A better architecture would have the management engine off the main processor: leave the CPU for running user processes. Consider a small ARM processor running an RTOS, and if that did much of the work the kernel does. A modern PC is already like a mini heterogeneous supercomputer. Put the GUI on a core on the GPU card, so the CPU can send basic instructions and data (what they're moving towards anyway, but stick the windowing system the other end of the pcie bus, next to the GPU, possibly in the screen. Do likewise for io, network, and audio. Economies of scale is how the price is kept low. The trend has been to lump more and more work on faster and faster CPUs. Now that CPUs are no longer increasing in single core speed as they once did, performance gains will come from not using a 95W intel chip for what can be done just as well on a 1W ARM or MIPs chip. As for verifying security, use public key crypto, signiatures and hashing to verify what us run (optionally), and separate OS install and normal runtime at a hardware level: have a small SSD for the core OS files (2GB would be plenty for a number of OSs side by side). In OS install mode, you can copy the files across, and do hardware diagnostics etc., but not run normal user software like Firefox. In normal run mode, the switch engages a hardware write blocker to the small OS SSD, and in signed mode the image needs to be signed by the hardware vendor somehow. Then malware cannot corrupt your OS as happens so often.
Digital copying is easy, sharing with friends is natural and human. The media industry is built on the principle of taking away our ability to copy and share, and on the idea that it is hard to do so. What would be rightfully ours under the original copyright laws is no longer ours, what we would have the right to do in the absence of copyright laws we no longer have the right to do. As for youtube, it is built on the hard work of those who invented the hardware and software technology to make it possible, and the efforts of many users. A little copyright infringment happens as 'collateral damage', and that is largely because copyright at present is so distant from what is natural, easy and straightforward. We could function without Trent's last album quite happily, but the ability to share information, events and enthusiasm is so much more important.
Suppose we took driving as an example, saying 'driving snobs who examine our driving students are not helping prepare them for the future, by insisting they know the basics of the highway code and car maintenance'. The result is a generation with terrible habits, like the difference between how people drive in the UK (which is bad enough) and how they drive in India. The latter with code would result in a mass proliferation of lines of code which simply do not f***ing work. It is bad enough at present with the proprietary priests in charge. Software needs to be simple, well designed, well thought through, reliable, and just f***ing work. If not, it is better not written. If that means fewer kids 'writing code' then that is not a bad thing, compared to millions of kids writing bad code, knowing no better, and being taught that those with high ideals of code which actually works properly are just 'backward luddites of an obsolete priesthood'.
Having a maths background, the ability to have x and X refer to different things is natural, useful, and the inability can be annoying. I haven't read the article, but essentially what I imagine is that, given a mismatch when checking exactly (that is, hash and compare to stored hash), it is then easy for the server to calculate say 1000 likely variants assuming the presence of a typographical error, and try those. The insecurity is mitigated by making your minimum length 2-3 characters more.
Something most people, scientific or otherwise, fail to do when reasoning from evidence is to consider all possible interpretations which are consistent with that evidence, right down to underlying metaphysical assumptions. Most working scientists do not have time to think such things through, of course. But to think about what, given what we know (rather than what we have inferred, tested and assume true), could be true is important: or put another way, knowing what we cannot rule out, but must assume for practical purposes. The silliest bit of TFA is the idea that you can reliably compute probabilities of us living in a simulation given only what we can empirically observe. Unless we see a smoking gun indicating that we are certainly in a simulation, we simply cannot know. Naturally this does not fall within the realms of empirical science (since we cannot empirically test hypotheses), but assuming things for convenience and then defending your assumptions by claiming that alternatives are unfalsifiable is a bad debating habit best consigned to the dustbin. You can find similar things turning up in the foundations of mathematics in things like set theory axioms (e.g. what can and cannot be proved with the Axiom of Choice, or by assuming the generalised continuum hypothesis, or assuming the Riemann hypothesis). In such situations, people will make an assumption and proceed. The fool, however, forgets he/she made that assumption and then kids themselves with the apparent rigour of the argument they constructed from it.
If we can explain that there _is_ an outside, but that we have no means to probe it, that is something, and if we can explain why the contrary is not consistent with empirical evidence (assuming one could do such a thing), then that too is something. It does not give predictive power in the way many theories do, but it does rule out the possibility of certain theories working (in the way that Bell's Theorem rules out some physical theories).
If you think about how a blu ray player works, for example, turning the video into raw pixels then sending over HDMI is actually quite stupid. Rather, if your display can decompress h264 in hardware, you can just stream the raw h264. With a few decent royalty free standards things could work so much better, albeit against a number of entrenched proprietary interests (which is why I don't hold my breath).
