Customer sets up a contract, with semi-sane requirements and sane deadline.
Contract is approved, but requirements change, deviating increasingly far from sane.
There are punitive charges for not meeting the deadline. The developer is simply unable to meet all the requirements of the contract in full, on time.
Solution: As deadline comes, wrap up and release the half-made, definitely not ready for market product that "technically" meets all the requirements "on paper" - everything works, but nearly nothing works correctly. Then finish development while calling it "maintenance". Deadline is met, requirements will be met *eventually*, no punitive charges, essentially the best outcome possible in given situation.
Yes. I can already predict the optimal solution the AI finds:
Build one "stock" airplane, with control systems linked to the AI for remote control. Arm it with a full loadout of air to ground missiles. Bomb the hell out of all the decision-makers until they come up with sane specs, keep repeating as necessary.
Only with uniform prices. Say, revenue of McDonalds from Russia doesn't reflect their popularity there, because if they charged the same prices as in USA, they'd be without customers.
Hey... it looks like you nailed the point behind the success of BASIC!
It was flat as a table. All variables were globals; its loops, subroutines and conditionals behaved like a GOTO with a little extra flavor added, essentially no hierarchy whatsoever, just sequential execution line by line with jumps here and there. And beginners loved it, while experts lamented the "negative learning".
Facilitating copyright infringement. Specifically, providing means to perform copyright infringement in a way that makes finding the infringers exceptionally difficult.
Boils down to accessory to a crime.
He took sufficient precautions to create a solid plausible deniability defense - solid enough that the case against him simply cannot be won. Everyone knows he's guilty, but he made damn sure there would be no solid proof of that - there's only a glaringly specific set of circumstances surrounding the inexplicable absence or invalidity of proofs. It's just like everyone *knows* Al Capone was a gangster, but to this day there are no legally valid proofs of that, despite nobody disputing that fact.
That's in silicon, so no easy hotfix for existing CPUs, but that would be essentially what's needed to protect new releases. Would work on Meltdown no problem.
Spectre would be harder to overcome as it utilizes timing mechanisms - essentially a calculation that is already cached ends way faster than a brand new one, which allows to guess results of prior calculations. The attack is extremely slow though, taking literally many hours of operation at full CPI load to read a significant part of memory - and it's read-only; can leak keys/hashes but can't directly escalate privileges or execute malicious code. I guess detecting the exploit program's activity and disabling it before it obtains anything of value would be easier than protecting the CPU (which currently essentially depends on flushing the cache before switching to protected content, so that no successful cache hits are there to be found.)
The basic problem with the Spectre is not branch predictor and speculation, but weak hashes used to identify cache entries, that allow for purposeful collision.
A speculative execution down a branch is canceled, but the results are cached; the cache entry having a certain identifier based on the branch address and some other details. Next time the program enters that branch, the results are there, free to pick from the cache, following the identifier.
The problem is the identifier is *not unique*. A specifically engineered branch entered from a specific address at a completely different location - say, userspace - will create a hash collision and the CPU will retrieve exactly the same results - even if it has no privileges to access them. In normal use this collision is pretty much impossible to occur accidentally, but properly engineered piece of code can mimic a privileged piece of code and get - or inject - speculative execution results.
Had the caching been based on hashes that are sized like the entire machine word of address, instead of a special selection of bits in a weirdo self-xored shift register, the problem wouldn't exist. It was a shortcut - a simplification of the predictor and cache architecture that instead of solidly, uniquely identifying an optimized code segment, stored its weak, easy to spoof footprint to find it for reuse.
> And by supporting that racist, misogynistic piece of crap they are contributing to a culture in this country that is racist and misogynistic.
Or maybe they just researched the white gloves crimes of Clinton, or got most blatantly, rudely cheated out of the chance to vote Sanders and they voted *the other candidate*?
Lots of Trump voters voted Trump only because they were sick and tired of empowering the Clintion mafia, and of being brainwashed that this is about genders, minorities, racial issues - while it was all about MONEY.
And the defenders of Clinton still keep on playing the whole gender/race red herring card. No, it's about the Clinton dynasty losing the crown to some upstart with his own business connections, disrupting the flow of money to Clintons and their secret supporters.
How "minute" a fringe are we talking about? 0.1%? 1%? 5%? I'd lean towards the latter; the fringe is vocal and influential. In a city of three hundred thousand, say 1/3 identifies as democrats; a riot of more than a thousand participants erupts, that's easily over 1%, They are dangerous people and the numbers are such that local police force can get overwhelmed... the threshold where random destruction of property turns into a civil war is really closer than the "just a fringe" apologists are willing to admit.
Oh, but that cost of labor somehow miraculously happened all by itself? Or some nefarious entity engineered it into the innocent Apple's products (and only Apple products!) with no Apple's involvement?
