Here's the trick, though: no flag-waving worker, no cones, no deviations from the known map beyond what's expected (any system like this must account for normal variations in surroundings), but the speed limit's still lower than normal for that stretch. No working equpiment at the moment means no broadcast signal, but that doesn't mean the posted limit doesn't apply. And no update to the database because nobody put it through, somebody dropped the ball. But the law still says that those signs are up, therefore the limit on the signs takes precedence over the database. And recognizing those signs reliably is not a trivial problem. A dirty sign that reads to the camera as a brown-and-yellow smudge is still a legal speed-limit sign.
I've seen the video. Now, I want to see that car drive on an overcast day with a lot of headlights creating a backdrop, reflections creating "lights" to compete with traffic lights, highly-patterned backgrounds that cause foreground objects to blend in, that kind of thing. Basically, take all the known weaknesses of the algorithms involved and design a scene specifically to break the algorithms, and prove to me that even in a deliberately hostile environment they keep working. Because there's a rule: "The absolute worst-case conditions you can imagine won't even come close to normal operating conditions.". I've got close to 3 decades designing, testing and breaking software. The Google videos look like the kind of smoke-tests I run to see if the basic logic's working. Once that's done, I start getting evil and trying to break the software, and that's where the real work begins. And all too often, it only takes me a few minutes to make the software crash. For instance, a traffic light where the lamps are as dim as is allowable, and half a block down a store has bright green lights out on a sidewalk overhang that're the same color as the traffic light's green lamp. Can the car pick out which green lamp's the traffic light and which is the store lamp?
How does it know what the speed limit is on a particular stretch of road? And what happens when the city changes the posted limit (eg.for construction work) and the car's database isn't updated? Since the car "knows" the speed limit is 55 there it's going to go 55 even though the posted limit is 25.
It should. The add-ons can be dumped into the folders, but the browser will leave them disabled and non-functioning until you manually enable them. At least until the adware makers start figuring out how to dig into the internals of the browser config files and modify things directly to convince the browser the add-ons have already been enabled. That's doable but not simple, so I expect it'll take a while for that to become common. And there's simple methods the browser can use to make that modification even more difficult, eg. tagging each enabled extension with an encrypted hash of the extension's file so that the adware would have to find the browser's encryption key before it could successfully modify the configuration.
Note that none of these will do anything about add-ons that convince the user to manually install them.
What I've seen done is a spammer gets an address book from one compromised account, and then proceeds to spam people in that address list forging the names of other people in the list on the assumption that if they know the holder of the compromised account they likely know each other too. The idea is to minimize the number of malicious e-mails appearing to come from any given source, so as to avoid getting noticed. And if you've received a malicious e-mail and clicked on a link in it, you have to assume your computer and your account are compromised too.
My analogy is a letter with my name and address written in the return-address space. Does that guarantee that the letter's from me? Of course not, anybody could write that in if they knew my address, and all it takes to find my address is to look me up in the phone book.
Long-established principle: while political speech may be protected, using my property for political speech against my wishes is not. You can get up on your soapbox all you want, but as a private citizen I don't have to let you use my front lawn to do it. Now, I'm paying for my cel phone and voice/data plan. That phone, and the minutes and messages on my plan, belong to me by virtue of my paying for them. You want to use up my voice minutes or my text messages from my cel-phone plan? Not without my permission. At the very minimum you shouldn't be allowed to use them up without replacing them at your own expense. I should have an unlimited-messages plan? Sorry, it's not your money and that's not your decision. You want me to have an unlimited-messages plan for your convenience, feel free to buy one for me and pick up the cost of it.
b) Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange;
Note the bolded text. The offer is good for any third party, not any third party possessing the binaries. And the offer is for source code only, not source code accompanied by binaries. According to the text of the GPL v2, if you are distributing under 3B then I do not need a copy of the binaries to avail myself of your offer to provide the source code.
I don't think so. "Third party" is, as you say, a specific legal term, but it is well-defined in law. The GPL doesn't provide any other definition, so the standard definition would be used in interpreting that language in the GPL.
