So, first, I'm a big fan of Mercurial, and our team just switched to it. The "gotchas" that I can enumerate are:
"I have to type TWO COMMANDS to get my bits onto the server???!!!!?!
"I have to type TWO COMMANDS to get new bits from the server???!?!?!?! (use "hg pull -u")
It's not quite as easy to secure through Apache -- if, say, you are using Trac for project management, that was what kept us using Svn for a while.
I'm not saying "impossible", I'm saying "not as easy", especially in an open source project.
The first two items, people eventually figure out how wonderful it is to be able to do local commits, without having to explicitly create/name/manage branches.
The third item, not an issue if you use ssh for distributed authentication to servers, which you surely will.
And learn to use the mute button, and be sure that you get a good quality connection, and use good microphones. CLI_I_G and DR__OUTS are really hard to stomach for a long meeting.
Yes, it's worth noting just how large automobile engines are nowadays. I learned to drive in a 2000lb car with 42bhp, and a top speed of about 75mph. Amusingly enough, if you take 100/75 (1.33), and cube it (2.35), times 42, gives 98.8 (the power required to overcome wind friction is cubic in speed). So my youthful experiments, are entirely consistent with your assertion. The Prius might do a little better than that, because of improvements in aerodynamics and tire design (the old Saabs were pretty good, but I think we do better now).
However, hill-climbing, is another matter entirely.
I have no trouble reporting bugs. If there's any obstacles along the way to reporting the bug, I bail, and simply post them on my blog. With snarky commentary, of course, because by then I'm in a bad mood.
Best of all, is when random strangers with the same bug, find my blogged bug report instead of the bug database, and then start posting comments on it.
Years ago, a friend of mine showed me the "killer app for Python", and it was the stack traces. When their (Python-based) product crashed in the field, its driver would bundle it all up in a zip file, and ask the user for permission to forward it to the developers, with the usual reassurances that all data discovered in the process would be treated as confidential. Bug report arrives, the stack trace (with variables etc, it's Python, right?) contains all the necessary info, the next interaction with the customer, is to tell them where to download the bug-fixed version of the product. "No time wasted on annoying human interaction, customer thinks we are attentive geniuses, everyone productive and happy."
My suspicion is, the robot would just go slower. Humans are famous for overestimating their luck and skill. It's also possible that the robot would do a better job even with everything equal -- if nothing else, it is ever-vigilant, and if has the skills to avoid 100% of pedestrian collisions, even assuming pedestrians "at fault", it can probably spot a moose, too. Remember, we don't see in IR.
And maybe travel just takes slightly longer, but you're not busy driving, so you can do something else with the time.
You assume that the robot would be stupid enough to travel at 100km/hr in a moose-crossing zone. Avoid the dangerous situation in the first place, you avoid the need to be able to handle it.
What they didn't tell you, was how smart the car actually is. When you program in a crappy little 2 mile trip, the car will refuse to me, and tell you to take a bicycle instead. The resulting regular exercise and improved cardiovascular health will save far more lives that merely avoiding crashes. (The part about the smart-ass car is a joke. The relative risk calculation, crash vs diseases of the unfit, is not.)
Presumably, the car will be black-boxed out the kazoo. If the robots are better than not just the average human, but better than the average careful human, then insurance companies will jump on it. They play the numbers game, not the "but what if it was ME????" game. If the robots are safer in aggregate, then they will be cheaper to insure, and the black boxes (which will cost little, compared to everything else) will generally provide a handy defense in court.
Pardon me, but you are misinformed, and have a terrible view of other people.
FIRST, WTF did you think the individual mandate was for, if not to prevent people from buying insurance only after they get sick? It's the quid-pro-quo of forbidding the insurance companies from dicking you around with pre-existing technicalities and requiring that they take-all-customers. (Would be more efficient to cut out the middlemen, but cost and quality were not as important as insurance industry campaign funding.)
SECOND, people DO care about their health, but they tend to discount the future horribly and believe that because of their all-encompassing above-averageness (the Lake Wobegon Effect), bad stuff will not happen to them. People do care about their health, they just don't do a very good job of putting that care into action. They don't want to retire in poverty either, but look at the state of most people's savings.
