiny and loud. You actually see a reasonable percentage of these exhaust kits that have performance effect (although it's not uncommon to see a real performance-enhancing exhaust kit combined with a performance-limiting but shiny noise producing tail).
what are you talking about? what "computer skills" are kids not going to be able to learn on the XO? Microsoft Word? oh, the horror.
please. OLPC isn't about vocational training for office clerks, it's about fundamental educational opportunities. textbook replacement, communication, experimentation, classroom tools, and so on. for all those things, it's a nearly ideal platform. the fact that it doesn't run MS Office is just gravy.
i don't disagree that there's plenty of other fat to trim, and i'd even agree with most of your examples. but that's no reason to give the single largest chunk of my income tax hit a free walk.
you're welcome to continue to think that, i guess, even though it's disgusting and abhorrent. it certainly isn't the way democratic government is meant to run, though. it's not what was advertised - a fair trial - and represents a remarkably bad way to start a government. you don't want a dressed-up lynching setting the tone for your legal system. Hussein's trial needed to be beyond reproach to get any sort of credibility; instead, it barely flirted with being passable.
on the "average people" thing: the point isn't that insurgents behave like average people, but that average people can easily become insurgents when put under the wrong conditions; conditions like an invasion and subsequent occupation by a foreign power.
and the comparison to the nazis is unfounded. 1930s Germany had a pre-existing military machine, which became Nazi by fiat and definition; it's akin to saying our military is Republican now, but was Democratic under Clinton. it tells you nothing about the people who signed up.you're looking at it backwards from set theory. the statement isn't that the average Iraqi is an insurgent, but that the insurgents are, for the most part, average Iraqis. similarly, the average American isn't in the army, but the average US soldier comes from "average america", as opposed to being mercenaries, psycho killers, or foreign nationals.oh, i think it's fair to say the average Iraqi is plenty afraid of both the insurgents and the US military. yay, we're no longer the worst killers in the country. is that supposed to make things better?sure... when our money's on the line. the marines were on the shores of Tripoli because the protection money we'd been paying to the Barbary pirates got too expensive. that's right, our "long history" includes significant instances of paying off the same pirates and terrorists. Iran-Contra, anyone? How 'bout the fact that the Iraqi military, while killing masses of its own people, was being supported by the US?that's just a lie. Bush went to great pains to convince America that Iraq was somehow a threat to us, while he knew (or reasonably should have known) that it was obviously false. the WMD line, the original justification for the war, was a lie. the "liberating the Iraqi people" line, the backpedal after the WMD lie was exposed, was a lie. he's been upfront about nothing.
you mention the taliban and afghanistan. see, if that were what it was about, great - somebody comes after us, we're going to knock 'em down. but we left afghanistan before the job was done to go after someone who was no threat to us whatsoever. we let bin Laden go when we had him cornered. and as a result, the ex-Taliban now control a significant portion of Afghanistan again, with al Queda having free run of those areas.
the Taliban was a repressive, authoritarian regime doing serious harm to the Afghan people even before they funded and supported al Queda in their attack on the United States. they were as much a terrorist state as you can get, and they deserved to go down. sadly, Bush decided Iraq was a more lucrative target.
and Saddam Hussein was tried by a kangaroo court and lynched by a mob. the entire thing makes me sick.
sure, but the point is they're not fixed groups; there's changeover. if all we had to do was kill the terrorists who were there when Bush declared an end to major combat operations, he would've been right, and most of our troops would've been home by now. given that this crossover happens, it's important to look at how and why. otherwise we'll end up fighting most of the 20-something million Iraqis eventually.
the elimination of people from the military and government for political reasons is hardly new, though. probably the most significant 20th century example was Stalin's purge of the Soviet military. that resulted in a well-funded, well-equipped, powerful military... that was totally incompetent. the resulting whooping that the Soviets took from the vastly outnumbered Finns during the Winter War gave Hitler the guts to try and invade Russia. sadly, i suspect we're in the middle of the first 21st century example of the same mess, with us as Russia and Iraq as Finland.