And no, the GPL v3 doesn't allow distributing only to people who got the software directly from the distributor. Section 6 paragraph B (which applies here) says you have to provide the code to anyone who possesses the binaries. If Harry gets the program from you, and then Harry passes a copy along to me, I have a right to get the source code from you (Harry received the software under 6b, I received it under 6c, I possess the binaries thus can invoke 6b against you).
One thing's remained constant for a long time: the only option that permits you to give source code only to people who received the software directly from you is 3A (GPLv2) or 6A (GPLv3): distributing the source code along with the binaries.
A: Distribute the source code along with the binaries.
B: Make a written offer to distribute the source code to any third party.
C: Pass along the written offer you received for source code. Available only for non-commercial distribution where you received the software under option B.
If you're distributing the source along with the binaries you aren't obliged to make the source available independently at all (anyone who got a binary also got the source code, your obligations are fulfilled). But if you're distributing the binaries without source the GPL v2 doesn't permit you to only give source code to people who have the binaries, you have to give it to anyone who asks.
GPL v3 would allow you to only give source code to people who have the binaries. Beware, though, because it doesn't allow you to only give source code to people who got the binaries from you. If they got the binaries from someone who got them from someone who got them from someone who got them from you, you're still on the hook to provide source code.
The GPL doesn't mandate that the software/source code be released for free, it mandates that the source code MUST accompany the binary if (and only if) the binary is distributed.
Yes and no. Source code only has to be made available if the binaries are distributed, that much is correct. But the GPL v2 does not mandate that the source code must accompany the binaries. For commercial distribution, what it mandates (down in section 3, paragraphis A and B) is that either:
The source code must be distributed along with the binaries, or
You must offer the source code, at a cost of no more than the cost of your making and delivering the copy, to any third party who asks for it.
So yes, if you put up a port of DOSBox in the app store and don't include source code in the package, I can indeed come along and demand the source code from you without ever buying a copy of your app. If you refuse to provide it, you're in breach of your license to distribute the DOSBox code because you're failing to comply with section 3 regarding availability of the source code (no source in the binary package means paragraph A doesn't apply, your failure to make it available to any third party means paragraph B doesn't apply, and since you're distributing commercially through the app store paragraph C doesn't apply).
For myself, I'd be happy with a solid 20 megabit connection. What I really want is:
20 megabits upstream as well as downstream, so things like VoIP don't choke on multiple users and I can upload files to an external server without causing congestion.
A connection that stays 20 megabits instead of getting choked down to a fraction of that when a bunch of neighbors start using their connections heavily.
Which means gigabit or better to the local distribution point, and neighborhood infrastructure that can handle the aggregate bandwidth. Most of the problems I have aren't my individual connection's bandwidth, it's the shared local infrastructure between my home's connection point and the ISP that's insufficient for the bandwidth of all the connected subscribers. Fix that and give me symmetric bandwidth and I'll be a happy camper.
That's the basic problem I see. For the example given of an open() call, what happens when the file doesn't exist? It could be an error, resulting in the program throwing up a message and aborting the operation. Or it could be a perfectly normal and expected occurrence, resulting in the program continuing as if it'd opened the file, immediately hit an EOF and closed the file. So with the proposed mechanism, every time I see an operation I need to jump to the bottom of the file and find it's error block to see if it's actually an error or not. That breaks a basic rule for coding: don't make the reader jump all over the place to see all of what's happening.
Worse, the same file-not-found condition could be an error at one spot in a routine and a normal result at another. That means you not only need the error-block construct, you need an additional construct at the point of call to associate different error blocks with the different open() calls. And you haven't eliminated the jumping-all-over-the-place problem.
I'd note that in this case the review went beyond merely "I think they did bad work.". As noted here the contractor's allegations are of false statements of fact, eg. that the review stated that the contractor had invoiced for work not performed. That's not a matter of opinion, that's a statement of fact that's either true or false and if she did falsely claim she was invoiced for work nor performed when either she wasn't invoiced or the work was in fact performed (even if performed badly) it is legitimately actionable. The contractor may or may not be able to prove his case in court, but it's not the usual nebulous "they said bad things about me" lawsuit.