I am, however, not optimistic that this algorithm would do any better than simply combining family history with lifestyle, or simply (especially in California) going gung-ho for getting people off their lard butts and out on their feet and on bicycles (where most people live, the weather is pretty darn nice).
I don't understand your remark, at all, about health care spending. Since most other first-world countries have some form of universal care, that would tend to result in over-reported spending (if I understand you correctly). Yet they are reported to spend less -- do I therefore infer that their spending is even lower than reported in the statistics?
And your claim about serious illness prognosis -- I am not sure what to make of that. Consider that they live longer. Longevity is a great metric, because we're really good at telling when people are dead (much better than determining if they are "seriously ill", for example). The two metrics are crossed -- I have to assume that (1) we use a more relaxed definition of "serious" in this country, hence that makes us look good or (2) serious illness is so rare, that other factors overwhelm it in the longevity results (in which case, shouldn't I also consider those other factors, too?), or (3) we have a higher rate of serious illness, so even though we treat it better, in the end, it kills a proportionally larger share of people here, thus resulting in reduced longevity.
So what do you make of this? Given the squishiness and ambiguity surrounding your statistic, why do you find it so interesting? I'm a big fan of not dying, from any cause. That seems like a very important metric, yet it isn't for you. Why is that?
One could also raise taxes. In the last decades, my federal income tax has been cut, my state income tax has been cut, and I've had several chances to take advantage of lower interest rates to refinance my mortgage. I recall the Clinton years, that was not so bad, was it?
It's also somewhat interesting to compare how much of our GDP we spend on medical care, versus any other country. Given that by most metrics, we do worse than so many countries that spend much less, arguably, the place where we are objectively pissing away several percent of our GDP, is in the medical/insurance industry.
So what are your travel choices, what's the distance involved? Bicycles, you at least get exercise. E-assist will help you travel longer distances. And the exercise is an enormous quality-of-life thing, that most Americans don't know that they are missing. I've tried working (laptop) on a bus, that's not so good, and subways are too crowded, but the Acela, now that is just dandy (except that the wireless is oversubscribed).
One hopes you checked elevations on that route out of FL (says the guy who grew up there). Some places (e.g., Pinellas County) have got precious few roads out, and most of them are vulnerable to high water (4 bridges, + 19, 582, 580).
Generally, I just be sure I know how to use the water heater as a water supply, plus we keep a fair amount of food in inventory, plus we have a wood stove, plus wood. House is on high enough ground that if that volcano in the Canary Islands slid, we'd be ok. If I had to get somewhere, I would take a bicycle -- runs on peanut butter, sugar, and vegetable oil, not impeded by traffic jams, can carry it over/around obstacles, and all the bubbas think it's just a toy so they'll ignore it.
Squirrel's ok. Had moose, too, but never at the same time, so I missed my chance to say "Vee are haffink Moose and Skvirrel for dinner!".
But I've never cleaned game, so I'm a little unclear on that concept. After what the little squirrel f*ckers did to some fruit tree seedlings this winter (*), I could certainly see eating them, and I'm not sure I'd bet on a vegetable garden doing well unless I did.
(*) snow was 2-3 feet deep in the backyard for a long long time, could not find their acorns, so they gnawed on everything they could find.
I notice talk of "weight savings in the components" but somehow they never got around to telling me what that particular nylon bike weighs in total. That suggests, to me, that the non-component part of the bicycle (e.g., the frame) is not that light, so the entire bike's weight is not nearly so impressive.
And this would be a big deal for a commuter-style bike like that; I played with a cheap folding bike once (Craigslist, yay) on a couple of trips, and the weight was a real issue whenever I was carrying it instead of riding it. Get that bike down below 15lbs, that would be interesting. (And if they DID, and failed to mention it, get better marketing.)
It's not effective enough, though it depends a lot on your threat model. Given that nation-sponsored malware has now been deployed and observed, and I think we need to be more careful.