The sensible thing is to have the GPU, or at least some of the GPU, in the display, along with a lightweight processor so that the display alone can run a decent GUI. Essentially something like the Quartz display server, but with a few extras. First, you can set up surfaces in the display GPU and stream content to them via either HDMI/VGA/DVI interfaces etc. and over a gigabit ethernet cable, possibly even wifi. You need a programmable API for how the GUI actually works, etc. Basically, you end up reinventing X11, but with a decent drawing API which is built around the capabilities of a modern GPU. I would do likewise in the main machine, having a small ARM chip handling IO tasks, and bringing up the main processors. Move certain parts of security onto these devices, and you can have a setup where rogue software running on the main processors can do little harm. In addition, the main processors can be used totally for running user programs, and can be powered down if necessary, leaving something with the power of a raspberry pi for basic IO and device management. If the GUI is running in the external display, this lightweight io processor needs only send basic instructions to the display to power a reasonable gui interface, and do stuff like the ILO features you find on rack servers. Given how cheap these small chips are, it is IMHO quite stupid to have the main processors doing the housekeeping tasks they do on current machines.
The basic rule should be that interfaces are not copyrightable, but implementations are. If the contents of/usr/include can be treated as not subject to the GPL, that is no problem. If the 'commonly available' bit applies as the op-ed mentions, then would this not apply at least to all 'freeware'? And to the space filler that comes on a new laptop's hard drive? Copyright needs fixing anyway.
To put my grammar nazi hat on, it is not incorrect to refer to a female using 'he', but it is incorrect to refer to a male using 'she'. Women are special in the language because they can have babies whilst males cannot, and this little fact, sexist or otherwise, is critical to keeping the human race going. Something like that.
Have a facebook account for business stuff, don't use it for personal stuff. Basically when you're getti g f@#$ed by facebook, make sure to wear a computational condom.
If you do, on an older machine, you may well get driver issues and software incompatibilities. I have a number of laptops from circa 2010-2012, and Windows 10 is basically unusable on them. I am gradually purchasing SSDs and reinstalling Windows 7 on them (may as well do the SSD upgrade while I'm at it). I have a friend who's laptop failed to do the update properly and was left in an unbootable state, at which point he was talking about taking it to a repair man to get the data off and buying a new laptop. That is the kind of nightmare the non-tech-savvy are put into. As for security problems, if Microsoft really cared about security, they would do a better job of designing Windows. They intentionally put in may hidden cubbyholes for proprietary software and oem customisations to take advantage of, and much malware thrives on this sort of thing. If companies like MS made things as simple as possible, security would be much more straightforward.
1. Port Windows and Mac OS X to run on top of a Linux kernel (and merging the best bits of each) so that all have a shared foundation. 2. Allowing apps for each platform (and you may as well do likewise with Android etc.) to send objects via simple IPC and shared memory (data structures like those you see in clojure, are a good idea here) basically an improvement upon the CLR idea MS has. 3. Open sourcing much of those foundations. 4. Putting ARM cores like those on smartphones onto the GPU, and running most of the GUI on the GPU. This is a throwback to X11, but based around modern GPUs with the pcie backplane being the network. 5. Having the GUI frontend be a separate small OS running on said ARM cores, which both runs the GUI, sound and such, and brings up the main processors, which are then freed up for the general purpose processing tasks they are best at. 6. Moving away from binary code to higher level code (android runtime sort of illustrates this) which can be comipiled either AOT or JIT when loaded onto a system. 7. Using dynamic compilation for both performance and security purposes (this entail rethinking the syscall interface, so that a process can only access the syscalls it needs: something akin to capability security, which can be achieved via the AOT and JIT compilation so that a process is limited in what syscalls it can make: do not allow processes to create executable binary code without explicit permission, and so on. This would make reverse engineering much easier, which is why things probably aren't heading this way, but Free Software would not suffer in the same way. (The thesis on the Synthesis OS, from quite a few years back, is worth a quick perusal.) 8. Do likewise as the coprocessed GUI for sound, and synchronosed sound and graphics, and IO (rather than taxing the main processors with the overhead of USB, having a small ARM core or similar doing this would get us back some of the advantages Firewire traditionally had over USB. The thing is, small ARM cores (or similar) as we find in mobile phones, raspberry pis and so on, can be cheaply added to e.g. a GPU, and since main processors (intel and amd) are hitting a wall with single core performance, it is sensible to start offloading to coprocessors as we had to do in the old days. But these days a small ARM core together with a specialised processor would be the way to go. Making it ubiquitous would lead to economies of scale (provided patent nightmares don't rear their ugly head as they tend to). Having done the above, OS architecture would need a bit of chanigng.