Before a state can force on-line merchants to collect sales tax on it's residents it must provide a way for merchants to inquire, in real time and without cost to the merchant, as to what the correct sales tax rate is for any customer from that state. If the merchant uses this service, the answer is to be considered authoritative and any errors are Not The Merchant's Problem.
The state has to provide a single place to send sales-tax payments to. The merchant may be required to detail the customer address and invoice amount for each transaction. Once payment is sent to this address, the merchant has no further obligations and it's the state's responsibility to distribute any portion of the sales tax due to other entities within the state.
IOW, if a state wants merchants to collect taxes for it, they have to take responsibility for telling merchants how much to collect and for actually collecting it. They don't get to fob the hard work off on the merchants.
We've tried several. The big problem is that the Treasury won't simply ditch the dollar bill in favor of coins. They try to issue the coins along with dollar bills, so of course people treat the coins as collectibles instead of currency and they never catch on.
The right way to do it is to just do it: issue the coins and stop issuing dollar bills. Bills would still be in circulation and accepted, but no new $1 bills would be issued to banks. If a bank orders $1s, it'd receive them in coins. Any dollar bills received by the Federal Reserve banks or the Treasury would be destroyed, and any replacement would be done with coins. I figure within a year we'd be on coins completely, most dollar bills would've been returned as worn and destroyed and with no new bills being issued coins would become so prevalent that they wouldn't be collectibles anymore.
Actually there are scientific facts. A really simple example would be in measuring gravity. If you drop a ball off a tower, the fact that it fell and the time it took to reach the ground are facts. The formula describing how it falls and the values of the various constants in it would be a theory, and how well the formula fits the observed facts is a measure of how good the theory is, but the facts still stand on their own independent of any theory.
Similarly with evolution. We can take critters with very short lifespans (which allows us to effectively compress time and see the effects of hundreds of generations in months) and observe that speciation does in fact occur. Evolution is a theory explaining how and why it occurs and giving us a framework for predicting when and how it'll occur outside the lab, but the fact that speciation did occur is just that: an observed fact. And one that can be repeatably observed, confirming that it's not a fluke.
If Issa's behind it, it's not being done because it benefits the public. My guess is this bill's proposed for the same kind of reasons California's Proposition 33 on auto insurance rates was proposed. That proposition would've changed the law to permit insurance companies to offer lower rates to drivers who'd had insurance for more than a certain length of time. Sounds good, right? Here's the catch: California law already requires insurance companies to offer best rates based only on driving record, regardless of whether you've had insurance before or not (yes, it's possible to have a driver's license and a driving record without owning a car of your own). So the proposed change wouldn't lower rates for people with a good driving record who'd had insurance for a while, but it'd permit insurance companies to raise rates for people with a good driving record who hadn't had insurance or had had a break in coverage. When all the major insurance companies are backing the proposal whole-heartedly you know it's not because it'll lower rates across the board, if they wanted to do that they could just do it. My guess is this proposal is to try and cut off legislation to do things like prohibit cableco/telco ISPs from giving their own VoIP services priority on the network while throttling traffic for Google Voice and/or Skype to make those competing services less attractive.
And when discussing the Three Laws, remember to read the stories first. Almost all of Asimov's Robots stories were about how the Three Laws broke down in practice. If one author can come up with so many interesting ways to break them, I'm not sanguine about the changes of anyone implementing them in any reasonable fashion.
Classic case: how does the robot react to the posited swerving-schoolbus scenario where all of it's available options involve violating the First Law? It can choose which humans it's going to injure, but not whether it'll injure one or more.
I'd point out a difference in scale, though. What you describe would be, in the system I work on, one single major class in one component, probably running in it's own thread. A single component would have anywhere from half-a-dozen to a couple dozen major classes handling the major aspects of it's operation. A change can involve code spanning 3-4 different components.
The code wasn't always like this. 10 years ago it was small and simple. But that was 10 years worth of additional features, contradictory design decisions and conflicting requirements ago. As time goes by I'm more and more convinced that an article in an IEEE journal had it right: major software has a lifespan of about 7 years, and if you want to be successful you have to plan on rewriting it every 5-7 years based on what you're learning as you maintain it in production.
