The problem isn't government though - its Americas government.. Or Europe's or UK's government or.. Anyway efficiency and competence are the most important thing - whether its private, public, socialist, liberal, Republican or whatever..
In the US the biggest problems include corporate socialism, wealth and social inequality, the slush of money the richest 0.01% bring to the table, and basic political stupidity.. In Europe the biggest problems are things like lowest bid tendering, privatisation, the single currency, the open single market, corporate socialism, excessive bureaucracy, and basic political stupidity..
All we need is intelligent, competent, efficient, compassionate, moral, and strong government. Maybe we need a new "Moon Program" to replace our politicians with something better - like machines. (Disclaimer : I am working on developing such machines)
You are absolutely right, Strong AI is non-deterministic, well not strictly deterministic. In some ways its quite 'un-computer-like' and although Strong AI could theoretically be run on ordinary hardware in practice this is unlikely to work. One of the reasons is basic low level memory management, which has long been one of the biggest stumbling points of strong AI. The machine needs a persistent and stable memory that is constantly online and this is surprisingly difficult to achieve. (heaps and lists and stacks are not reliable enough)
My early attempts based on C++ tried to build a system of 'encapsulation' using doubly recursive pointers, but in the end it kept becoming a tangled mess of spaghetti and just wouldn't work... The language simply wasn't designed to do it. C++ has plenty of encapsulation on the source code text side but on the run-time side there is basically none at all. I tried writing the core in machine code but that didn't get very far either. - The real problem is in the way the memory manager and CPU work together - controlled by the operating system. I needed total control of this and couldn't get it - the thing is also enormously complicated and poorly documented. Its not a place OS and hardware vendors want anyone looking. The basic problem though was that I was trying to get the machine to do something it really wasn't designed to do. I came to the conclusion again and again that building a custom CPU was the only answer, and eventually gave up on it as an impossible problem.. (I have considered the alternative of building a virtual machine solution but this would still have similar problems)
Ten years later and I discovered people are building systems using custom CPU's built in FPGA's. The nightmare now is learning enough Verilog and digital electronics to get the thing working... The biggest problem is that it requires more than one CPU and more than one FPGA, and this ultimately means building a custom motherboard. Fortunately electronics design software is now advanced enough to do this with relatively little work (with a couple of electronics engineers).. The total budget is probably less than $100,000 to $200,000. At a pinch the prototype can be built using FPGA dev boards putting it within the reach of one person and maybe $10,000.. plus of course another 10 years...
I have worked on and off on Strong AI for 25 years... 20 years ago (in 1995) I had a workable design for building a working machine but I calculated that it would need a budget of about $100 million.. In 2003 the base cost was about the same - though the final machine would fit inside a PC case instead of multiple filing cabinets.
A couple of years ago I restarted work on the thing, because technology has advanced. Now the whole machine should fit inside a 1 foot cube, but now the minimum base dev costs have shrunk to about $10,000.. (based on FPGA) Its still a 10 year project because it requires a special purpose computer. It turns out to be 'easier' to build a complete new architecture from the ground up than to reverse engineer an OS like Linux to make it compatible with strong AI, the machine needs custom hardware to work anyway...
Of course the machine will still need many millions to go from a prototype to a production machine. Strong AI has some weird problems with security and that makes everything so much more difficult - it will probably end up as being classified as both military and civilian and being quite unlike anything we have seen before.. And yes Strong AI will eventually be a multi-trillion dollar global market... but try telling an investor that their trillion dollar return is twenty or thirty years away.. 'High risk' doesn't even begin to cover it.. : )
NASA isn't the key point, the V rocket program was. Without Hitler's investment in the V rocket program we might not be in orbit even today.. The whole problem was that rockets weren't taken seriously and might never have received the investment they needed - the V program took them from being a toy to sophisticated machines able to reach supersonic speeds and touch the edge of orbit.
