There's a reason in the art world if a painting cannot be tracked through it's whole life it's first considered a fake.
Except of course for all the paintings not discovered to be by someone considered important until years, decades, or centuries after the work was created. Something that's actually done fairly routinely.
In which case the burden of proof lies on the discoverer, unless I'm greatly mistaken. If you discover that a piece is by a famous artist and attempt to sell it as such, I imagine the first question anyone asks will be "How do you know?" I think this was the point the parent post was trying to make -- such things are considered fakes until proven otherwise.
All you're saying is that this method doesn't solve all the world's problems, but we already knew that. Think of it as a more environmentally friendly way of producing carbon black, that happens to clean up some uses of natural gas as a side effect, rather than the other way around. Sure, it's not a world-changing effect, but it's helpful nonetheless.
No, that's the whole point. The ozone molecule consumed in oxidizing the H2 to H2O is irrelevant. However, by the time that occurs, the H2 has floated up to an altitude that normally has almost no water present. The microscopic ice crystals that result from that reaction are far from harmless, though: it is on their surfaces that the CFC reactions occur. Adding more ice increases the ozone damage caused by the already-present CFCs.
Water ice is a potent ozone depleter at those altitudes. Normally, though, the water freezes out before getting there. Hydrogen, however, survives quite easily to that altitude where it turns into water ice. A small amount of ice is naturally occurring, which is why CFCs are a problem -- but that amount is so small that hydrogen pollution is a relevant source of it.
It uses less energy and less methane than normal carbon black production. True, a lot of methane passes through the reactor, but most of it leaves in enriched form that can easily be used somewhere else. The methane *consumed* is less than a conventional carbon black production process. (The carbon black market is huge -- it's a major component of tire rubber and other industrial plastics.)
You're looking at it backwards. It should really be viewed as a more efficient way of producing carbon black (which there is a huge market for, btw -- it's a major component of tire rubber, rubber hoses, and similar plastics) that happens to have some nice side effects (like producing an enriched natural gas with cleaner combustion properties).
The current carbon black production techniques involve sooty combustion of hydrocarbon fuels; the energy from that process is normally wasted, since it's in a form that is difficult to recapture. This process manages to waste less energy, since the electricity input is modest and some of the electrical energy and fossil fuel energy spend making the carbon black is stored in the H2, which can be used productively by enriching the unused portion of the natural gas stream.
(Also, there's no reason the electricity to run this *has* to come from fossil fuels. It could come from nuclear or renewable sources. It's the same as electric cars -- saying "but the electricity comes from fossil fuels!" is true but misses the point -- it's easier to swap out your electric source later on than to swap your car / chemical plant. Going to a process that can easily choose a cleaner energy source is a good thing, even if that source won't be available immediately.)
I'm not really buying the idea that hydrogen-enriched natural gas will burn more cleanly. It will produce less CO2, true, but at the price of less energy per unit volume. And natural gas can already be burned less completely.
Combustion chemistry is best described as really weird. Different fuels have a large impact on how much nitrogen burns to nitrogen oxides, as well as how completely the fuel burns. Details of the combustion environment (mixing, combustion time, combustion temp, pressure, etc) also have a huge impact. There is plenty of evidence that adding H2 to normal hydrocarbon fuels makes them burn both more completely and with less NOx production. Oxygen-bearing fuels (eg ethanol added to gasoline) can also have similar effects. Normally adding H2 has a large enough energy cost that it isn't viable, but if this process can do it easily and efficiently, that's interesting.
You're absolutely correct, however the mechanisms by which that happens are anything but trivial. Toward the end of TFA they address this -- apparently normal natural gas pipeline materials don't get brittle if the H2 content is kept to around 10%. You'll still see some leakage, but the leak rate through most metals is fairly low. Plastics show high leak rates, as do some metals like aluminum (I think; not my area of expertise, and I haven't looked up details recently), and some glasses.