The big problem I have with your statement is that often the problem with the software is the basic design. It's 10 years old, completely insufficient for what's being asked of it today, and has so many kludges and hacks bolted on that you can't do anything outside a tight set of constraints without irreparably breaking something else. Usually it gets to this state because management or the team leadership won't permit changes to the design (the usual justification is that there's too much risk and no business benefit right now). Once code's in that state, new code has to follow the old design to work and it's that old design that's adding complications and keeping new requirements from being cleanly implemented. You can't design new code, because the constraints imposed by the design it has to live inside are too tight. You can't implement new policies because they break existing code. You can't clean up code because every bit you try to clean up requires you to first clean up 8 other bits, and eventually you hit a stop-point in the recursion where cleaning up that level requires that the program design be changed, see above.
Yes, good code should never get to this state. But the problem cases aren't good code, that's why they're problem cases.
Make it more difficult to use reply-all. There's times when you really do need everyone on the original recipient list to see your response, but it should take extra steps to do that.
Enforce top-quoting and trimming of text. We have e-mail clients that can thread messages and maintain history. We don't need the entire thread in every single message. If the message text hasn't been trimmed, the e-mail client should insert an extra step to insure the user actually intended to quote the entire original message. Top-quoting helps enforce that by putting the cursor at the bottom of the message instead of at the top. That forces the user to either change locations or trim text to get their reply seen. If you're quoting so much of the message that your reply isn't visible from the top of the e-mail, you're probably quoting more than's needed.
Start using wiki and forum software to replace mailing lists. They offer notifications and alerts (so you know when people have responded to a thread you're involved in), RSS feeds (so you can conveniently see new and updated threads) and all the bells and whistles. They're not hard to set up. And best of all, they don't get purged (and the information lost) as mailboxes get cleaned up. Institutional memory goes a long way towards heading off future problems. It's a lot easier to deal with issues when someone can go "I think I remember something like this, lemme do a quick search... oh right, that's what it was, and we did X last time, let's try that.".
AdBlock Plus is fairly benign, if the advertisers stop to think about it. It permits exceptions, and allows users to temporarily unblock ads. If the advertisers run it out of business, I can always put into place alternatives to it. A DNS wildcard zone file that resolves all hosts in a domain to the router, a small HTTP/HTTPS server that returns "404 Not Found" for all requests, and an afternoon's work to cons up a script to convert a list of advertiser domains into named.conf syntax is all it takes to make their domains cease to exist completely as far as the local network's concerned. And it's fairly easy to add that to a flash image for most consumer routers, not much harder to add in an automatic update of the blackhole named.conf snippet from my server.
No, the average user won't be able to do any of that. They don't have to. They complain about ads to their techie friend, he offers to upgrade their router to eliminate the problem, from their point of view the problem ceases to be. And instead of one organization to deal with when it comes to ad blocking, now they have innumerable unidentified techies and modified routers and no way to argue their case with each individual consumer.
I think the advertisers ought to think about something: it's a lack of discretion that's brought things to this point. They couldn't be happy with only a few people blocking their ads, they had to get louder and more obnoxious in an attempt to force everybody to look at them and in the process convinced more and more people to use more and more effective blocking methods. So they want to continue getting ever more obnoxious? Not likely to do any better than it has to date. I'd liken it to the door-to-door salesman: when the homeowner says "Sorry, not interested." and starts to close the door, ramming your shoulder into the door and forcing it back open is not going to win you a second hearing and a sale. And if you take it any further in your attempts to force your way in and force the homeowner to listen to you, you run the risk of the homeowner calling the cops on you and really ruining your day.
Individual bits may, but to make use of that you have to separate those bits out from the kernel and distribute them as independent parts. When combined with the kernel, you have to distribute on terms that fit all of the kernel. So if the kernel as a whole is "GPLv2 only" and one driver is "GPLv2 or any later version", you can distribute just the driver under GPLv3 but when you distribute the whole kernel you have to follow GPLv2 because if you used GPLv3 you'd be violating the license to the rest of the kernel (it only permits GPLv2).