Once you can cross the coulomb barrier and can stay there, your not far from getting all the toys - like total atomic conversion, bending space and time - gravity engines, contained singularities, force fields, manipulating energy and entropy. From there going for fusion looks a little - 'primitive'.
Of course crossing that barrier and staying there also looks like one of the hardest things imaginable - a contained quark gluon plasma.. I like working on left-field 'impossible' problems..:)
There is a much simpler way to stop them if you want to and its instant. In any decent bank account with online control there should be a section in the settings for allowed payees, all you have to do is remove PayPal from that list and it stops them from taking money instantly. I did this about four years ago and now only use PayPal through my credit cards - it all works perfectly, is instant and if things go wrong mostly easy to fix...
If you smash open your skull you will find some shit inside. You can use it to fertilize the garden. : ) Seriously though if you really want to campaign against nuclear power then you should campaign against the one source that really does kill people in large numbers - the sun. (est 500,000+ per year)
Radiation from Fukushima has probably killed less than ~ 10 people. However Fukushima has killed something like 2,000 to maybe 10,000 people - because Japan switched its nuclear stations off and switched to coal and oil - which are literally 10,000 time more dangerous than nuclear.. Real problem, nuclear is scary to simple people.
The real problem is that all real research into nuclear power has all but stopped largely thanks to the anti-nuclear lobby. Given that pollution from burning coal & oil has killed something like 70 million to over 100 million people since 1945 the safety bar for nuclear is set insanely high.. With money more efficient fuel cycles and technologies could b developed.
Of course with say $10-20 billion and a more efficient bureaucracy the development time to working commercial fusion plants could be reduced from 40+ years to something like 10 or 20 years. (using a technique like parallel development) Fusion would probably render all fission systems obsolete..
Sorry but the USA Today article is scientifically illiterate. I know Von Braun's early designs quite well and they were created before he had the experience of Apollo. (he was also a primary designer within Apollo) The problem with those early designs was that they were too large and were hopelessly cumbersome and expensive - try (at least) 10 to 20 times the final cost of the actual program that was built.
I'm actually a fan of using nuclear powered reusable orbital lifters but they were definitely a step too far in 1960, and would be a complex technology to develop even today. Look at the Skylon program in the UK - with tech well beyond the 1960's, 70's, or 80's. Apollo was exactly the right tech at the right time, it was the political cowardice that came later that screwed up the space program. Richard Nixon, wreaked the plan to extend the technology of Apollo and instead chose the Shuttle which was the direct design successor to those earlier reusable designs. But in the 80's the real disaster came when Ronald Reagan crippled the shuttle by cancelling the nuclear space tug, leaving the system restricted to low Earth orbit. The space tug system was a good idea, with a modular sensible design approach and a high level of adaptability - an approach that could still work today. Three space tugs locked together could form the core of a manned Mars transfer vehicle, and they would have enabled missions like satellite repair or building a Lunar colony.
No. The reason the US never returned to the Moon was for one reason - the cost of the Vietnam war. It was the war that was a huge waste of money, and that achieved little or nothing, and wasted 3 million human lives. If America hadn't faced that cost Astronauts would have walked on Mars by 1985. Those programs are only stepping stones to future exploration, but they also had a huge boosting effect on science, and they helped open up the whole of humanities future. If you cant see that you are mentally blind. Short term thinking - that's what the cow does.
".. They were right to keep it in the lab and not just start prototyping."
Only if you want to wait till 2050 for a working commercial plant. Single nation development - double efficiency, less bureaucracy, 50% final cost, 50% + faster. Its also much faster to modify a design during construction than to wait until every last detail has been worked out first to the last minutiae, especially as designs are likely to continue to evolve ad infinitum anyway. Three programs - Apollo, Manhattan project, V rocket program, all used techniques that compressed lead time. The irony is that by trying to spend money more quickly they actually ended up being cheaper in total.. An opposite example today, the James Web telescope - double original budget, why ? reducing spend rate, cutting interim budgets, and delaying final completion. They call it the 'dead hand' of bureaucracy for a reason.