The hydrogen leakage is a bigger deal as an ozone-depleting pollutant than as an energy loss. The reactions by which CFCs destroy ozone occur on the surface of ice crystals. Normally, water vapor all freezes out of the atmosphere before it gets that high, so there isn't much ice at ozone layer altitudes. Hydrogen, however, happily floats up that high, where it is oxidized by the ozone and forms ice.
On the other hand, there is a large market for carbon black. If you remove the carbon and sell it, while getting the energy from the hydrogen, your biogas is now carbon-negative, which is even better. Whether it could be economical or not likely depends on things like cap and trade -- with no incentives for being carbon-negative, it probably doesn't make sense economically, with them it might, depending on the size.
There are some straightforward compaction algorithms for non power-of-two sizes. The simplest approach is to take n symbols in your alphabet, treat it as an n-digit number base b (the number of different symbols), and convert that to base two. You'll use at most ceiling(n * log2(b)) bits.
You can be more sophisticated by using a compression algorithm of some sort (Huffman with a standardized dictionary, for a simple example). Anything that does better than the above n * log2(b) will produce a variable length output, though, which means that while you could usually fit more than 160 characters into 140 bytes, sometimes the limit would be lower (since rare characters take more bits to encode).
As long as the standard quality version is free, and the price is small, there are plenty of youtube videos I'd pay $0.01-$0.10 to see the high quality version of. A lot of the nifty science demos would look cooler in HD, for example.
I would consider paying, but there would be several hurdles. I'd prefer optional tipping, provided there was a dead simple way to tip a tiny amount, but I might consider paying even if it was a more traditional model.
First, I would only be willing to pay for higher-quality versions. If there isn't a low-quality (ie current normal youtube quality) free, I'm not interested in paying sight unseen.
Second, it needs to be a true micropayment, and they need to somehow make it really trivial to use. I'm not particularly interested in giving them blanket access to my bank account, and I'm not particularly interested in worrying about how much is left in my special youtube account and periodically transferring money. Yes, I realize that doing both of those is probably impossible right now. Their problem, not mine.
Third, they need to provide a download option, at least on larger things. I'm not interested in watching a TV show in my browser, or in having it stop halfway through because my Internet connection hiccuped and it couldn't keep streaming.
And fourth, it needs to be per-video, not per-viewing. I don't want to count the times I've seen something cool on youtube and then later pulled it up to play it for a friend on my computer. I don't mind paying a couple pennies for the good quality version of a neat video, but I mind paying it repeatedly.
That's the goal of the owners and marketers. I suspect most reporters hold with the older ideals. And take a look at who implemented this idea, and who spoke out against it...
I meant Linux development support -- it would need a lot more memory and an MMU to get Linux running on it.
As a 32-bit device at 20 MIPS per core and 8 cores, it has significantly more performance than the 8-bit AVRs (40 MIPS max, I believe) and PICs (12 MIPS) in most applications. The PICs will outperform it if you need a lot of multiplies -- hardware multiply of 8 by 8 to 16 bit result in a single cycle; that can do a 16x16 to 32 in about 2us, iirc. A single propeller core looks like it would be a bit slower at that, but if you can get several cores working on the problem, it might do better.
For performance in a hobbyist-friendly DIP package, though, the real competitor is the 16-bit dsPIC line. Those have a pair of 40-bit accumulators and a DSP engine that can do a single cycle A += B*C, where B and C are 16 bit variables in memory and A is the accumulator. And it will increment the registers that store the pointers, so you really can do an n-element 16-bit dot product in n cycles, running at 40 MIPS.
If you're interfacing purely with other digital hardware, I suspect that in many cases you could replace the hardware peripherals of a PIC with a cog, at a cost in software complexity. The end result is that you have one main thread talking to several peripherals, much like in a PIC.