How about we simply test for impairment? It shouldn't matter why you're impaired, only whether you're impaired, right? So use the same roadside tests we currently use to determine whether a driver's impaired or not. Not the breathalyzer, the ones that test the driver's reactions and coordination. If the driver's impaired, he won't be able to pass the tests, If he can pass the tests, he's not impaired (or at least not impaired beyond what we otherwise allow on the roads). If he turns out to be impaired, you can sort out exactly why he's impaired after you've gotten him off the road. I'd think that tests for impairment should be a lot easier to come up with, assuming we even need to come up with them (I'd think that the roadside tests we used for decades before portable breathalyzers came along should do OK, otherwise why did we ever consider them valid?).
Note to police: the threat to the public is not driving while under the influence of alcohol or marijuana or the like. The threat is driving while sufficiently impaired to pose a danger to others. For myself I don't care if the other guy's reaction times were hosed because he was stoned, because he was drunk, because he was on prescription pain meds or because he was just too tired, the damage to others when he causes an accident because his reaction times are hosed is the same in all 4 cases.
Slashdot Mirror
Here's the trick, though: no flag-waving worker, no cones, no deviations from the known map beyond what's expected (any system like this must account for normal variations in surroundings), but the speed limit's still lower than normal for that stretch. No working equpiment at the moment means no broadcast signal, but that doesn't mean the posted limit doesn't apply. And no update to the database because nobody put it through, somebody dropped the ball. But the law still says that those signs are up, therefore the limit on the signs takes precedence over the database. And recognizing those signs reliably is not a trivial problem. A dirty sign that reads to the camera as a brown-and-yellow smudge is still a legal speed-limit sign.
I've seen the video. Now, I want to see that car drive on an overcast day with a lot of headlights creating a backdrop, reflections creating "lights" to compete with traffic lights, highly-patterned backgrounds that cause foreground objects to blend in, that kind of thing. Basically, take all the known weaknesses of the algorithms involved and design a scene specifically to break the algorithms, and prove to me that even in a deliberately hostile environment they keep working. Because there's a rule: "The absolute worst-case conditions you can imagine won't even come close to normal operating conditions.". I've got close to 3 decades designing, testing and breaking software. The Google videos look like the kind of smoke-tests I run to see if the basic logic's working. Once that's done, I start getting evil and trying to break the software, and that's where the real work begins. And all too often, it only takes me a few minutes to make the software crash. For instance, a traffic light where the lamps are as dim as is allowable, and half a block down a store has bright green lights out on a sidewalk overhang that're the same color as the traffic light's green lamp. Can the car pick out which green lamp's the traffic light and which is the store lamp?
How does it know what the speed limit is on a particular stretch of road? And what happens when the city changes the posted limit (eg.for construction work) and the car's database isn't updated? Since the car "knows" the speed limit is 55 there it's going to go 55 even though the posted limit is 25.
It should. The add-ons can be dumped into the folders, but the browser will leave them disabled and non-functioning until you manually enable them. At least until the adware makers start figuring out how to dig into the internals of the browser config files and modify things directly to convince the browser the add-ons have already been enabled. That's doable but not simple, so I expect it'll take a while for that to become common. And there's simple methods the browser can use to make that modification even more difficult, eg. tagging each enabled extension with an encrypted hash of the extension's file so that the adware would have to find the browser's encryption key before it could successfully modify the configuration.
Note that none of these will do anything about add-ons that convince the user to manually install them.
What I've seen done is a spammer gets an address book from one compromised account, and then proceeds to spam people in that address list forging the names of other people in the list on the assumption that if they know the holder of the compromised account they likely know each other too. The idea is to minimize the number of malicious e-mails appearing to come from any given source, so as to avoid getting noticed. And if you've received a malicious e-mail and clicked on a link in it, you have to assume your computer and your account are compromised too.
My analogy is a letter with my name and address written in the return-address space. Does that guarantee that the letter's from me? Of course not, anybody could write that in if they knew my address, and all it takes to find my address is to look me up in the phone book.