Ur where did the root of the technology for Apollo come from? A lot of it came from the German V rocket scientists so was paid for by Hitler. Other tech was brought from the British, who also later also sold it on to the French to create the first Arianne rockets. Of course most of the actual heavy design work on the tech was purely American. Some like the flight computers was ultimately appropriated from the US ICBM programs.
The Russians basically brought or stole most of their later tech from the same sources. They didn't steal so much from the Americans which is why their moon program failed.
The equations of motion mean that light travels with infinite velocity. The resonance or curvature of space time limits this infinity to the speed we know.
(Momentum p = mv, so v = p/m, so for light v = h/0 . (h = any value except zero) )
How about a business form of over 400 pages telling you how they've simplified and made the process of filling in the form easier, and with enough levels of parenthesis and bureaucracy to drive even lawyers crazy... The EU has literally millions of published pages of rules and regulations and its dead weight is completely suffocating Europe.. Regulating the curvature of bananas is only one among many.. many over demanding rules... Insane no. Dead stupid and vastly over bureaucratic yes.
In the Victorian era the Conservatives were on the 'other' side. Taking money from the child prostitution industry they fended off age of consent laws for at least 30 years... When I learned it for a while I couldn't help calling them the 'Pedo Party'. (All done in the name of money - in my book that makes them worse than the pedophiles..)
Given other child sex scandals here in the UK its quite clear that there was a plot right across the establishment, at least up until a few years ago.. hundreds of children trafficked over decades - some murdered - and VIP pedophiles actively protected by the police. Brings a whole new meaning to 'Police Pedophile Unit'.. Then there was Rotherham.. police incompetence honed to an artistic peak ..or collusion..
It has nothing to do with housing status Chav is simply a particular kind of low level criminal.. basically what we might would call street scum. They are also a grouping a bit like Skinheads or Mods or Rockers or Goths with a particular fashion look. There are lots of different types of chav - from people who just dress in chav fashion, to pot heads and junkies, all the way to violent thugs and real criminals. Kind of like a white version of US rap culture.. Classic example of US Chav - Britney Spears.
BTW labelling them as 'council housed' suggests the author is on the UK far right - version of Tea Party, Murdock newspaper reader..
"We should should worry about overpopulation from pinhead-dancing angels too."
Human population is 7.3 billion and increasing by 80 million per year and rising. (Birth rate 380,000 per day, death rate 160,000 per day.) Even if population does level off at about 10 billion that's still an awful lot of people. Today about half the total population live in stone age level poverty and regularly go hungry and the world is already under pretty heavy food stress. Today the global gap between poor and rich is slowly equalising and the poorest are slowly starting to rise out of poverty- but this is set to hugely increase the global demand for food and resources. (by about 1/4 to 3/8) Disaster is not certain, but not worrying about over-population and over-stress on Earths resources and eco-system is insane.
I'm working on Strong AI and the design I am working on will probably have either 8 or 12 primary CPU's. The overall capacity might not be far different to a PC with a GPU but the circuits need to be laid out very differently. My design will be built on custom special purpose CPU designs built on FPGA chips. The first four are the core system and will run as 4 way redundant. The rest will (mostly) be visual processing CPUs, divided between front end and back end. This roughly fits with the human brain where 1/3 to 1/2 the total volume is taken up with visual processing reasoning. So no it will not quite run on a GPU but almost..
Great quote cargo cult science is a great name for it. I know - am working on (against) what has to be one of the biggest examples in the whole of science - general relativity. In short general relativity has a really stupid flaw that makes the whole theory a complete nonsense, but this flaw only really appears in the FTL part of its geometry. One fairly simple way of describing the flaw - The speed of light C isn't just a speed but also has a direction (is a vector). Put light as vector into the system and space time fits into three dimensions instead of four, it turns out that time is point like and not an actual dimension. - Space time was actually 'borrowing' the dimension from the spatial direction of light or other relativistic objects. Dimensional time still exists but only as an abstraction or on quantum scales. There are many physicists who know about this and other problems with general relativity, but there is an extremely aggressive establishment that cannot tolerate any substantive criticism of the theory. .