I've engineered devices which would need to be serviced later. I'd lose my job if I made them such that a routine maintenance procedure like battery replacement took more than five minutes to perform, and more than ten minutes to train someone to do it.
I bet if I lost my job that way, though, Apple would welcome me with open arms and request me to develop a device as complex and difficult to repair as only possible.
Somehow none of these "miracles" happen to any other brands... Only Apple. How is it that every company in the world can get this right, only where it comes to Apple suddenly you're faced with insurmountable mountain of problems?
Unless you've been smoking near your iPhone. Or farted. Or looked at it wrong. Apple is extremely happy to disclaim any responsibility for repairs. And the reason why the procedure is difficult and labor-intensive = expensive? Because Apple deliberately made it so. Other companies somehow can keep battery replacement easy.
Only because Apple made the procedure hard in the first place.
Replacement batteries for other phones are often $5, and easy enough to replace that there's no chance to damage the phone.
"These are not bullshit excuses, these are genuine problems!" is an old Apple bullshit mantra - these are "genuine problems" which Apple deliberately chose to create artificially in the first place.
But you can't fail to pull the lever, and the outcome is strictly random (or otherwise tweaked in a way beyond your control).
Make this a genuine *skill* challenge, with clear goal, failure mode, and actually challenging (not a mere formality). Of course losing deprives you of the winnings. Set the difficulty in such a way that, say, 50% players can finish it, and top 30-40% can finish it consistently, every time, while bottom 30-40% fail it consistently.
yandere dating simulator game, active all the time, monitoring your phone activity and sending you creepy messages from your virtual psycho girlfriend after you call a contact named with a female name, or it overhears a female voice over the microphone.
Water in concentrations of ~0.5% is present in Martian soil; it's also heavily saturated with various salts, primarily perchlorates, so obtaining and purifying it is a serious scientific and technological challenge - but not unsurmountable one. I guess a machine that strip-mines Mars for water would help; the base would recirculate all water it can retrieve, so only "leakage" losses and expansion needs would need to be supplied from the soil.
Nitpicking: they had been flown, but in a blunt-nosed configuration.
Customer sets up a contract, with semi-sane requirements and sane deadline.
Contract is approved, but requirements change, deviating increasingly far from sane.
There are punitive charges for not meeting the deadline. The developer is simply unable to meet all the requirements of the contract in full, on time.
Solution: As deadline comes, wrap up and release the half-made, definitely not ready for market product that "technically" meets all the requirements "on paper" - everything works, but nearly nothing works correctly. Then finish development while calling it "maintenance". Deadline is met, requirements will be met *eventually*, no punitive charges, essentially the best outcome possible in given situation.
Yes. I can already predict the optimal solution the AI finds:
Build one "stock" airplane, with control systems linked to the AI for remote control. Arm it with a full loadout of air to ground missiles. Bomb the hell out of all the decision-makers until they come up with sane specs, keep repeating as necessary.
Only with uniform prices. Say, revenue of McDonalds from Russia doesn't reflect their popularity there, because if they charged the same prices as in USA, they'd be without customers.
Hey... it looks like you nailed the point behind the success of BASIC!
It was flat as a table. All variables were globals; its loops, subroutines and conditionals behaved like a GOTO with a little extra flavor added, essentially no hierarchy whatsoever, just sequential execution line by line with jumps here and there. And beginners loved it, while experts lamented the "negative learning".
Yes, of course, Big Al was just an innocent used furniture salesman, and he missed the date on his taxes.
Facilitating copyright infringement. Specifically, providing means to perform copyright infringement in a way that makes finding the infringers exceptionally difficult.
Boils down to accessory to a crime.
He took sufficient precautions to create a solid plausible deniability defense - solid enough that the case against him simply cannot be won. Everyone knows he's guilty, but he made damn sure there would be no solid proof of that - there's only a glaringly specific set of circumstances surrounding the inexplicable absence or invalidity of proofs. It's just like everyone *knows* Al Capone was a gangster, but to this day there are no legally valid proofs of that, despite nobody disputing that fact.
That's in silicon, so no easy hotfix for existing CPUs, but that would be essentially what's needed to protect new releases. Would work on Meltdown no problem.
Spectre would be harder to overcome as it utilizes timing mechanisms - essentially a calculation that is already cached ends way faster than a brand new one, which allows to guess results of prior calculations. The attack is extremely slow though, taking literally many hours of operation at full CPI load to read a significant part of memory - and it's read-only; can leak keys/hashes but can't directly escalate privileges or execute malicious code. I guess detecting the exploit program's activity and disabling it before it obtains anything of value would be easier than protecting the CPU (which currently essentially depends on flushing the cache before switching to protected content, so that no successful cache hits are there to be found.)
It's quite untrue.
