While I have not the slightest doubt about the Chinese authorities' interests in surveillance, the ISA level seems a deeply strange layer to futz around with in order to further that goal. Are they planning on adding a MOVSCP(Move Word From String to Communist Party) instruction or something?
It seems that somebody looking to build bugs would be focusing on a mechanism a trifle higher-level: whatever 'TPM-but-don't-call-it-a-TPM-because-NIH-is-serious-business' standard they are plugging away at would be one logical place to look. Any 'National Operating System' type initiative would also be worth a look(though, realistically, retail spying on end user devices is kind of a pain in the ass, and vulnerable to discovery by hacker types, and you can just bug the telcos and ISPs instead, CALEA-or-nastier style, with much less fuss).
I'd be much more inclined to suspect some combination of quasi-mercantalist desire to avoid paying license fees to western tech outfits(and provide a convenient 'hook' by which foreign outfits who wish to score Chinese contracts can be forced to collaborate with whoever produces the blessed ISA) and a desire to (try) and prevent their government infrastructure being riddled with spots of code rotting on legacy architectures because some contractor who hasn't been in business in a decade had experience with it...
Unless the grandparent poster is making a wildly strong claim, it is unlikely that 'sets' as in 'axiomatic set theory' are intuitive; but that use of essentially setlike operators on comparatively small(and definitely finite) numbers of things is intuitive... Just stay away from some of the behaviorally-challenged edge cases.
It is my understanding that the 'medfield' platform is missing some legacy details(like a PCI bus, or something that looks like one) without which XP, and possibly later versions, simply won't run.
It's an x86; but it isn't really a member of the venerable 'IBM compatible' wintel family...
I know this isn't going to be a popular sentiment on/., but a Steam Linux client is going to please the Linux community for all of about 5 minutes. The applause won't even have died down before they're bitching that there aren't enough games, it's not open source, it doesn't look right in their obscure distro of choice, etc.
The Linux community *should* embrace and celebrate this, but my experience has been that a large (or at least largely vocal) part of that community is made up of idealists and professional bitchers who think everything should be open source and free. Introducing a closed source client that charges for games into that group isn't going to please them. Nothing is going to please them.
Okay, now everyone mod me troll for pointing out something you know is true.
I suspect that it won't be a major issue: Obviously, Free Software Only people aren't going to bite; but that is to be expected. Non-gamers won't care, also expected, and pretty much anybody gaming on Linux is already probably resigned to closed source binaries: their graphics drivers if nothing else(and presumably most of the games that they've coaxed into working under WINE(maybe there are a few OSS games with such strong Windows ties that WINE is easier than a port; but I'm having a hard time thinking of any). Intel OSS drivers are OK; but intel GPUs are not really gaming material. AMD is on the right trajectory; but the latest more-or-less-fully-ironed-out FOSS 3D support is for R200 parts, which aren't exactly screamers, and Nvidia's position on OSS drivers is "Well, it needs to be good enough so that the customer can see what they are doing as they download and install our binary driver."
I don't know how the numbers break down between purist users and nonpurist users; but the ratio of 'do-unto-others' purists to everybody else is tiny. Even the big, bad, Godfather of GNU himself merely advises that using closed software is not a good idea, and requests that you comply with the license of GPL software you use. Not terribly scary.
On a somewhat different topic, this linux release of theirs might have some ties to the persistent rumors of some sort of Valve-blessed hardware configuration providing a console-like package. If they suspect that they can even break even on Steam/Source for Linux, that might improve their prospects of being able to release a valvebox spec that leaves buyers with the extra $100 to spend on games, rather than on Windows. Even people who don't care about freedom care about free, after all.
For semi or permanently patient-connected devices, I'd really want to see some good, old-fashioned, physical interlocks where possible.
If, say, a device needs some sort of adjustment from time to time, it wouldn't be terribly taxing to have normally-open reed switch that physically disconnnects the external programming interface unless 'activated' by shoving a magnetic key into the programming slot. It doesn't stop a truly malicious actor, subtly planting malware to strike during planned program/sync periods; but it does mean that the entire world isn't fuzzing your security 24/7...
It would also be nice if (analogous to devices that have contact with patient tissue/fluids) any programmer/interface device used to manipulate implants and near-implants could be 'cleaned' by verifiably flashing a manufacturer approved memory state onto the unit before use.
The more complex(and, to look at history so far, probably not solvable) problem are the bigger devices, almost all built around embedded computers running obsolete OSes that you aren't allowed to change; but which you want to be able to communicate with other devices so as to swap diagnostic data, schedules, and who knows what else about. These are also the ones where massively expensive downtime, massive data breaches, or lingering infections requiring something little short of burning everything would be most likely to crop up.