"... The main problem with an in vivo approach would be targeting specific locations or cells rather than miniaturization. How do you get the nanobots to go only where you want them to go?"
Only one of about a dozen critical problems with using real nanobots. The most obvious of course is that 'nanobots' don't actually yet exist.. For a brief period I worked on assemblers way back around 1991, they were about 10 to 20 years away then - and they are still about 10 to 20 years away today..
To build an assembler for real the starting price is at least 1 to 2 billion $ and you also need working Strong AI tech.. and that 10 years.. Then add in another 10 to 20 years to get the things up to speed as a medical technology.:) The main danger is dying od old age first..
Plutonium RTG's are safe enough to use inside the human body. Unfortunately they are very expensive to make & very difficult to licence. For a drone a bigger problem is that the energy output per unit weight is to low to actually power a flying machine.. It would be possible using a small RTG to charge a larger secondary Lithium Ion battery to fly the drone - but pretty pointless. (Cycle : sits for 10 hours charging then fly's for 10 minutes.)
You forgot two other options the system leaves men.. 4. choose gay - solves all the problems with women. or 5. catch a woman and keep her locked up in your basement for 20 years.
Sigh. To 'fire' something into the sun you need to decelerate it by 29 km/s to take it out of Earths orbit, then you need to accelerate it towards the star. A pretty expensive trajectory.
BTW anyone calling people at random 'space nutters' must be a cretin, probably humping their sister or Mum.
Who cares if they don't work. Lets build another ten thousand. Its worth it for the subsidy alone.. Wind turbines are the oil industries solution to the threat of 'carbon free energy' and will keep them in business for another 100 years.
Slashdot Mirror
The problem isn't government though - its Americas government.. Or Europe's or UK's government or.. Anyway efficiency and competence are the most important thing - whether its private, public, socialist, liberal, Republican or whatever..
In the US the biggest problems include corporate socialism, wealth and social inequality, the slush of money the richest 0.01% bring to the table, and basic political stupidity..
In Europe the biggest problems are things like lowest bid tendering, privatisation, the single currency, the open single market, corporate socialism, excessive bureaucracy, and basic political stupidity..
All we need is intelligent, competent, efficient, compassionate, moral, and strong government. Maybe we need a new "Moon Program" to replace our politicians with something better - like machines. (Disclaimer : I am working on developing such machines)
You are absolutely right, Strong AI is non-deterministic, well not strictly deterministic. In some ways its quite 'un-computer-like' and although Strong AI could theoretically be run on ordinary hardware in practice this is unlikely to work. One of the reasons is basic low level memory management, which has long been one of the biggest stumbling points of strong AI. The machine needs a persistent and stable memory that is constantly online and this is surprisingly difficult to achieve. (heaps and lists and stacks are not reliable enough)
My early attempts based on C++ tried to build a system of 'encapsulation' using doubly recursive pointers, but in the end it kept becoming a tangled mess of spaghetti and just wouldn't work... The language simply wasn't designed to do it. C++ has plenty of encapsulation on the source code text side but on the run-time side there is basically none at all.
I tried writing the core in machine code but that didn't get very far either. - The real problem is in the way the memory manager and CPU work together - controlled by the operating system. I needed total control of this and couldn't get it - the thing is also enormously complicated and poorly documented. Its not a place OS and hardware vendors want anyone looking.
The basic problem though was that I was trying to get the machine to do something it really wasn't designed to do. I came to the conclusion again and again that building a custom CPU was the only answer, and eventually gave up on it as an impossible problem.. (I have considered the alternative of building a virtual machine solution but this would still have similar problems)
Ten years later and I discovered people are building systems using custom CPU's built in FPGA's. The nightmare now is learning enough Verilog and digital electronics to get the thing working... The biggest problem is that it requires more than one CPU and more than one FPGA, and this ultimately means building a custom motherboard. Fortunately electronics design software is now advanced enough to do this with relatively little work (with a couple of electronics engineers).. The total budget is probably less than $100,000 to $200,000.