The basic problem with the Spectre is not branch predictor and speculation, but weak hashes used to identify cache entries, that allow for purposeful collision.
A speculative execution down a branch is canceled, but the results are cached; the cache entry having a certain identifier based on the branch address and some other details. Next time the program enters that branch, the results are there, free to pick from the cache, following the identifier.
The problem is the identifier is *not unique*. A specifically engineered branch entered from a specific address at a completely different location - say, userspace - will create a hash collision and the CPU will retrieve exactly the same results - even if it has no privileges to access them. In normal use this collision is pretty much impossible to occur accidentally, but properly engineered piece of code can mimic a privileged piece of code and get - or inject - speculative execution results.
Had the caching been based on hashes that are sized like the entire machine word of address, instead of a special selection of bits in a weirdo self-xored shift register, the problem wouldn't exist. It was a shortcut - a simplification of the predictor and cache architecture that instead of solidly, uniquely identifying an optimized code segment, stored its weak, easy to spoof footprint to find it for reuse.
What? Never heard of angular temperature? 90 degrees Kelvin is either right or hot, while 0 degrees Kelvin is cold ahead.
> And by supporting that racist, misogynistic piece of crap they are contributing to a culture in this country that is racist and misogynistic.
Or maybe they just researched the white gloves crimes of Clinton, or got most blatantly, rudely cheated out of the chance to vote Sanders and they voted *the other candidate*?
Lots of Trump voters voted Trump only because they were sick and tired of empowering the Clintion mafia, and of being brainwashed that this is about genders, minorities, racial issues - while it was all about MONEY.
And the defenders of Clinton still keep on playing the whole gender/race red herring card. No, it's about the Clinton dynasty losing the crown to some upstart with his own business connections, disrupting the flow of money to Clintons and their secret supporters.
Because you say so? Usually civil wars start with civil unrest, public protests escalating in violence.
The problem is in exact numbers.
How "minute" a fringe are we talking about? 0.1%? 1%? 5%? I'd lean towards the latter; the fringe is vocal and influential. In a city of three hundred thousand, say 1/3 identifies as democrats; a riot of more than a thousand participants erupts, that's easily over 1%, They are dangerous people and the numbers are such that local police force can get overwhelmed... the threshold where random destruction of property turns into a civil war is really closer than the "just a fringe" apologists are willing to admit.
Only after a world-wide scandal. Formerly, you'd have it replaced for $80.
Oh, but that cost of labor somehow miraculously happened all by itself? Or some nefarious entity engineered it into the innocent Apple's products (and only Apple products!) with no Apple's involvement?
...and still end up charging some $15 more than in case of other brands.
I've engineered devices which would need to be serviced later. I'd lose my job if I made them such that a routine maintenance procedure like battery replacement took more than five minutes to perform, and more than ten minutes to train someone to do it.
I bet if I lost my job that way, though, Apple would welcome me with open arms and request me to develop a device as complex and difficult to repair as only possible.
Somehow none of these "miracles" happen to any other brands... Only Apple. How is it that every company in the world can get this right, only where it comes to Apple suddenly you're faced with insurmountable mountain of problems?
Unless you've been smoking near your iPhone. Or farted. Or looked at it wrong. Apple is extremely happy to disclaim any responsibility for repairs. And the reason why the procedure is difficult and labor-intensive = expensive? Because Apple deliberately made it so. Other companies somehow can keep battery replacement easy.
Only because Apple made the procedure hard in the first place.
Replacement batteries for other phones are often $5, and easy enough to replace that there's no chance to damage the phone.
"These are not bullshit excuses, these are genuine problems!" is an old Apple bullshit mantra - these are "genuine problems" which Apple deliberately chose to create artificially in the first place.
But you can't fail to pull the lever, and the outcome is strictly random (or otherwise tweaked in a way beyond your control).
Make this a genuine *skill* challenge, with clear goal, failure mode, and actually challenging (not a mere formality). Of course losing deprives you of the winnings. Set the difficulty in such a way that, say, 50% players can finish it, and top 30-40% can finish it consistently, every time, while bottom 30-40% fail it consistently.
yandere dating simulator game, active all the time, monitoring your phone activity and sending you creepy messages from your virtual psycho girlfriend after you call a contact named with a female name, or it overhears a female voice over the microphone.
Oh, but dear sir, it's not so simple! How sure are you the phone won't be 59% slower, or 61% slower?
What a move! Not only get the water problem solved, but also force a competitor in the space race to handle the heavy expenses of delivery!
Water in concentrations of ~0.5% is present in Martian soil; it's also heavily saturated with various salts, primarily perchlorates, so obtaining and purifying it is a serious scientific and technological challenge - but not unsurmountable one. I guess a machine that strip-mines Mars for water would help; the base would recirculate all water it can retrieve, so only "leakage" losses and expansion needs would need to be supplied from the soil.