Having somebody write an assembler virus for your specific brand of pacemaker and somehow sneak it in to your programming session is so much effort that it's practically a complement. The fact that millions of dollars of fancy medical gear will be good for little more than spewing the Klez worm the moment somebody's precious 'airgap' is compromised by a cleaner replugging a wire is simply a sad inevitability...
I see two major areas of concern with, arguably, quite different requirements:
1. Implants/embedded systems with some measure of field-programmability: On the plus side, these are much more likely to be running something fairly esoteric, possibly not even an OS at all, possibly some RTOS or embedded OS. They are also likely(for the moment) to have only short-range connection capabilities, quite possibly over a somewhat obscure protocol. This makes them low risk devices in terms of untargeted worm/phishing/etc. attacks, by virtue of limited connection and oddity of software. On the minus side, being directly connected to the patient, these offer a handy target for personally-directed sabotage, possibly from a surprising distance, depending on the whims of the RF gods(surely, the first person to reinact the classic 'sniper on the roof, suit with bodyguards crossing the parking lot toward the armored limo' scene; but with a rifle-stocked Yagi and lethal exploit code for the suit's pacemaker will be awarded a signed copy of every cyberpunk book of note).
2. Systems that have much more in common with the PLCs and management console computer systems that we are always complaining about in factory scenarios. That box running WinNT SP2 connected to a monstrously expensive diagnostic science machine, etc. etc. These are much more prosaic, just badly patched and outdated WinSomething boxes that really ought to be air-gapped properly, which makes them much more likely to suffer lots, and lots, and lots of expensive downtime when they eventually cave to the demand for electronic transmission of radiology data to another hospital for a consult and hook the sucker to the internet....
'Type 1' stuff seems like it would be best off with a "When in doubt, don't" approach: Don't interpret unsigned inputs, use very short range(inductive rather than RF, say) interfaces. It won't be perfect; but it'll at least confine the universe of potential hackers to people who could have just shived you anyway.
'Type 2' is where the mess really hits. Like industrial stuff, the economics of ripping out expensive capital investments are Deeply Unexciting; but persuading the vendor to deliver a service contract that doesn't read "Fuck you. Buy a Model N+1" is going to be a challenge. Also the (by no means necessarily false) promises of various 'telemedicine' applications are going to be constantly tugging at the people who run that stuff, urging them to connect it up. That isn't go to go well at all...
Indeed. And "email it to myself" has got to be the #2 backupstrategy(second only to childish hope) among the non-techies of the world.
My impression, at the time, was that Google was trying to, without stirring up a fuss, discourage the practice from becoming too common, too simple, or too widespread. A few geeks playing around with FUSE, or sending themselves zip chunks? Cost of doing business. Some $10 utility that 'turns your Gmail into a drive letter!" making the cover of PC Magazine? Not so good. So, they quietly pointed out that this was a ToS violation, that they could actually nuke your account if they wanted to, and largely left it at that. I don't think that anybody ever got nuked; but I never heard of anybody trying to commercialize backup software that made setting this up easy or selling some dodgy backup service using gmail accounts as a backend(don't laugh, it's no crazier than the 'hack unsecured corporated PBXes, sell cut-rate international phone cards' scam...)
That's why I started with "if you are interested in modest acceleration". Your rotation-for-pseudo-gravity needs can be conveniently handled by materials with good tensile strength(their stiffness sucks; but ropes, cords, cables, straps, and the like get truly impressive tensile strength per unit weight). Only if you fancy a trip into an atmosphere, or swift changes in velocity do you really need heavy structural durability...
It isn't clear that there will be a significant difference in editorial control:
With closed access journals, researcher submits paper hoping to improve his CV(sometimes even surrenders copyright). Peers in the field review paper, usually for free/status associated with being a reviewer for a serious journal, Journal turns around and sells the finished work back to the libraries.
Under the most common 'open' model, the costs of publishing are most commonly moved from the library end to the research end, by having a submission fee for papers that is one part of the cost of research, rather than having library/journal subscription fees as one part of the cost of research. That's the major economic change.
I'm sure that, in practice, the shakeups surrounding moves from one model to the other will sometimes be accompanied by moves toward tyrannical editorial control or toward broader transparency; but those will really be orthogonal to the funding model.
Harvard, with it's massive endowment, pretends that it cannot afford this? That is utter BS.