At a pinch the prototype can be built using FPGA dev boards putting it within the reach of one person and maybe $10,000.. plus of course another 10 years...
I have worked on and off on Strong AI for 25 years... 20 years ago (in 1995) I had a workable design for building a working machine but I calculated that it would need a budget of about $100 million.. In 2003 the base cost was about the same - though the final machine would fit inside a PC case instead of multiple filing cabinets.
A couple of years ago I restarted work on the thing, because technology has advanced. Now the whole machine should fit inside a 1 foot cube, but now the minimum base dev costs have shrunk to about $10,000.. (based on FPGA) Its still a 10 year project because it requires a special purpose computer. It turns out to be 'easier' to build a complete new architecture from the ground up than to reverse engineer an OS like Linux to make it compatible with strong AI, the machine needs custom hardware to work anyway...
Of course the machine will still need many millions to go from a prototype to a production machine. Strong AI has some weird problems with security and that makes everything so much more difficult - it will probably end up as being classified as both military and civilian and being quite unlike anything we have seen before.. And yes Strong AI will eventually be a multi-trillion dollar global market. .. but try telling an investor that their trillion dollar return is twenty or thirty years away.. 'High risk' doesn't even begin to cover it.. : )
NASA isn't the key point, the V rocket program was. Without Hitler's investment in the V rocket program we might not be in orbit even today.. The whole problem was that rockets weren't taken seriously and might never have received the investment they needed - the V program took them from being a toy to sophisticated machines able to reach supersonic speeds and touch the edge of orbit.
Once you can cross the coulomb barrier and can stay there, your not far from getting all the toys - like total atomic conversion, bending space and time - gravity engines, contained singularities, force fields, manipulating energy and entropy. From there going for fusion looks a little - 'primitive'.
Of course crossing that barrier and staying there also looks like one of the hardest things imaginable - a contained quark gluon plasma.. I like working on left-field 'impossible' problems.. :)
There is a much simpler way to stop them if you want to and its instant. In any decent bank account with online control there should be a section in the settings for allowed payees, all you have to do is remove PayPal from that list and it stops them from taking money instantly. I did this about four years ago and now only use PayPal through my credit cards - it all works perfectly, is instant and if things go wrong mostly easy to fix...
If you smash open your skull you will find some shit inside. You can use it to fertilize the garden. : ) Seriously though if you really want to campaign against nuclear power then you should campaign against the one source that really does kill people in large numbers - the sun. (est 500,000+ per year)
Radiation from Fukushima has probably killed less than ~ 10 people. However Fukushima has killed something like 2,000 to maybe 10,000 people - because Japan switched its nuclear stations off and switched to coal and oil - which are literally 10,000 time more dangerous than nuclear.. Real problem, nuclear is scary to simple people.
The real problem is that all real research into nuclear power has all but stopped largely thanks to the anti-nuclear lobby. Given that pollution from burning coal & oil has killed something like 70 million to over 100 million people since 1945 the safety bar for nuclear is set insanely high.. With money more efficient fuel cycles and technologies could b developed.
Of course with say $10-20 billion and a more efficient bureaucracy the development time to working commercial fusion plants could be reduced from 40+ years to something like 10 or 20 years. (using a technique like parallel development) Fusion would probably render all fission systems obsolete..
But not very stable.. :))
A muon is like an electron but is type II matter so is unstable.. Could work great in an accelerator.
Sorry but the USA Today article is scientifically illiterate. I know Von Braun's early designs quite well and they were created before he had the experience of Apollo. (he was also a primary designer within Apollo) The problem with those early designs was that they were too large and were hopelessly cumbersome and expensive - try (at least) 10 to 20 times the final cost of the actual program that was built.