Long-established principle: while political speech may be protected, using my property for political speech against my wishes is not. You can get up on your soapbox all you want, but as a private citizen I don't have to let you use my front lawn to do it. Now, I'm paying for my cel phone and voice/data plan. That phone, and the minutes and messages on my plan, belong to me by virtue of my paying for them. You want to use up my voice minutes or my text messages from my cel-phone plan? Not without my permission. At the very minimum you shouldn't be allowed to use them up without replacing them at your own expense. I should have an unlimited-messages plan? Sorry, it's not your money and that's not your decision. You want me to have an unlimited-messages plan for your convenience, feel free to buy one for me and pick up the cost of it.
To quote from the GPL v2, section 3 paragraph B:
Note the bolded text. The offer is good for any third party, not any third party possessing the binaries. And the offer is for source code only, not source code accompanied by binaries. According to the text of the GPL v2, if you are distributing under 3B then I do not need a copy of the binaries to avail myself of your offer to provide the source code.
I don't think so. "Third party" is, as you say, a specific legal term, but it is well-defined in law. The GPL doesn't provide any other definition, so the standard definition would be used in interpreting that language in the GPL.
And no, the GPL v3 doesn't allow distributing only to people who got the software directly from the distributor. Section 6 paragraph B (which applies here) says you have to provide the code to anyone who possesses the binaries. If Harry gets the program from you, and then Harry passes a copy along to me, I have a right to get the source code from you (Harry received the software under 6b, I received it under 6c, I possess the binaries thus can invoke 6b against you).
One thing's remained constant for a long time: the only option that permits you to give source code only to people who received the software directly from you is 3A (GPLv2) or 6A (GPLv3): distributing the source code along with the binaries.
Apparently a common misconception, but a misconception nonetheless. If you read the GPL v2 section 3 on distribution in object-code form, you'll find you have 3 options:
If you're distributing the source along with the binaries you aren't obliged to make the source available independently at all (anyone who got a binary also got the source code, your obligations are fulfilled). But if you're distributing the binaries without source the GPL v2 doesn't permit you to only give source code to people who have the binaries, you have to give it to anyone who asks.
GPL v3 would allow you to only give source code to people who have the binaries. Beware, though, because it doesn't allow you to only give source code to people who got the binaries from you. If they got the binaries from someone who got them from someone who got them from someone who got them from you, you're still on the hook to provide source code.
Yes and no. Source code only has to be made available if the binaries are distributed, that much is correct. But the GPL v2 does not mandate that the source code must accompany the binaries. For commercial distribution, what it mandates (down in section 3, paragraphis A and B) is that either:
So yes, if you put up a port of DOSBox in the app store and don't include source code in the package, I can indeed come along and demand the source code from you without ever buying a copy of your app. If you refuse to provide it, you're in breach of your license to distribute the DOSBox code because you're failing to comply with section 3 regarding availability of the source code (no source in the binary package means paragraph A doesn't apply, your failure to make it available to any third party means paragraph B doesn't apply, and since you're distributing commercially through the app store paragraph C doesn't apply).
For myself, I'd be happy with a solid 20 megabit connection. What I really want is:
Which means gigabit or better to the local distribution point, and neighborhood infrastructure that can handle the aggregate bandwidth. Most of the problems I have aren't my individual connection's bandwidth, it's the shared local infrastructure between my home's connection point and the ISP that's insufficient for the bandwidth of all the connected subscribers. Fix that and give me symmetric bandwidth and I'll be a happy camper.
That's the basic problem I see. For the example given of an open() call, what happens when the file doesn't exist? It could be an error, resulting in the program throwing up a message and aborting the operation. Or it could be a perfectly normal and expected occurrence, resulting in the program continuing as if it'd opened the file, immediately hit an EOF and closed the file. So with the proposed mechanism, every time I see an operation I need to jump to the bottom of the file and find it's error block to see if it's actually an error or not. That breaks a basic rule for coding: don't make the reader jump all over the place to see all of what's happening.
Worse, the same file-not-found condition could be an error at one spot in a routine and a normal result at another. That means you not only need the error-block construct, you need an additional construct at the point of call to associate different error blocks with the different open() calls. And you haven't eliminated the jumping-all-over-the-place problem.