The question I want to answer, for a given movie, is whether I will like it, not some abstract average viewer. What IMDB should do is: 1. Require users to have accounts to vote 2. Analyse correlations between votes for various movies, for a given user 3. When rating movies for me, produce a number of ratings based on groupings from the analysis 4. Allow me to select a few movies I like and don't like to seed the process. The data gathered would be valuable (so the effort could pay for itself). Further, if there is a large male/female discrepancy, for example, the rating system can point this out.
Firstly, that a digital computer contained within this universe can accurately replicate the behaviour of a brain in real time, let alone the behaviour of the brain coupled with its body.
Secondly, even if one comes up with a passable approximation, I remain to be convinced that my conscious experience will be transferred into the digital version so that the 'digital me' will not be a simulated prediction of who I am. The 'experts' tend to handwave around the difficult parts of matters like this, saying that it's just a matter of having enough instructions per second without any justification that there is an 'enough ips' or that such an 'enough ips' is physically feasible. (For where I am coming from with the latter bit, the horrors of Ramsey Theory come to mind.)
I only use Windows for running Windows only software, like VSTs and DAWs; only use my mac for running music software. All else I use Linux. Having to use the mouse to launch my main apps, rather than a two-key combo is a real annoyance when going back to them. Likewise, not being able to write short scripts with ease is again an annoyance. Not being able to install to a USB stick, taking it from machine to machine without activation problems is another annoyance.
With xfce, for common apps, the procedure: 1. Think of what key combination is most obvious 2. If it is not used, bind the app to that key combo 3. If it is used, add a modifier key, else choose next most obvious Super-T for terminal, Super-B for browser, Super-E for file manager (inherited from Win95), Super-hash for virt-manager or x2goclient, and so on.
In the US, you have weakly regulated free market capitalism where companies are charged with making profits for their shareholders, and some praise this system... then you complain about Apple exploiting tax regulations so as to improve its profitability to its shareholders.
Capitalism stands on self-organisation and enlightened self-interest. When there was the threat of the USSR, this kept greed and short-sightedness in check. In the aftermath of the Cold War, however, that is giving way to unrestrained self-interest. Competition is good to a point, but expending resources to compete rather than produce, and to alter one's strategy away from the overall common good is what is happening. I imagine game theorists saw it coming to a certain degree.
A major problem is one of semantics. Does the answer to 'do we live in a computer simulation?' depend on what exactly we mean by 'computer simulation'? If not, why not (has anybody made any kind of coherent case for this?), and if so, who is bothering to give sufficiently precise detail about what they mean that one can give the question a meaningful answer. I tend to lean towards the 'yes' position, but I tend to leave what is meant by 'computer simulation' sufficiently flexible as to be able to fudge things later. One can always fudge a 'no' answer by taking the 'that's not a computer simulation' objection and repeating it ad nauseam like a broken record.
A universal tradeable resource token is what money is. In earlier parts of civilisation things were simpler, and a money-trade system of resource organisation probably worked reasonably. The modern world has grown complex to a degree that organising resources using money no longer works well. There are too many middlemen, and too many ways for 'value' to leak out in the money system. With regards to the basic resources needed for living, there are surely better ways of organising the human and material resources we have.
In the war between facts and dogma, facts have a habit of coming second. Facts are hard to think through and analyse properly, and proper analyses are detailed and tough to understand. Dogma doesn't have any of these drawbacks.
But it is well known that the momentum of a photon depends on its frequency. If the device spits out microwaves in two directions, the same number of photons per second, but the microwaves come out in one direction are of a higher frequency than the other, then more momentum will be emitted in photons in one direction than the other. I would expect the quantities involved to be miniscule, but that's the first guess I had in mind when I read about this.