Harvard's Libraries say that they can't afford this(and given the relatively thin slice of the cash that the libraries see, is quite possibly true. Universities aren't unified entities. If anything, they are rather more compartmentalized than corporations(who, for accounting purposes if nothing else, have all sorts of internal distinctions of their own). The Endowment is practically an in-house investment fund, not a petty cash jar that the libraries can get access to easily. There are probably all manner of horribly complex distinctions(some largely accounting fictions, some fairly real(a professor in something biomed who maintains his lab and underlings on grants, say, is practically a tenant rather than an employee)).
Plus(in a happy confluence of self-interest and altruism) paying for these journals because they can afford it would be a losing choice for Harvard and just about all the other schools:
Why? Since Harvard is made of money, they really don't want to face the classic "How much does it cost?" "Well, how much do you have in your pockets?" chat with the sales rep. That's an express ticket to paying 50% more per year. Poorer schools don't want to end up paying 'industry standard' rates dictated by what richer schools can afford; but their faculty are also less likely to have the clout to just say "Dear Elsevier, fuck you." without damaging their careers.
Harvard has the cash and prestige to afford it(if they would just cut their libraries a slightly larger slice, from what my friends associated with the system tell me, the libraries are surprisingly starved given their reputation); but this also means that they have the best chance to draw a line and end the practice, instead...
If I remember, this is also why Google made some quiet, but threatening, noises about what they would do to anybody who made serious use of the cute little 'gmailFS' FUSE projects that are available to slap a filesystem-like structure on top of your Gmail storage space.
It isn't rocket-surgery that Gmail quotas are often largely underused, and the stuff that is used is rich with delicious keywords to be mined any monetized, while bulk file storage brings out the packrat in people, and frequently ends up containing big huge lumps of 'boring-and-probably-pirated-.iso-I-might-need-again' which aren't worth much to the marketdroids...
The problem goes deeper than that: Unless you go with encryption-on-client(ideally handled by a dedicated security processor, so that the key is never available to the potentially untrustworthy uploader-agent), which is comparatively rare because it breaks handy features like 'access from the web' and deduplication, the cloud storage provider gets to paw through your files by design.
Now, I do have to wonder why Google used a proprietary client(given their history of, for instance, OSSing their updater widget in order to calm people's fears about what it might be up to) when your data will be showing up on their servers in short order anyway, and file transfer over the internet isn't exactly an area of cutting-edge research.(Hi rysnc, how's it going?). One would think that an OSSed client would provide minimal competitive advantage to others, while helping to alleviate the 'our google overlords creep me out' response.
More generally, though, there really isn't a 'clientless'(ie. client is installed by default) option at present. The browser-based upload widgets are hacky as hell and often flake out on larger files, the java/activeX ones are incrementally more reliable but far more demanding and dodgy. FTP is horribly insecure and crotchety, SFTP causes barely a ripple outside a few geek circles. WebDAV seems to have gone nowhere for something like two decades now, some sort of NFS/SMB over VPN is ugly and wouldn't play nicely with many setups... A FUSE based FS would be nice for team linux; but arguably counts as a 'client' and doesn't help the majority of the market much...
I'd certainly trust an OSS client over a closed one; but it's hard to hold the need for a client of some kind against them at the moment.
Do you need heavy support structures if you are building in space?
They are certainly a convention in art and film; but if you are interested in modest acceleration(eg. solar sails, ion engines, and other stuff where fuel weight doesn't kill you) in a nearly total vacuum you enjoy considerable freedom to build massive structures out of toothpicks and mylar, with structural concerns only kicking in in pressurized sections of the craft or anything designed to re-enter a planetary atmosphere...
You wouldn't want to take it to the extremes commonly practiced by the software industry; but it wouldn't entirely surprise me if the 'ship now, fix later' model, within limits, actually holds up pretty well as long as only robots and expendable people(and no radioisotopic generators) are involved...
If space travel is going to be anything but a toy(outside of a few commercially viable niches for small satellites doing very valuable things in earth orbit), launch costs need to fall. If launch costs fall, the economics of being a raging perfectionist about every bit of gear you send up start looking less sensible compared to taking the risk of possibly having to send up a second one.
Concerns about space junk ruining valuable orbital slots will place a lower bound on how careless the world's nation states will let you be, and public opinion will be touchy about any accidents involving radioactive sources or high-profile humans burning up in the atmosphere; but nobody cares about the occasional satellite dropping dead or unmanned satellite launch blowing up...
Platinum has the advantage of being reasonably intrinsically useful(a brilliant catalyst for a variety of applications(certain fuel cell designs, for one), nice and corrosion resistant, in addition to being pretty and rare); but the price would certainly plummet if supply increased dramatically.