I'm actually a fan of using nuclear powered reusable orbital lifters but they were definitely a step too far in 1960, and would be a complex technology to develop even today. Look at the Skylon program in the UK - with tech well beyond the 1960's, 70's, or 80's. Apollo was exactly the right tech at the right time, it was the political cowardice that came later that screwed up the space program.
Richard Nixon, wreaked the plan to extend the technology of Apollo and instead chose the Shuttle which was the direct design successor to those earlier reusable designs. But in the 80's the real disaster came when Ronald Reagan crippled the shuttle by cancelling the nuclear space tug, leaving the system restricted to low Earth orbit.
The space tug system was a good idea, with a modular sensible design approach and a high level of adaptability - an approach that could still work today. Three space tugs locked together could form the core of a manned Mars transfer vehicle, and they would have enabled missions like satellite repair or building a Lunar colony.
No. The reason the US never returned to the Moon was for one reason - the cost of the Vietnam war. It was the war that was a huge waste of money, and that achieved little or nothing, and wasted 3 million human lives. If America hadn't faced that cost Astronauts would have walked on Mars by 1985. Those programs are only stepping stones to future exploration, but they also had a huge boosting effect on science, and they helped open up the whole of humanities future. If you cant see that you are mentally blind. Short term thinking - that's what the cow does.
".. They were right to keep it in the lab and not just start prototyping."
Only if you want to wait till 2050 for a working commercial plant. Single nation development - double efficiency, less bureaucracy, 50% final cost, 50% + faster. Its also much faster to modify a design during construction than to wait until every last detail has been worked out first to the last minutiae, especially as designs are likely to continue to evolve ad infinitum anyway. Three programs - Apollo, Manhattan project, V rocket program, all used techniques that compressed lead time. The irony is that by trying to spend money more quickly they actually ended up being cheaper in total..
An opposite example today, the James Web telescope - double original budget, why ? reducing spend rate, cutting interim budgets, and delaying final completion. They call it the 'dead hand' of bureaucracy for a reason.
Ur where did the root of the technology for Apollo come from? A lot of it came from the German V rocket scientists so was paid for by Hitler. Other tech was brought from the British, who also later also sold it on to the French to create the first Arianne rockets. Of course most of the actual heavy design work on the tech was purely American. Some like the flight computers was ultimately appropriated from the US ICBM programs.
The Russians basically brought or stole most of their later tech from the same sources. They didn't steal so much from the Americans which is why their moon program failed.
The equations of motion mean that light travels with infinite velocity. The resonance or curvature of space time limits this infinity to the speed we know.
(Momentum p = mv, so v = p/m, so for light v = h/0 . (h = any value except zero) )
How about a business form of over 400 pages telling you how they've simplified and made the process of filling in the form easier, and with enough levels of parenthesis and bureaucracy to drive even lawyers crazy... The EU has literally millions of published pages of rules and regulations and its dead weight is completely suffocating Europe..
Regulating the curvature of bananas is only one among many.. many over demanding rules... Insane no. Dead stupid and vastly over bureaucratic yes.
In the Victorian era the Conservatives were on the 'other' side. Taking money from the child prostitution industry they fended off age of consent laws for at least 30 years... When I learned it for a while I couldn't help calling them the 'Pedo Party'. (All done in the name of money - in my book that makes them worse than the pedophiles..)
Given other child sex scandals here in the UK its quite clear that there was a plot right across the establishment, at least up until a few years ago.. hundreds of children trafficked over decades - some murdered - and VIP pedophiles actively protected by the police. Brings a whole new meaning to 'Police Pedophile Unit'.. Then there was Rotherham .. police incompetence honed to an artistic peak . .or collusion..