The simple way is to connect the LED anode to the PIC pin, and cathode to ground. Then drive the PIC pin to ground. Change the pin to a digital input and wait for it to trip high. The photodiode current combines with the diode capacitance and pin capacitance to determine the rate of charge, and the digital input makes an appropriate comparator. Time how long it takes the pin to go high, and you have your light measurement. The A/D might work, but it presents more load and is slower and more complicated to use. And of course, if you want you can then drive the pin high to turn on the LED.
In this mode the LED is wired as a solar cell; it is always forward-biased. Normal photodiode sensors are usually run in a reverse-biased mode for increased sensitivity, but that isn't a requirement. In fact, if you get a largish LED, the solar cell current is large enough to measure directly on a multimeter -- a few hundred nA in moderate light.
(Not my idea, and I don't remember where I first saw it.)
The Propeller looks really interesting. I might get one to play with, but I'm disappointed by the lack of Linux support.
I'm also surprised they don't have *any* hardware peripherals on-chip. I'm used to working with the PIC microcontrollers, which give me tons of things like UARTs, USB interface, SPI controller, CAN bus, A/D converters, timers, PWM output, comparators, etc. Obviously some of that can be implemented with one of the cogs, but some of it would be hard or impossible. The lack of hardware multiply and divide support is also annoying.
Most of my projects involve interfacing with the outside world in some non-trivial fashion; the Propeller doesn't do much to make that easy, which is disappointing. I do appreciate that they have a DIP package, though -- most high-performance microcontrollers these days are surface-mount only, which makes breadboards somewhat tricky.
The low-power, low-voltage op amps are impressive -- I'll see your LMC6462 and raise you an LT6003: 1.6 to 16 volts, 1uA supply, though the input resistance is slightly worse at 10GOhm (differential) to 2TOhm (common mode).
In some ways more impressive, imho, are the high speed precision op amps. Take a look at the LT1468, for example -- 90MHz, 75uV offset, and settles to 150uV in under a microsecond.
On the other hand, most of my breadboards still begin life with a uA741 or LM324 -- I'd much rather let the smoke out of a cheap op amp than an expensive one. Once the smoke stays in, I'll swap it for the one that will actually act as a precision part.
The operational basics of the 555 are completely explained by a half-page functional block diagram. You can easily fit all the important max ratings and speeds and such on the other half of the page. Even the 10F200 has a 96 page data sheet (though to be fair, to be that thorough about the 555 would probably require 2 or 3 pages, not just one).
The PIC has a lot going for it when compared to a 555, but simplicity is not one of those things.
The PIC does it with three external components -- a regulator and a capacitor for power, and a resistor to help drive the LED. If you run at lower supply voltages you can omit the resistor and use the output impedance of the PIC instead, provided you don't care about tweaking the power consumption. Lower parts count and less board area is cheaper, and the PIC is only marginally more expensive than the 555.
And not only is the PIC cheaper, it can do a better job for most circuits. It will operate a more accurate long-period timer without precision components, and do it at sub-microamp power consumption.
At the extreme end, I can make a PIC blink an LED only when it's dark out, using only a CR2032 coin cell, a PIC, and LED. Let's see you get a 555 to do anything useful with two external components, including the battery.
Of course, I say all that, but I prefer to build my circuits with op amps instead of PICs, and I debug them with a Tek 561A. Heck, in the right context a medium speed op amp or three has more compute power than a PIC...
Now, this is one student's transcription as best as he could. The story changes a lot with each telling, but it's always hilarious. The best part of it, of course, is that the professor either totally believes it or is the best troll ever.
Troll? Hardly. That precise format is how all the best ghost stories get told. It sounds to me like most of the audience simply wasn't used to oral storytelling as an art form.
If you are going to explain that it is a joke, you might as well not bother in the first place since explaining takes away all the fun.
Make your snide remarks and take your moderations like a man.
If you're going to try to avoid the downmods and explain your reasoning, you might as well not bother in the first place since explaining takes away all the fun.