I'd note that in this case the review went beyond merely "I think they did bad work.". As noted here the contractor's allegations are of false statements of fact, eg. that the review stated that the contractor had invoiced for work not performed. That's not a matter of opinion, that's a statement of fact that's either true or false and if she did falsely claim she was invoiced for work nor performed when either she wasn't invoiced or the work was in fact performed (even if performed badly) it is legitimately actionable. The contractor may or may not be able to prove his case in court, but it's not the usual nebulous "they said bad things about me" lawsuit.
I'd be fine with this, with two conditions:
IOW, if a state wants merchants to collect taxes for it, they have to take responsibility for telling merchants how much to collect and for actually collecting it. They don't get to fob the hard work off on the merchants.
It depends. In California, that answer according to California Labor Code sections 2870-2872 the answer is a very definitive "No.".
We've tried several. The big problem is that the Treasury won't simply ditch the dollar bill in favor of coins. They try to issue the coins along with dollar bills, so of course people treat the coins as collectibles instead of currency and they never catch on.
The right way to do it is to just do it: issue the coins and stop issuing dollar bills. Bills would still be in circulation and accepted, but no new $1 bills would be issued to banks. If a bank orders $1s, it'd receive them in coins. Any dollar bills received by the Federal Reserve banks or the Treasury would be destroyed, and any replacement would be done with coins. I figure within a year we'd be on coins completely, most dollar bills would've been returned as worn and destroyed and with no new bills being issued coins would become so prevalent that they wouldn't be collectibles anymore.
Actually there are scientific facts. A really simple example would be in measuring gravity. If you drop a ball off a tower, the fact that it fell and the time it took to reach the ground are facts. The formula describing how it falls and the values of the various constants in it would be a theory, and how well the formula fits the observed facts is a measure of how good the theory is, but the facts still stand on their own independent of any theory.
Similarly with evolution. We can take critters with very short lifespans (which allows us to effectively compress time and see the effects of hundreds of generations in months) and observe that speciation does in fact occur. Evolution is a theory explaining how and why it occurs and giving us a framework for predicting when and how it'll occur outside the lab, but the fact that speciation did occur is just that: an observed fact. And one that can be repeatably observed, confirming that it's not a fluke.
If Issa's behind it, it's not being done because it benefits the public. My guess is this bill's proposed for the same kind of reasons California's Proposition 33 on auto insurance rates was proposed. That proposition would've changed the law to permit insurance companies to offer lower rates to drivers who'd had insurance for more than a certain length of time. Sounds good, right? Here's the catch: California law already requires insurance companies to offer best rates based only on driving record, regardless of whether you've had insurance before or not (yes, it's possible to have a driver's license and a driving record without owning a car of your own). So the proposed change wouldn't lower rates for people with a good driving record who'd had insurance for a while, but it'd permit insurance companies to raise rates for people with a good driving record who hadn't had insurance or had had a break in coverage. When all the major insurance companies are backing the proposal whole-heartedly you know it's not because it'll lower rates across the board, if they wanted to do that they could just do it. My guess is this proposal is to try and cut off legislation to do things like prohibit cableco/telco ISPs from giving their own VoIP services priority on the network while throttling traffic for Google Voice and/or Skype to make those competing services less attractive.
And when discussing the Three Laws, remember to read the stories first. Almost all of Asimov's Robots stories were about how the Three Laws broke down in practice. If one author can come up with so many interesting ways to break them, I'm not sanguine about the changes of anyone implementing them in any reasonable fashion.
Classic case: how does the robot react to the posited swerving-schoolbus scenario where all of it's available options involve violating the First Law? It can choose which humans it's going to injure, but not whether it'll injure one or more.
I'd point out a difference in scale, though. What you describe would be, in the system I work on, one single major class in one component, probably running in it's own thread. A single component would have anywhere from half-a-dozen to a couple dozen major classes handling the major aspects of it's operation. A change can involve code spanning 3-4 different components.
The code wasn't always like this. 10 years ago it was small and simple. But that was 10 years worth of additional features, contradictory design decisions and conflicting requirements ago. As time goes by I'm more and more convinced that an article in an IEEE journal had it right: major software has a lifespan of about 7 years, and if you want to be successful you have to plan on rewriting it every 5-7 years based on what you're learning as you maintain it in production.