There are relatively few elements that are genuinely without practical applications(some of the shorter-lived radioactive ones are probably too hot to handle but fade too quickly to be useful industrial or medical emitters); but some get bumped into the status of 'financial instrument with a few esoteric applications' by their scarcity.
Cowards!!! Launch early, launch often. (Or just give the coders the honor of being test pilots. That will make those code monkeys program it real good the first time...)
It is possible that the researcher is somehow still a clueless bleeding-heart about math exams and drills, even after making it through a hard science curriculum; but I suspect that there is a much more sensible core to the idea(albeit one that can be achieved in large part just by waiting for the examinee to answer before posing the next question, rather than with the fancy apparatus)...
Especially for drills/practice, it is considered pedagogically wasteful to either waste a student's time on problems they can crush trivially or problems that they can only bang their head futilely against. The ideal is to keep the student at the edge of their ability, emphasizing areas where they are unacceptably weak; but not so weak as to be unlikely to benefit from the practice.
If the objective is a final evaluation of some sort, adjusting the difficulty makes no sense. However, if it is a practice exercise of some kind, continuous modification of difficulty according to understanding is a virtue.
It's all fun and games to laugh at the nannybot; but one should probably spare a moment's concern for the much-cherished illusion of 'agency', which is not done much good by any result that allows an individual's mental processes to be inferred before they've even become aware of them...
The researcher's suggestion for on-the-fly difficulty adjustment seems (if not overtly wrongheaded) a waste of scanner hardware, just waiting 20 seconds will give you the same data. The interesting bit is that actually making the mistake is apparently something you do fairly late in the game, compared to the occult mental activity...
Statistics like that are almost enough to make you believe that people had finally worked out the practical applications of 'statistics' and 'empiricism'.
This pleasant feeling only lasts until the next barrage of polling about the existence of guardian angels or horoscopes or whether coffee enemas cure cancer; but so it goes.
In all seriousness, this article manages to have a very important point(trust is an extremely valuable asset in a society, far cheaper and more pleasant than the alternatives of investing in lots and lots of contract lawyers and prisons); but its pessimism masks the counterpoint that loss of trust isn't exactly some sort of mental pathology. If anything, continued trust in the face of getting screwed over is pathological. It is important to distinguish the trust-loss scenarios where paranoia is the problem(eg. violent crime, for most of us. It's available 24/7, anything messy that happens worldwide; but actual levels are deeply unimpressive by historical standards) and trust-loss scenarios where the problem is that they really are out to get you(If you trust banks, I have a loss-proof CDO tranche to sell you)...
I'm not enough of a sociologist to say how much, as a matter of particular historical fact, was purely jealousy/grant competition; but there is a reasonably compelling strand of argument to be made in favor of the anti-physicist's view:
Namely, that the need to build very large apparatus suggested that physics had exhausted(whether by superior study or by a quirk of the laws of nature) the low-hanging fruits earlier than some other subjects, which put it further up the unpleasant slope of diminishing returns. If the biologists, say, are in the situation where they can barely send a grad student anywhere on earth without turning up a few dozen novel species; or the chemists have a variety of plans that involve nothing more exciting than a small-bore NMR apparatus and a few tens of thousands in reagents, it isn't entirely illogical to feel slighted by multi-billion dollar plans to conduct ever more subtle bug-hunts for subatomic particles.
We haven't been building steadily larger, more complex, and harder-to-get-funded physics machines just for giggles, we've been building them because the previous ones failed to smash particles hard enough.
It's not as though scientists like grantwriting and political money-grubbing to get their science widgets built. It's just that, so far, each generation of Big Apparatus has managed to peel back another layer of detail that bears no signs of being the deepest one. There is, after all, absolutely nothing that requires the laws of physics to be discernible with apparatus(or minds) of modest size. It is entirely possible that this isn't just the comparatively petty matter of deciding which politicians sign the checks; but of whether we will be able to declare victory within the limits of all the mass and energy within our reach. If it turns out that chasing elementary particles into their spider holes requires an accelerator that runs around the edge of the Kuiper belt, or a Pulsar caliber beam source, who do we get to complain to?
Unfortunately, if TFA is to be believed, the waffling was on the part of 'communications' flacks, not actual scientists. To be sure, scientists are unlikely to be enthusiastic about being misquoted(though, if they've done anything high profile before, probably view it as inevitable, no matter How Slowly And Loudly They Explain Their Work With Small Words...); but it isn't even clear that the email chain manages to involve any scientists, let alone giving them the final word on their research.
Slashdot Mirror
While I have not the slightest doubt about the Chinese authorities' interests in surveillance, the ISA level seems a deeply strange layer to futz around with in order to further that goal. Are they planning on adding a MOVSCP(Move Word From String to Communist Party) instruction or something?