Actually, it doesn't appear to be so much a range problem as a line of sight one. Behind the monitor is bad, but a couple inches away and not behind the metal monitor casing is fine. I have no need for 28-ft range on my mouse, so I think I'd prefer the additional battery life (allegedly 1-year, but I haven't tested that yet).
If you have an ultrasonic cleaner, pull the batteries and drop it in. Alcohol is preferred because it will dry faster than water (and work slightly better, but I doubt it matters). Either way, you should make sure it's thoroughly dry before replacing the batteries. This step may take a while if you used water.
For keyboards, I've done that (requires a large ultrasonic cleaner -- I used one at work, since they're too expensive to buy one myself) and also just tossed it in the top rack of the dishwasher. Again, allow to dry very thoroughly (12-24 hours if you don't disassemble for drying) before applying power.
I have one of the Logitech cordless (non-bluetooth) mice. It takes AA batteries (alkaline only, not NiMH or NiCd, unfortunately). It has a low battery indicator, so I don't worry about it dieing unexpectedly. Of course, in the four months (maybe five, I don't remember exactly) I've had it, that indicator has yet to come on, so I don't actually know how much time is left when it does.
Interference is annoying. It took me a little while to realize that I can't plug the receiver into the back of the computer or the back of the monitor. I got a short USB extension cable and the receiver now sits on my desk behind the keyboard, and it works wonderfully. I'd hoped the range would be longer, but this isn't a problem. Overall, I'm fairly happy with it.
They merge into a single black hole, spherical except for the deformations resulting from spin. If they're in close orbit, they'll lose energy to gravity waves and other forms of radiation (both Hawking and synchrotron) and spiral into each other. The gravity waves should be quite strong -- one of the sources that LIGO etc ought to be able to detect.
Of course, IANAA either, so I might be off base here, but that's my recollection as to what happens.
Slashdot Mirror
Except of course for all the paintings not discovered to be by someone considered important until years, decades, or centuries after the work was created. Something that's actually done fairly routinely.
In which case the burden of proof lies on the discoverer, unless I'm greatly mistaken. If you discover that a piece is by a famous artist and attempt to sell it as such, I imagine the first question anyone asks will be "How do you know?" I think this was the point the parent post was trying to make -- such things are considered fakes until proven otherwise.
All you're saying is that this method doesn't solve all the world's problems, but we already knew that. Think of it as a more environmentally friendly way of producing carbon black, that happens to clean up some uses of natural gas as a side effect, rather than the other way around. Sure, it's not a world-changing effect, but it's helpful nonetheless.
No, that's the whole point. The ozone molecule consumed in oxidizing the H2 to H2O is irrelevant. However, by the time that occurs, the H2 has floated up to an altitude that normally has almost no water present. The microscopic ice crystals that result from that reaction are far from harmless, though: it is on their surfaces that the CFC reactions occur. Adding more ice increases the ozone damage caused by the already-present CFCs.
Water ice is a potent ozone depleter at those altitudes. Normally, though, the water freezes out before getting there. Hydrogen, however, survives quite easily to that altitude where it turns into water ice. A small amount of ice is naturally occurring, which is why CFCs are a problem -- but that amount is so small that hydrogen pollution is a relevant source of it.
It uses less energy and less methane than normal carbon black production. True, a lot of methane passes through the reactor, but most of it leaves in enriched form that can easily be used somewhere else. The methane *consumed* is less than a conventional carbon black production process. (The carbon black market is huge -- it's a major component of tire rubber and other industrial plastics.)
You're looking at it backwards. It should really be viewed as a more efficient way of producing carbon black (which there is a huge market for, btw -- it's a major component of tire rubber, rubber hoses, and similar plastics) that happens to have some nice side effects (like producing an enriched natural gas with cleaner combustion properties).