The oscillators in the circuits to power LEDs, are typically part of a switching power supply, that runs at 100s of kiloHertz. The PWM dimming CAN be run slowly, but that is considered poor practice. The LEDs can easily be switched at tens of kiloHertz. The controller I ran across for "dimmable LEDs on a Triac circuit" (most home dimmers) http://www.national.com/pf/LM/LM3445.html points out that it has no 120Hz flicker; they PWM much faster than that.
I am almost certain that if/when you see flicker in a LED string, it is because of extreme cost cutting. I have not read any recommendations of over-volting, except yours (and obviously, since I knew "LM3445" from memory, I have read a bit).
I don't mean to be insulting, but you have no idea what you are talking about. When there is flicker, it is because someone is using a cheap-and-sleazy rectifier and regulator. As a rule, there is no gain to intermittently overdriving an LED to get "more (average) light"; you would get more (average) light if you ran it at a slightly higher DC voltage, for a given power input and a given heat dissipation. They lose efficiency as you increase their current, that's why you are better off with a constant current.
And I really, truly will not see any flicker from the LEDs I have installed, no matter how much I waved my fingers. I put a scope across one to check for any voltage wobble, and I could only see the vaguest change at the highest time resolution (i.e., faster than 10 kHz).
Oy. You don't want to get me started on bicycles and LEDs (I use a hub dynamo, custom control circuit, the weeds get deeper). I've got a blog (wordpress), there's a dozen or so entries circling around the issue of getting lights right on a bicycle). The way I solve the problem you describe is to have two white LEDs front, one spot, one flat and wide, and put a mirror on top of them to knock the top off.
It's your assertion that the efficiency is 100%. It isn't, both because (1) you aren't using a heat pump, and because of (2) thermodynamic and transmission losses in the electrical generation. If you took the same fuel used to generate the electricity and burned it in a furnace, you'd get about double the heat. These are two frequently-used different ways of heating a house that are more efficient than your claimed "100%", so clearly, it's not 100%.
In fewer words, if you think that resistive electrical heating is efficient, you're probably making a mistake somewhere.
Slashdot Mirror
Zen sarcasm.
Yeah, wait till they drop a rat down your shorts to sniff for explosives.
The first two items, people eventually figure out how wonderful it is to be able to do local commits, without having to explicitly create/name/manage branches. The third item, not an issue if you use ssh for distributed authentication to servers, which you surely will.
And learn to use the mute button, and be sure that you get a good quality connection, and use good microphones. CLI_I_G and DR__OUTS are really hard to stomach for a long meeting.
Yes, it's worth noting just how large automobile engines are nowadays. I learned to drive in a 2000lb car with 42bhp, and a top speed of about 75mph. Amusingly enough, if you take 100/75 (1.33), and cube it (2.35), times 42, gives 98.8 (the power required to overcome wind friction is cubic in speed). So my youthful experiments, are entirely consistent with your assertion. The Prius might do a little better than that, because of improvements in aerodynamics and tire design (the old Saabs were pretty good, but I think we do better now).
However, hill-climbing, is another matter entirely.
Now see, that is just the sort of error message that the Apple Human Interface Guidelines recommend against.
I have no trouble reporting bugs. If there's any obstacles along the way to reporting the bug, I bail, and simply post them on my blog. With snarky commentary, of course, because by then I'm in a bad mood.
Best of all, is when random strangers with the same bug, find my blogged bug report instead of the bug database, and then start posting comments on it.
Years ago, a friend of mine showed me the "killer app for Python", and it was the stack traces. When their (Python-based) product crashed in the field, its driver would bundle it all up in a zip file, and ask the user for permission to forward it to the developers, with the usual reassurances that all data discovered in the process would be treated as confidential. Bug report arrives, the stack trace (with variables etc, it's Python, right?) contains all the necessary info, the next interaction with the customer, is to tell them where to download the bug-fixed version of the product. "No time wasted on annoying human interaction, customer thinks we are attentive geniuses, everyone productive and happy."