It has nothing to do with housing status Chav is simply a particular kind of low level criminal.. basically what we might would call street scum. They are also a grouping a bit like Skinheads or Mods or Rockers or Goths with a particular fashion look. There are lots of different types of chav - from people who just dress in chav fashion, to pot heads and junkies, all the way to violent thugs and real criminals. Kind of like a white version of US rap culture.. Classic example of US Chav - Britney Spears.
BTW labelling them as 'council housed' suggests the author is on the UK far right - version of Tea Party, Murdock newspaper reader..
"We should should worry about overpopulation from pinhead-dancing angels too."
Human population is 7.3 billion and increasing by 80 million per year and rising. (Birth rate 380,000 per day, death rate 160,000 per day.) Even if population does level off at about 10 billion that's still an awful lot of people. Today about half the total population live in stone age level poverty and regularly go hungry and the world is already under pretty heavy food stress. Today the global gap between poor and rich is slowly equalising and the poorest are slowly starting to rise out of poverty- but this is set to hugely increase the global demand for food and resources. (by about 1/4 to 3/8) Disaster is not certain, but not worrying about over-population and over-stress on Earths resources and eco-system is insane.
I'm working on Strong AI and the design I am working on will probably have either 8 or 12 primary CPU's. The overall capacity might not be far different to a PC with a GPU but the circuits need to be laid out very differently.
My design will be built on custom special purpose CPU designs built on FPGA chips. The first four are the core system and will run as 4 way redundant. The rest will (mostly) be visual processing CPUs, divided between front end and back end. This roughly fits with the human brain where 1/3 to 1/2 the total volume is taken up with visual processing reasoning.
So no it will not quite run on a GPU but almost..
Great quote cargo cult science is a great name for it. I know - am working on (against) what has to be one of the biggest examples in the whole of science - general relativity. In short general relativity has a really stupid flaw that makes the whole theory a complete nonsense, but this flaw only really appears in the FTL part of its geometry.
One fairly simple way of describing the flaw - The speed of light C isn't just a speed but also has a direction (is a vector). Put light as vector into the system and space time fits into three dimensions instead of four, it turns out that time is point like and not an actual dimension. - Space time was actually 'borrowing' the dimension from the spatial direction of light or other relativistic objects. Dimensional time still exists but only as an abstraction or on quantum scales. There are many physicists who know about this and other problems with general relativity, but there is an extremely aggressive establishment that cannot tolerate any substantive criticism of the theory. .
"... The main problem with an in vivo approach would be targeting specific locations or cells rather than miniaturization. How do you get the nanobots to go only where you want them to go?"
Only one of about a dozen critical problems with using real nanobots. The most obvious of course is that 'nanobots' don't actually yet exist.. For a brief period I worked on assemblers way back around 1991, they were about 10 to 20 years away then - and they are still about 10 to 20 years away today..
To build an assembler for real the starting price is at least 1 to 2 billion $ and you also need working Strong AI tech.. and that 10 years.. Then add in another 10 to 20 years to get the things up to speed as a medical technology. :) The main danger is dying od old age first..
Plutonium RTG's are safe enough to use inside the human body. Unfortunately they are very expensive to make & very difficult to licence. For a drone a bigger problem is that the energy output per unit weight is to low to actually power a flying machine.. It would be possible using a small RTG to charge a larger secondary Lithium Ion battery to fly the drone - but pretty pointless. (Cycle : sits for 10 hours charging then fly's for 10 minutes.)
You forgot two other options the system leaves men.. 4. choose gay - solves all the problems with women. or 5. catch a woman and keep her locked up in your basement for 20 years.
Sigh. To 'fire' something into the sun you need to decelerate it by 29 km/s to take it out of Earths orbit, then you need to accelerate it towards the star. A pretty expensive trajectory.
BTW anyone calling people at random 'space nutters' must be a cretin, probably humping their sister or Mum.
Who cares if they don't work. Lets build another ten thousand. Its worth it for the subsidy alone.. Wind turbines are the oil industries solution to the threat of 'carbon free energy' and will keep them in business for another 100 years.