The current carbon black production techniques involve sooty combustion of hydrocarbon fuels; the energy from that process is normally wasted, since it's in a form that is difficult to recapture. This process manages to waste less energy, since the electricity input is modest and some of the electrical energy and fossil fuel energy spend making the carbon black is stored in the H2, which can be used productively by enriching the unused portion of the natural gas stream.
(Also, there's no reason the electricity to run this *has* to come from fossil fuels. It could come from nuclear or renewable sources. It's the same as electric cars -- saying "but the electricity comes from fossil fuels!" is true but misses the point -- it's easier to swap out your electric source later on than to swap your car / chemical plant. Going to a process that can easily choose a cleaner energy source is a good thing, even if that source won't be available immediately.)
I'm not really buying the idea that hydrogen-enriched natural gas will burn more cleanly. It will produce less CO2, true, but at the price of less energy per unit volume. And natural gas can already be burned less completely.
Combustion chemistry is best described as really weird. Different fuels have a large impact on how much nitrogen burns to nitrogen oxides, as well as how completely the fuel burns. Details of the combustion environment (mixing, combustion time, combustion temp, pressure, etc) also have a huge impact. There is plenty of evidence that adding H2 to normal hydrocarbon fuels makes them burn both more completely and with less NOx production. Oxygen-bearing fuels (eg ethanol added to gasoline) can also have similar effects. Normally adding H2 has a large enough energy cost that it isn't viable, but if this process can do it easily and efficiently, that's interesting.
You're absolutely correct, however the mechanisms by which that happens are anything but trivial. Toward the end of TFA they address this -- apparently normal natural gas pipeline materials don't get brittle if the H2 content is kept to around 10%. You'll still see some leakage, but the leak rate through most metals is fairly low. Plastics show high leak rates, as do some metals like aluminum (I think; not my area of expertise, and I haven't looked up details recently), and some glasses.
The hydrogen leakage is a bigger deal as an ozone-depleting pollutant than as an energy loss. The reactions by which CFCs destroy ozone occur on the surface of ice crystals. Normally, water vapor all freezes out of the atmosphere before it gets that high, so there isn't much ice at ozone layer altitudes. Hydrogen, however, happily floats up that high, where it is oxidized by the ozone and forms ice.
On the other hand, there is a large market for carbon black. If you remove the carbon and sell it, while getting the energy from the hydrogen, your biogas is now carbon-negative, which is even better. Whether it could be economical or not likely depends on things like cap and trade -- with no incentives for being carbon-negative, it probably doesn't make sense economically, with them it might, depending on the size.
A mass spectrometer can operate on a few milligrams of carbon. That means you need perhaps as much as 50 microliters of whiskey, or about 0.0017 oz.
Burning $0.50 worth of whiskey makes sense to me when testing a $20,000 bottle that has a greater than 50% chance of being a fake.
There are some straightforward compaction algorithms for non power-of-two sizes. The simplest approach is to take n symbols in your alphabet, treat it as an n-digit number base b (the number of different symbols), and convert that to base two. You'll use at most ceiling(n * log2(b)) bits.
You can be more sophisticated by using a compression algorithm of some sort (Huffman with a standardized dictionary, for a simple example). Anything that does better than the above n * log2(b) will produce a variable length output, though, which means that while you could usually fit more than 160 characters into 140 bytes, sometimes the limit would be lower (since rare characters take more bits to encode).
As long as the standard quality version is free, and the price is small, there are plenty of youtube videos I'd pay $0.01-$0.10 to see the high quality version of. A lot of the nifty science demos would look cooler in HD, for example.
I would consider paying, but there would be several hurdles. I'd prefer optional tipping, provided there was a dead simple way to tip a tiny amount, but I might consider paying even if it was a more traditional model.
First, I would only be willing to pay for higher-quality versions. If there isn't a low-quality (ie current normal youtube quality) free, I'm not interested in paying sight unseen.