My suspicion is, the robot would just go slower. Humans are famous for overestimating their luck and skill. It's also possible that the robot would do a better job even with everything equal -- if nothing else, it is ever-vigilant, and if has the skills to avoid 100% of pedestrian collisions, even assuming pedestrians "at fault", it can probably spot a moose, too. Remember, we don't see in IR.
And maybe travel just takes slightly longer, but you're not busy driving, so you can do something else with the time.
You assume that the robot would be stupid enough to travel at 100km/hr in a moose-crossing zone. Avoid the dangerous situation in the first place, you avoid the need to be able to handle it.
What they didn't tell you, was how smart the car actually is. When you program in a crappy little 2 mile trip, the car will refuse to me, and tell you to take a bicycle instead. The resulting regular exercise and improved cardiovascular health will save far more lives that merely avoiding crashes. (The part about the smart-ass car is a joke. The relative risk calculation, crash vs diseases of the unfit, is not.)
Presumably, the car will be black-boxed out the kazoo. If the robots are better than not just the average human, but better than the average careful human, then insurance companies will jump on it. They play the numbers game, not the "but what if it was ME????" game. If the robots are safer in aggregate, then they will be cheaper to insure, and the black boxes (which will cost little, compared to everything else) will generally provide a handy defense in court.
If it gets to the point that the RIAA lawyers are arguing against the use of de-duplicating storage systems, I think they've probably lost.
Pardon me, but you are misinformed, and have a terrible view of other people.
FIRST, WTF did you think the individual mandate was for, if not to prevent people from buying insurance only after they get sick? It's the quid-pro-quo of forbidding the insurance companies from dicking you around with pre-existing technicalities and requiring that they take-all-customers. (Would be more efficient to cut out the middlemen, but cost and quality were not as important as insurance industry campaign funding.)
SECOND, people DO care about their health, but they tend to discount the future horribly and believe that because of their all-encompassing above-averageness (the Lake Wobegon Effect), bad stuff will not happen to them. People do care about their health, they just don't do a very good job of putting that care into action. They don't want to retire in poverty either, but look at the state of most people's savings.
I am, however, not optimistic that this algorithm would do any better than simply combining family history with lifestyle, or simply (especially in California) going gung-ho for getting people off their lard butts and out on their feet and on bicycles (where most people live, the weather is pretty darn nice).
I think your pun detector is a little rusty.
I don't understand your remark, at all, about health care spending. Since most other first-world countries have some form of universal care, that would tend to result in over-reported spending (if I understand you correctly). Yet they are reported to spend less -- do I therefore infer that their spending is even lower than reported in the statistics?
And your claim about serious illness prognosis -- I am not sure what to make of that. Consider that they live longer. Longevity is a great metric, because we're really good at telling when people are dead (much better than determining if they are "seriously ill", for example). The two metrics are crossed -- I have to assume that (1) we use a more relaxed definition of "serious" in this country, hence that makes us look good or (2) serious illness is so rare, that other factors overwhelm it in the longevity results (in which case, shouldn't I also consider those other factors, too?), or (3) we have a higher rate of serious illness, so even though we treat it better, in the end, it kills a proportionally larger share of people here, thus resulting in reduced longevity.
So what do you make of this? Given the squishiness and ambiguity surrounding your statistic, why do you find it so interesting? I'm a big fan of not dying, from any cause. That seems like a very important metric, yet it isn't for you. Why is that?
One could also raise taxes. In the last decades, my federal income tax has been cut, my state income tax has been cut, and I've had several chances to take advantage of lower interest rates to refinance my mortgage. I recall the Clinton years, that was not so bad, was it?
It's also somewhat interesting to compare how much of our GDP we spend on medical care, versus any other country. Given that by most metrics, we do worse than so many countries that spend much less, arguably, the place where we are objectively pissing away several percent of our GDP, is in the medical/insurance industry.
So what are your travel choices, what's the distance involved? Bicycles, you at least get exercise. E-assist will help you travel longer distances. And the exercise is an enormous quality-of-life thing, that most Americans don't know that they are missing. I've tried working (laptop) on a bus, that's not so good, and subways are too crowded, but the Acela, now that is just dandy (except that the wireless is oversubscribed).