It seems that somebody looking to build bugs would be focusing on a mechanism a trifle higher-level: whatever 'TPM-but-don't-call-it-a-TPM-because-NIH-is-serious-business' standard they are plugging away at would be one logical place to look. Any 'National Operating System' type initiative would also be worth a look(though, realistically, retail spying on end user devices is kind of a pain in the ass, and vulnerable to discovery by hacker types, and you can just bug the telcos and ISPs instead, CALEA-or-nastier style, with much less fuss).
I'd be much more inclined to suspect some combination of quasi-mercantalist desire to avoid paying license fees to western tech outfits(and provide a convenient 'hook' by which foreign outfits who wish to score Chinese contracts can be forced to collaborate with whoever produces the blessed ISA) and a desire to (try) and prevent their government infrastructure being riddled with spots of code rotting on legacy architectures because some contractor who hasn't been in business in a decade had experience with it...
Those poor TSA agents thought that it was the CIA's cocaine they were waving through. They were just doing their jobs.
Unless the grandparent poster is making a wildly strong claim, it is unlikely that 'sets' as in 'axiomatic set theory' are intuitive; but that use of essentially setlike operators on comparatively small(and definitely finite) numbers of things is intuitive... Just stay away from some of the behaviorally-challenged edge cases.
It is my understanding that the 'medfield' platform is missing some legacy details(like a PCI bus, or something that looks like one) without which XP, and possibly later versions, simply won't run.
It's an x86; but it isn't really a member of the venerable 'IBM compatible' wintel family...
I know this isn't going to be a popular sentiment on /., but a Steam Linux client is going to please the Linux community for all of about 5 minutes. The applause won't even have died down before they're bitching that there aren't enough games, it's not open source, it doesn't look right in their obscure distro of choice, etc.
The Linux community *should* embrace and celebrate this, but my experience has been that a large (or at least largely vocal) part of that community is made up of idealists and professional bitchers who think everything should be open source and free. Introducing a closed source client that charges for games into that group isn't going to please them. Nothing is going to please them.
Okay, now everyone mod me troll for pointing out something you know is true.
I suspect that it won't be a major issue: Obviously, Free Software Only people aren't going to bite; but that is to be expected. Non-gamers won't care, also expected, and pretty much anybody gaming on Linux is already probably resigned to closed source binaries: their graphics drivers if nothing else(and presumably most of the games that they've coaxed into working under WINE(maybe there are a few OSS games with such strong Windows ties that WINE is easier than a port; but I'm having a hard time thinking of any). Intel OSS drivers are OK; but intel GPUs are not really gaming material. AMD is on the right trajectory; but the latest more-or-less-fully-ironed-out FOSS 3D support is for R200 parts, which aren't exactly screamers, and Nvidia's position on OSS drivers is "Well, it needs to be good enough so that the customer can see what they are doing as they download and install our binary driver."
I don't know how the numbers break down between purist users and nonpurist users; but the ratio of 'do-unto-others' purists to everybody else is tiny. Even the big, bad, Godfather of GNU himself merely advises that using closed software is not a good idea, and requests that you comply with the license of GPL software you use. Not terribly scary.
On a somewhat different topic, this linux release of theirs might have some ties to the persistent rumors of some sort of Valve-blessed hardware configuration providing a console-like package. If they suspect that they can even break even on Steam/Source for Linux, that might improve their prospects of being able to release a valvebox spec that leaves buyers with the extra $100 to spend on games, rather than on Windows. Even people who don't care about freedom care about free, after all.
For semi or permanently patient-connected devices, I'd really want to see some good, old-fashioned, physical interlocks where possible.
If, say, a device needs some sort of adjustment from time to time, it wouldn't be terribly taxing to have normally-open reed switch that physically disconnnects the external programming interface unless 'activated' by shoving a magnetic key into the programming slot. It doesn't stop a truly malicious actor, subtly planting malware to strike during planned program/sync periods; but it does mean that the entire world isn't fuzzing your security 24/7...
It would also be nice if (analogous to devices that have contact with patient tissue/fluids) any programmer/interface device used to manipulate implants and near-implants could be 'cleaned' by verifiably flashing a manufacturer approved memory state onto the unit before use.
The more complex(and, to look at history so far, probably not solvable) problem are the bigger devices, almost all built around embedded computers running obsolete OSes that you aren't allowed to change; but which you want to be able to communicate with other devices so as to swap diagnostic data, schedules, and who knows what else about. These are also the ones where massively expensive downtime, massive data breaches, or lingering infections requiring something little short of burning everything would be most likely to crop up.