the role of Germany has not yet been cast, but Iran's audition is going well.
oh, yeah: and get the fsck'ing mercs out! that's just stupid.
the fundamentals of the idea are good, but i think the XO isn't the right vehicle (for this; i'm a big supporter generally and encourage everyone to check out the Give One Get One program).
your idea is based on two points: first, do something constructive instead of destructive, and second, encourage communication. both are excellent.
the military is actually doing some of the former, but not nearly enough. we did a lot of damage to infrastructure in Iraq during the invasion, and the standards there were never all that high. the army's sole purpose at this point (really since the "Mission Accomplished" nonsense) should be building infrastructure (water, sewage, housing, power, transportation, and telecommunications) and defense of those assets. if the brass needs to feel like they're doing something more "active", let the marines (who're more "surgical" in their actions, and a lot smaller) go after only confirmed foreign military units acting in Iraq; better yet, though, stick 'em on border patrol and training. the key point here is let the Iraqi military and police forces go after Iraqi terrorists; the US needs to stop killing Iraqis if we're going to stop generating new terrorists.
the second point is trickier. the poster below who noted the huge cultural issues is right: direct communication isn't likely to do much in many cases. i don't have a good answer for fixing the American side of the equation, but the Iraqi side is much easier. Iraq is still a resource-rich country; fix the infrastructure so that business can work normally and normal trade will shortly resume. it's not quite as direct as letter writing, but international trade with close neighbors does a good job of helping people get along better. this has been true for at least a millennium (contrast the attitudes of eastern and western Christians towards their Muslim neighbors around the time of the crusades, for example), and has only increased with the industrial and then technological revolutions.
the idea has flaws (the cultural barrier noted by your sibling is very significant), but is not totally off-base. your criticism of terrorists vs. average people is a far more troubling over-simplification, on two counts.
first, terrorists need a support network. aside from the active combatants, they need to get food and supplies from others. they need people to hide them, or at the very least turn a blind eye. convert the masses, and the operational support the terrorists rely on goes away. the insurrection will be very short-lived at that point.
second, the terrorists are average people, at least in bulk. sure, there's a few ex-Iraqi-military types, and the rumors of Iranian and Syrian military are probably not entirely false, but they certainly aren't the majority of the combatants. the occupation and our current (and recent) actions turn average people into terrorists, or at the very least give them a good shove in that direction.
a day or two after Bush talked about how the Iraqi Olympic football (soccer) team was such a shining example of the liberation of Iraq i read an interview with one of the members saying if he wasn't at the Olympics, he'd be home fighting in Fallujah (this is when that was the current hotspot) - against the Americans. so much for being greeted as liberators. and this is hardly an entirely foreign concept: i think Bush is probably our worst president ever and is actively undermining the foundations upon which the Republic is built, but if the Chinese decided we needed a regime change, invaded and occupied us, i don't have to squint very hard to see myself out in the streets fighting them off. because it's not their country, and they should get the hell out. occupation and military action breeds resentment and hostility, universally.
no, it's not controversial to point out that different religions are different. but you're ignoring history. i agree, Jesus' portrayal in Christian texts is much better (by any secular or objective standards) than Mohammed's portrayal in Islamic texts. but you need to keep in mind that every major Christian denomination claims the old testament as part of their Bible. that means Christians don't get to just point to Jesus and say "see? all good!". we also have to own up to David, a military conquerer with... "significant moral blemishes." we have to own up to Abraham's marriage practices. Noah's drunkenness and the resulting incest (and these were the best folks God could find!). i could believe that, on some percentage basis, the Koran is more violent than the Old Testament (i don't believe that at current, based on my reading, but i've only read the Koran once, don't own one to compare, and probably wouldn't bother with the lengthy textual comparison if i did (i'm not that bored). the point is i could be convinced). my point is that the differences are, at best, one of degree, and not a huge degree, either.