Second, it needs to be a true micropayment, and they need to somehow make it really trivial to use. I'm not particularly interested in giving them blanket access to my bank account, and I'm not particularly interested in worrying about how much is left in my special youtube account and periodically transferring money. Yes, I realize that doing both of those is probably impossible right now. Their problem, not mine.
Third, they need to provide a download option, at least on larger things. I'm not interested in watching a TV show in my browser, or in having it stop halfway through because my Internet connection hiccuped and it couldn't keep streaming.
And fourth, it needs to be per-video, not per-viewing. I don't want to count the times I've seen something cool on youtube and then later pulled it up to play it for a friend on my computer. I don't mind paying a couple pennies for the good quality version of a neat video, but I mind paying it repeatedly.
That's the goal of the owners and marketers. I suspect most reporters hold with the older ideals. And take a look at who implemented this idea, and who spoke out against it...
I meant Linux development support -- it would need a lot more memory and an MMU to get Linux running on it.
As a 32-bit device at 20 MIPS per core and 8 cores, it has significantly more performance than the 8-bit AVRs (40 MIPS max, I believe) and PICs (12 MIPS) in most applications. The PICs will outperform it if you need a lot of multiplies -- hardware multiply of 8 by 8 to 16 bit result in a single cycle; that can do a 16x16 to 32 in about 2us, iirc. A single propeller core looks like it would be a bit slower at that, but if you can get several cores working on the problem, it might do better.
For performance in a hobbyist-friendly DIP package, though, the real competitor is the 16-bit dsPIC line. Those have a pair of 40-bit accumulators and a DSP engine that can do a single cycle A += B*C, where B and C are 16 bit variables in memory and A is the accumulator. And it will increment the registers that store the pointers, so you really can do an n-element 16-bit dot product in n cycles, running at 40 MIPS.
If you're interfacing purely with other digital hardware, I suspect that in many cases you could replace the hardware peripherals of a PIC with a cog, at a cost in software complexity. The end result is that you have one main thread talking to several peripherals, much like in a PIC.
The simple way is to connect the LED anode to the PIC pin, and cathode to ground. Then drive the PIC pin to ground. Change the pin to a digital input and wait for it to trip high. The photodiode current combines with the diode capacitance and pin capacitance to determine the rate of charge, and the digital input makes an appropriate comparator. Time how long it takes the pin to go high, and you have your light measurement. The A/D might work, but it presents more load and is slower and more complicated to use. And of course, if you want you can then drive the pin high to turn on the LED.
In this mode the LED is wired as a solar cell; it is always forward-biased. Normal photodiode sensors are usually run in a reverse-biased mode for increased sensitivity, but that isn't a requirement. In fact, if you get a largish LED, the solar cell current is large enough to measure directly on a multimeter -- a few hundred nA in moderate light.
(Not my idea, and I don't remember where I first saw it.)
The Propeller looks really interesting. I might get one to play with, but I'm disappointed by the lack of Linux support.
I'm also surprised they don't have *any* hardware peripherals on-chip. I'm used to working with the PIC microcontrollers, which give me tons of things like UARTs, USB interface, SPI controller, CAN bus, A/D converters, timers, PWM output, comparators, etc. Obviously some of that can be implemented with one of the cogs, but some of it would be hard or impossible. The lack of hardware multiply and divide support is also annoying.
Most of my projects involve interfacing with the outside world in some non-trivial fashion; the Propeller doesn't do much to make that easy, which is disappointing. I do appreciate that they have a DIP package, though -- most high-performance microcontrollers these days are surface-mount only, which makes breadboards somewhat tricky.
The low-power, low-voltage op amps are impressive -- I'll see your LMC6462 and raise you an LT6003: 1.6 to 16 volts, 1uA supply, though the input resistance is slightly worse at 10GOhm (differential) to 2TOhm (common mode).
In some ways more impressive, imho, are the high speed precision op amps. Take a look at the LT1468, for example -- 90MHz, 75uV offset, and settles to 150uV in under a microsecond.