Having somebody write an assembler virus for your specific brand of pacemaker and somehow sneak it in to your programming session is so much effort that it's practically a complement. The fact that millions of dollars of fancy medical gear will be good for little more than spewing the Klez worm the moment somebody's precious 'airgap' is compromised by a cleaner replugging a wire is simply a sad inevitability...
I see two major areas of concern with, arguably, quite different requirements:
1. Implants/embedded systems with some measure of field-programmability: On the plus side, these are much more likely to be running something fairly esoteric, possibly not even an OS at all, possibly some RTOS or embedded OS. They are also likely(for the moment) to have only short-range connection capabilities, quite possibly over a somewhat obscure protocol. This makes them low risk devices in terms of untargeted worm/phishing/etc. attacks, by virtue of limited connection and oddity of software. On the minus side, being directly connected to the patient, these offer a handy target for personally-directed sabotage, possibly from a surprising distance, depending on the whims of the RF gods(surely, the first person to reinact the classic 'sniper on the roof, suit with bodyguards crossing the parking lot toward the armored limo' scene; but with a rifle-stocked Yagi and lethal exploit code for the suit's pacemaker will be awarded a signed copy of every cyberpunk book of note).
2. Systems that have much more in common with the PLCs and management console computer systems that we are always complaining about in factory scenarios. That box running WinNT SP2 connected to a monstrously expensive diagnostic science machine, etc. etc. These are much more prosaic, just badly patched and outdated WinSomething boxes that really ought to be air-gapped properly, which makes them much more likely to suffer lots, and lots, and lots of expensive downtime when they eventually cave to the demand for electronic transmission of radiology data to another hospital for a consult and hook the sucker to the internet....
'Type 1' stuff seems like it would be best off with a "When in doubt, don't" approach: Don't interpret unsigned inputs, use very short range(inductive rather than RF, say) interfaces. It won't be perfect; but it'll at least confine the universe of potential hackers to people who could have just shived you anyway.
'Type 2' is where the mess really hits. Like industrial stuff, the economics of ripping out expensive capital investments are Deeply Unexciting; but persuading the vendor to deliver a service contract that doesn't read "Fuck you. Buy a Model N+1" is going to be a challenge. Also the (by no means necessarily false) promises of various 'telemedicine' applications are going to be constantly tugging at the people who run that stuff, urging them to connect it up. That isn't go to go well at all...
Indeed. And "email it to myself" has got to be the #2 backupstrategy(second only to childish hope) among the non-techies of the world.
My impression, at the time, was that Google was trying to, without stirring up a fuss, discourage the practice from becoming too common, too simple, or too widespread. A few geeks playing around with FUSE, or sending themselves zip chunks? Cost of doing business. Some $10 utility that 'turns your Gmail into a drive letter!" making the cover of PC Magazine? Not so good. So, they quietly pointed out that this was a ToS violation, that they could actually nuke your account if they wanted to, and largely left it at that. I don't think that anybody ever got nuked; but I never heard of anybody trying to commercialize backup software that made setting this up easy or selling some dodgy backup service using gmail accounts as a backend(don't laugh, it's no crazier than the 'hack unsecured corporated PBXes, sell cut-rate international phone cards' scam...)
That's why I started with "if you are interested in modest acceleration". Your rotation-for-pseudo-gravity needs can be conveniently handled by materials with good tensile strength(their stiffness sucks; but ropes, cords, cables, straps, and the like get truly impressive tensile strength per unit weight). Only if you fancy a trip into an atmosphere, or swift changes in velocity do you really need heavy structural durability...
It isn't clear that there will be a significant difference in editorial control:
With closed access journals, researcher submits paper hoping to improve his CV(sometimes even surrenders copyright). Peers in the field review paper, usually for free/status associated with being a reviewer for a serious journal, Journal turns around and sells the finished work back to the libraries.
Under the most common 'open' model, the costs of publishing are most commonly moved from the library end to the research end, by having a submission fee for papers that is one part of the cost of research, rather than having library/journal subscription fees as one part of the cost of research. That's the major economic change.
I'm sure that, in practice, the shakeups surrounding moves from one model to the other will sometimes be accompanied by moves toward tyrannical editorial control or toward broader transparency; but those will really be orthogonal to the funding model.
Harvard, with it's massive endowment, pretends that it cannot afford this? That is utter BS.