The big problem I have with your statement is that often the problem with the software is the basic design. It's 10 years old, completely insufficient for what's being asked of it today, and has so many kludges and hacks bolted on that you can't do anything outside a tight set of constraints without irreparably breaking something else. Usually it gets to this state because management or the team leadership won't permit changes to the design (the usual justification is that there's too much risk and no business benefit right now). Once code's in that state, new code has to follow the old design to work and it's that old design that's adding complications and keeping new requirements from being cleanly implemented. You can't design new code, because the constraints imposed by the design it has to live inside are too tight. You can't implement new policies because they break existing code. You can't clean up code because every bit you try to clean up requires you to first clean up 8 other bits, and eventually you hit a stop-point in the recursion where cleaning up that level requires that the program design be changed, see above.
Yes, good code should never get to this state. But the problem cases aren't good code, that's why they're problem cases.
AdBlock Plus is fairly benign, if the advertisers stop to think about it. It permits exceptions, and allows users to temporarily unblock ads. If the advertisers run it out of business, I can always put into place alternatives to it. A DNS wildcard zone file that resolves all hosts in a domain to the router, a small HTTP/HTTPS server that returns "404 Not Found" for all requests, and an afternoon's work to cons up a script to convert a list of advertiser domains into named.conf syntax is all it takes to make their domains cease to exist completely as far as the local network's concerned. And it's fairly easy to add that to a flash image for most consumer routers, not much harder to add in an automatic update of the blackhole named.conf snippet from my server.
No, the average user won't be able to do any of that. They don't have to. They complain about ads to their techie friend, he offers to upgrade their router to eliminate the problem, from their point of view the problem ceases to be. And instead of one organization to deal with when it comes to ad blocking, now they have innumerable unidentified techies and modified routers and no way to argue their case with each individual consumer.
I think the advertisers ought to think about something: it's a lack of discretion that's brought things to this point. They couldn't be happy with only a few people blocking their ads, they had to get louder and more obnoxious in an attempt to force everybody to look at them and in the process convinced more and more people to use more and more effective blocking methods. So they want to continue getting ever more obnoxious? Not likely to do any better than it has to date. I'd liken it to the door-to-door salesman: when the homeowner says "Sorry, not interested." and starts to close the door, ramming your shoulder into the door and forcing it back open is not going to win you a second hearing and a sale. And if you take it any further in your attempts to force your way in and force the homeowner to listen to you, you run the risk of the homeowner calling the cops on you and really ruining your day.
Individual bits may, but to make use of that you have to separate those bits out from the kernel and distribute them as independent parts. When combined with the kernel, you have to distribute on terms that fit all of the kernel. So if the kernel as a whole is "GPLv2 only" and one driver is "GPLv2 or any later version", you can distribute just the driver under GPLv3 but when you distribute the whole kernel you have to follow GPLv2 because if you used GPLv3 you'd be violating the license to the rest of the kernel (it only permits GPLv2).
How about we simply test for impairment? It shouldn't matter why you're impaired, only whether you're impaired, right? So use the same roadside tests we currently use to determine whether a driver's impaired or not. Not the breathalyzer, the ones that test the driver's reactions and coordination. If the driver's impaired, he won't be able to pass the tests, If he can pass the tests, he's not impaired (or at least not impaired beyond what we otherwise allow on the roads). If he turns out to be impaired, you can sort out exactly why he's impaired after you've gotten him off the road. I'd think that tests for impairment should be a lot easier to come up with, assuming we even need to come up with them (I'd think that the roadside tests we used for decades before portable breathalyzers came along should do OK, otherwise why did we ever consider them valid?).
Note to police: the threat to the public is not driving while under the influence of alcohol or marijuana or the like. The threat is driving while sufficiently impaired to pose a danger to others. For myself I don't care if the other guy's reaction times were hosed because he was stoned, because he was drunk, because he was on prescription pain meds or because he was just too tired, the damage to others when he causes an accident because his reaction times are hosed is the same in all 4 cases.