On the other hand, most of my breadboards still begin life with a uA741 or LM324 -- I'd much rather let the smoke out of a cheap op amp than an expensive one. Once the smoke stays in, I'll swap it for the one that will actually act as a precision part.
The operational basics of the 555 are completely explained by a half-page functional block diagram. You can easily fit all the important max ratings and speeds and such on the other half of the page. Even the 10F200 has a 96 page data sheet (though to be fair, to be that thorough about the 555 would probably require 2 or 3 pages, not just one).
The PIC has a lot going for it when compared to a 555, but simplicity is not one of those things.
The PIC does it with three external components -- a regulator and a capacitor for power, and a resistor to help drive the LED. If you run at lower supply voltages you can omit the resistor and use the output impedance of the PIC instead, provided you don't care about tweaking the power consumption. Lower parts count and less board area is cheaper, and the PIC is only marginally more expensive than the 555.
And not only is the PIC cheaper, it can do a better job for most circuits. It will operate a more accurate long-period timer without precision components, and do it at sub-microamp power consumption.
At the extreme end, I can make a PIC blink an LED only when it's dark out, using only a CR2032 coin cell, a PIC, and LED. Let's see you get a 555 to do anything useful with two external components, including the battery.
Of course, I say all that, but I prefer to build my circuits with op amps instead of PICs, and I debug them with a Tek 561A. Heck, in the right context a medium speed op amp or three has more compute power than a PIC...
Now, this is one student's transcription as best as he could. The story changes a lot with each telling, but it's always hilarious. The best part of it, of course, is that the professor either totally believes it or is the best troll ever.
Troll? Hardly. That precise format is how all the best ghost stories get told. It sounds to me like most of the audience simply wasn't used to oral storytelling as an art form.
If you are going to explain that it is a joke, you might as well not bother in the first place since explaining takes away all the fun.
Make your snide remarks and take your moderations like a man.
If you're going to try to avoid the downmods and explain your reasoning, you might as well not bother in the first place since explaining takes away all the fun.
Actually, it doesn't appear to be so much a range problem as a line of sight one. Behind the monitor is bad, but a couple inches away and not behind the metal monitor casing is fine. I have no need for 28-ft range on my mouse, so I think I'd prefer the additional battery life (allegedly 1-year, but I haven't tested that yet).
If you have an ultrasonic cleaner, pull the batteries and drop it in. Alcohol is preferred because it will dry faster than water (and work slightly better, but I doubt it matters). Either way, you should make sure it's thoroughly dry before replacing the batteries. This step may take a while if you used water.
For keyboards, I've done that (requires a large ultrasonic cleaner -- I used one at work, since they're too expensive to buy one myself) and also just tossed it in the top rack of the dishwasher. Again, allow to dry very thoroughly (12-24 hours if you don't disassemble for drying) before applying power.
I have one of the Logitech cordless (non-bluetooth) mice. It takes AA batteries (alkaline only, not NiMH or NiCd, unfortunately). It has a low battery indicator, so I don't worry about it dieing unexpectedly. Of course, in the four months (maybe five, I don't remember exactly) I've had it, that indicator has yet to come on, so I don't actually know how much time is left when it does.
Interference is annoying. It took me a little while to realize that I can't plug the receiver into the back of the computer or the back of the monitor. I got a short USB extension cable and the receiver now sits on my desk behind the keyboard, and it works wonderfully. I'd hoped the range would be longer, but this isn't a problem. Overall, I'm fairly happy with it.
They merge into a single black hole, spherical except for the deformations resulting from spin. If they're in close orbit, they'll lose energy to gravity waves and other forms of radiation (both Hawking and synchrotron) and spiral into each other. The gravity waves should be quite strong -- one of the sources that LIGO etc ought to be able to detect.
Of course, IANAA either, so I might be off base here, but that's my recollection as to what happens.