Harvard's Libraries say that they can't afford this(and given the relatively thin slice of the cash that the libraries see, is quite possibly true. Universities aren't unified entities. If anything, they are rather more compartmentalized than corporations(who, for accounting purposes if nothing else, have all sorts of internal distinctions of their own). The Endowment is practically an in-house investment fund, not a petty cash jar that the libraries can get access to easily. There are probably all manner of horribly complex distinctions(some largely accounting fictions, some fairly real(a professor in something biomed who maintains his lab and underlings on grants, say, is practically a tenant rather than an employee)).
Plus(in a happy confluence of self-interest and altruism) paying for these journals because they can afford it would be a losing choice for Harvard and just about all the other schools:
Why? Since Harvard is made of money, they really don't want to face the classic "How much does it cost?" "Well, how much do you have in your pockets?" chat with the sales rep. That's an express ticket to paying 50% more per year. Poorer schools don't want to end up paying 'industry standard' rates dictated by what richer schools can afford; but their faculty are also less likely to have the clout to just say "Dear Elsevier, fuck you." without damaging their careers.
Harvard has the cash and prestige to afford it(if they would just cut their libraries a slightly larger slice, from what my friends associated with the system tell me, the libraries are surprisingly starved given their reputation); but this also means that they have the best chance to draw a line and end the practice, instead...
If I remember, this is also why Google made some quiet, but threatening, noises about what they would do to anybody who made serious use of the cute little 'gmailFS' FUSE projects that are available to slap a filesystem-like structure on top of your Gmail storage space.
It isn't rocket-surgery that Gmail quotas are often largely underused, and the stuff that is used is rich with delicious keywords to be mined any monetized, while bulk file storage brings out the packrat in people, and frequently ends up containing big huge lumps of 'boring-and-probably-pirated-.iso-I-might-need-again' which aren't worth much to the marketdroids...
The problem goes deeper than that: Unless you go with encryption-on-client(ideally handled by a dedicated security processor, so that the key is never available to the potentially untrustworthy uploader-agent), which is comparatively rare because it breaks handy features like 'access from the web' and deduplication, the cloud storage provider gets to paw through your files by design.
Now, I do have to wonder why Google used a proprietary client(given their history of, for instance, OSSing their updater widget in order to calm people's fears about what it might be up to) when your data will be showing up on their servers in short order anyway, and file transfer over the internet isn't exactly an area of cutting-edge research.(Hi rysnc, how's it going?). One would think that an OSSed client would provide minimal competitive advantage to others, while helping to alleviate the 'our google overlords creep me out' response.
More generally, though, there really isn't a 'clientless'(ie. client is installed by default) option at present. The browser-based upload widgets are hacky as hell and often flake out on larger files, the java/activeX ones are incrementally more reliable but far more demanding and dodgy. FTP is horribly insecure and crotchety, SFTP causes barely a ripple outside a few geek circles. WebDAV seems to have gone nowhere for something like two decades now, some sort of NFS/SMB over VPN is ugly and wouldn't play nicely with many setups... A FUSE based FS would be nice for team linux; but arguably counts as a 'client' and doesn't help the majority of the market much...
I'd certainly trust an OSS client over a closed one; but it's hard to hold the need for a client of some kind against them at the moment.
Do you need heavy support structures if you are building in space?
They are certainly a convention in art and film; but if you are interested in modest acceleration(eg. solar sails, ion engines, and other stuff where fuel weight doesn't kill you) in a nearly total vacuum you enjoy considerable freedom to build massive structures out of toothpicks and mylar, with structural concerns only kicking in in pressurized sections of the craft or anything designed to re-enter a planetary atmosphere...
You wouldn't want to take it to the extremes commonly practiced by the software industry; but it wouldn't entirely surprise me if the 'ship now, fix later' model, within limits, actually holds up pretty well as long as only robots and expendable people(and no radioisotopic generators) are involved...
If space travel is going to be anything but a toy(outside of a few commercially viable niches for small satellites doing very valuable things in earth orbit), launch costs need to fall. If launch costs fall, the economics of being a raging perfectionist about every bit of gear you send up start looking less sensible compared to taking the risk of possibly having to send up a second one.
Concerns about space junk ruining valuable orbital slots will place a lower bound on how careless the world's nation states will let you be, and public opinion will be touchy about any accidents involving radioactive sources or high-profile humans burning up in the atmosphere; but nobody cares about the occasional satellite dropping dead or unmanned satellite launch blowing up...
Platinum has the advantage of being reasonably intrinsically useful(a brilliant catalyst for a variety of applications(certain fuel cell designs, for one), nice and corrosion resistant, in addition to being pretty and rare); but the price would certainly plummet if supply increased dramatically.