One hopes you checked elevations on that route out of FL (says the guy who grew up there). Some places (e.g., Pinellas County) have got precious few roads out, and most of them are vulnerable to high water (4 bridges, + 19, 582, 580).
Generally, I just be sure I know how to use the water heater as a water supply, plus we keep a fair amount of food in inventory, plus we have a wood stove, plus wood. House is on high enough ground that if that volcano in the Canary Islands slid, we'd be ok. If I had to get somewhere, I would take a bicycle -- runs on peanut butter, sugar, and vegetable oil, not impeded by traffic jams, can carry it over/around obstacles, and all the bubbas think it's just a toy so they'll ignore it.
Squirrel's ok. Had moose, too, but never at the same time, so I missed my chance to say "Vee are haffink Moose and Skvirrel for dinner!".
But I've never cleaned game, so I'm a little unclear on that concept. After what the little squirrel f*ckers did to some fruit tree seedlings this winter (*), I could certainly see eating them, and I'm not sure I'd bet on a vegetable garden doing well unless I did.
(*) snow was 2-3 feet deep in the backyard for a long long time, could not find their acorns, so they gnawed on everything they could find.
I notice talk of "weight savings in the components" but somehow they never got around to telling me what that particular nylon bike weighs in total. That suggests, to me, that the non-component part of the bicycle (e.g., the frame) is not that light, so the entire bike's weight is not nearly so impressive.
And this would be a big deal for a commuter-style bike like that; I played with a cheap folding bike once (Craigslist, yay) on a couple of trips, and the weight was a real issue whenever I was carrying it instead of riding it. Get that bike down below 15lbs, that would be interesting. (And if they DID, and failed to mention it, get better marketing.)
It's not effective enough, though it depends a lot on your threat model. Given that nation-sponsored malware has now been deployed and observed, and I think we need to be more careful.
The oscillators in the circuits to power LEDs, are typically part of a switching power supply, that runs at 100s of kiloHertz. The PWM dimming CAN be run slowly, but that is considered poor practice. The LEDs can easily be switched at tens of kiloHertz. The controller I ran across for "dimmable LEDs on a Triac circuit" (most home dimmers) http://www.national.com/pf/LM/LM3445.html points out that it has no 120Hz flicker; they PWM much faster than that.
I am almost certain that if/when you see flicker in a LED string, it is because of extreme cost cutting. I have not read any recommendations of over-volting, except yours (and obviously, since I knew "LM3445" from memory, I have read a bit).
I don't mean to be insulting, but you have no idea what you are talking about. When there is flicker, it is because someone is using a cheap-and-sleazy rectifier and regulator. As a rule, there is no gain to intermittently overdriving an LED to get "more (average) light"; you would get more (average) light if you ran it at a slightly higher DC voltage, for a given power input and a given heat dissipation. They lose efficiency as you increase their current, that's why you are better off with a constant current.
And I really, truly will not see any flicker from the LEDs I have installed, no matter how much I waved my fingers. I put a scope across one to check for any voltage wobble, and I could only see the vaguest change at the highest time resolution (i.e., faster than 10 kHz).
Oy. You don't want to get me started on bicycles and LEDs (I use a hub dynamo, custom control circuit, the weeds get deeper). I've got a blog (wordpress), there's a dozen or so entries circling around the issue of getting lights right on a bicycle). The way I solve the problem you describe is to have two white LEDs front, one spot, one flat and wide, and put a mirror on top of them to knock the top off.
It's your assertion that the efficiency is 100%. It isn't, both because (1) you aren't using a heat pump, and because of (2) thermodynamic and transmission losses in the electrical generation. If you took the same fuel used to generate the electricity and burned it in a furnace, you'd get about double the heat. These are two frequently-used different ways of heating a house that are more efficient than your claimed "100%", so clearly, it's not 100%.
In fewer words, if you think that resistive electrical heating is efficient, you're probably making a mistake somewhere.