There are relatively few elements that are genuinely without practical applications(some of the shorter-lived radioactive ones are probably too hot to handle but fade too quickly to be useful industrial or medical emitters); but some get bumped into the status of 'financial instrument with a few esoteric applications' by their scarcity.
I would suggest adding a watchdog to the system to handle that, but just ask Laika how well that works...
Cowards!!! Launch early, launch often. (Or just give the coders the honor of being test pilots. That will make those code monkeys program it real good the first time...)
It is possible that the researcher is somehow still a clueless bleeding-heart about math exams and drills, even after making it through a hard science curriculum; but I suspect that there is a much more sensible core to the idea(albeit one that can be achieved in large part just by waiting for the examinee to answer before posing the next question, rather than with the fancy apparatus)...
Especially for drills/practice, it is considered pedagogically wasteful to either waste a student's time on problems they can crush trivially or problems that they can only bang their head futilely against. The ideal is to keep the student at the edge of their ability, emphasizing areas where they are unacceptably weak; but not so weak as to be unlikely to benefit from the practice.
If the objective is a final evaluation of some sort, adjusting the difficulty makes no sense. However, if it is a practice exercise of some kind, continuous modification of difficulty according to understanding is a virtue.
It's all fun and games to laugh at the nannybot; but one should probably spare a moment's concern for the much-cherished illusion of 'agency', which is not done much good by any result that allows an individual's mental processes to be inferred before they've even become aware of them...
The researcher's suggestion for on-the-fly difficulty adjustment seems (if not overtly wrongheaded) a waste of scanner hardware, just waiting 20 seconds will give you the same data. The interesting bit is that actually making the mistake is apparently something you do fairly late in the game, compared to the occult mental activity...
Statistics like that are almost enough to make you believe that people had finally worked out the practical applications of 'statistics' and 'empiricism'.
This pleasant feeling only lasts until the next barrage of polling about the existence of guardian angels or horoscopes or whether coffee enemas cure cancer; but so it goes.
In all seriousness, this article manages to have a very important point(trust is an extremely valuable asset in a society, far cheaper and more pleasant than the alternatives of investing in lots and lots of contract lawyers and prisons); but its pessimism masks the counterpoint that loss of trust isn't exactly some sort of mental pathology. If anything, continued trust in the face of getting screwed over is pathological. It is important to distinguish the trust-loss scenarios where paranoia is the problem(eg. violent crime, for most of us. It's available 24/7, anything messy that happens worldwide; but actual levels are deeply unimpressive by historical standards) and trust-loss scenarios where the problem is that they really are out to get you(If you trust banks, I have a loss-proof CDO tranche to sell you)...
I'm not enough of a sociologist to say how much, as a matter of particular historical fact, was purely jealousy/grant competition; but there is a reasonably compelling strand of argument to be made in favor of the anti-physicist's view:
Namely, that the need to build very large apparatus suggested that physics had exhausted(whether by superior study or by a quirk of the laws of nature) the low-hanging fruits earlier than some other subjects, which put it further up the unpleasant slope of diminishing returns. If the biologists, say, are in the situation where they can barely send a grad student anywhere on earth without turning up a few dozen novel species; or the chemists have a variety of plans that involve nothing more exciting than a small-bore NMR apparatus and a few tens of thousands in reagents, it isn't entirely illogical to feel slighted by multi-billion dollar plans to conduct ever more subtle bug-hunts for subatomic particles.
Is this really a political problem?
We haven't been building steadily larger, more complex, and harder-to-get-funded physics machines just for giggles, we've been building them because the previous ones failed to smash particles hard enough.
It's not as though scientists like grantwriting and political money-grubbing to get their science widgets built. It's just that, so far, each generation of Big Apparatus has managed to peel back another layer of detail that bears no signs of being the deepest one. There is, after all, absolutely nothing that requires the laws of physics to be discernible with apparatus(or minds) of modest size. It is entirely possible that this isn't just the comparatively petty matter of deciding which politicians sign the checks; but of whether we will be able to declare victory within the limits of all the mass and energy within our reach. If it turns out that chasing elementary particles into their spider holes requires an accelerator that runs around the edge of the Kuiper belt, or a Pulsar caliber beam source, who do we get to complain to?
Unfortunately, if TFA is to be believed, the waffling was on the part of 'communications' flacks, not actual scientists. To be sure, scientists are unlikely to be enthusiastic about being misquoted(though, if they've done anything high profile before, probably view it as inevitable, no matter How Slowly And Loudly They Explain Their Work With Small Words...); but it isn't even clear that the email chain manages to involve any scientists, let alone giving them the final word on their research.
It's the 'Harper Regime', I'm afraid.