Slashdot Mirror
A better architecture would have the management engine off the main processor: leave the CPU for running user processes. Consider a small ARM processor running an RTOS, and if that did much of the work the kernel does. A modern PC is already like a mini heterogeneous supercomputer. Put the GUI on a core on the GPU card, so the CPU can send basic instructions and data (what they're moving towards anyway, but stick the windowing system the other end of the pcie bus, next to the GPU, possibly in the screen. Do likewise for io, network, and audio. Economies of scale is how the price is kept low. The trend has been to lump more and more work on faster and faster CPUs. Now that CPUs are no longer increasing in single core speed as they once did, performance gains will come from not using a 95W intel chip for what can be done just as well on a 1W ARM or MIPs chip. As for verifying security, use public key crypto, signiatures and hashing
to verify what us run (optionally), and separate OS install and normal runtime at a hardware level: have a small SSD for the core OS files (2GB would be plenty for a number of OSs side by side). In OS install mode, you can copy the files across, and do hardware diagnostics etc., but not run normal user software like Firefox. In normal run mode, the switch engages a hardware write blocker to the small OS SSD, and in signed mode the image needs to be signed by the hardware vendor somehow. Then malware cannot corrupt your OS as happens so often.
Digital copying is easy, sharing with friends is natural and human. The media industry is built on the principle of taking away our ability to copy and share, and on the idea that it is hard to do so. What would be rightfully ours under the original copyright laws is no longer ours, what we would have the right to do in the absence of copyright laws we no longer have the right to do. As for youtube, it is built on the hard work of those who invented the hardware and software technology to make it possible, and the efforts of many users. A little copyright infringment happens as 'collateral damage', and that is largely because copyright at present is so distant from what is natural, easy and straightforward. We could function without Trent's last album quite happily, but the ability to share information, events and enthusiasm is so much more important.
Suppose we took driving as an example, saying 'driving snobs who examine our driving students are not helping prepare them for the future, by insisting they know the basics of the highway code and car maintenance'. The result is a generation with terrible habits, like the difference between how people drive in the UK (which is bad enough) and how they drive in India. The latter with code would result in a mass proliferation of lines of code which simply do not f***ing work. It is bad enough at present with the proprietary priests in charge. Software needs to be simple, well designed, well thought through, reliable, and just f***ing work. If not, it is better not written. If that means fewer kids 'writing code' then that is not a bad thing, compared to millions of kids writing bad code, knowing no better, and being taught that those with high ideals of code which actually works properly are just 'backward luddites of an obsolete priesthood'.
Having a maths background, the ability to have x and X refer to different things is natural, useful, and the inability can be annoying. I haven't read the article, but essentially what I imagine is that, given a mismatch when checking exactly (that is, hash and compare to stored hash), it is then easy for the server to calculate say 1000 likely variants assuming the presence of a typographical error, and try those. The insecurity is mitigated by making your minimum length 2-3 characters more.
Something most people, scientific or otherwise, fail to do when reasoning from evidence is to consider all possible interpretations which are consistent with that evidence, right down to underlying metaphysical assumptions. Most working scientists do not have time to think such things through, of course. But to think about what, given what we know (rather than what we have inferred, tested and assume true), could be true is important: or put another way, knowing what we cannot rule out, but must assume for practical purposes. The silliest bit of TFA is the idea that you can reliably compute probabilities of us living in a simulation given only what we can empirically observe. Unless we see a smoking gun indicating that we are certainly in a simulation, we simply cannot know. Naturally this does not fall within the realms of empirical science (since we cannot empirically test hypotheses), but assuming things for convenience and then defending your assumptions by claiming that alternatives are unfalsifiable is a bad debating habit best consigned to the dustbin. You can find similar things turning up in the foundations of mathematics in things like set theory axioms (e.g. what can and cannot be proved with the Axiom of Choice, or by assuming the generalised continuum hypothesis, or assuming the Riemann hypothesis). In such situations, people will make an assumption and proceed. The fool, however, forgets he/she made that assumption and then kids themselves with the apparent rigour of the argument they constructed from it.
If we can explain that there _is_ an outside, but that we have no means to probe it, that is something, and if we can explain why the contrary is not consistent with empirical evidence (assuming one could do such a thing), then that too is something. It does not give predictive power in the way many theories do, but it does rule out the possibility of certain theories working (in the way that Bell's Theorem rules out some physical theories).
If you think about how a blu ray player works, for example, turning the video into raw pixels then sending over HDMI is actually quite stupid. Rather, if your display can decompress h264 in hardware, you can just stream the raw h264. With a few decent royalty free standards things could work so much better, albeit against a number of entrenched proprietary interests (which is why I don't hold my breath).
The sensible thing is to have the GPU, or at least some of the GPU, in the display, along with a lightweight processor so that the display alone can run a decent GUI. Essentially something like the Quartz display server, but with a few extras. First, you can set up surfaces in the display GPU and stream content to them via either HDMI/VGA/DVI interfaces etc. and over a gigabit ethernet cable, possibly even wifi. You need a programmable API for how the GUI actually works, etc. Basically, you end up reinventing X11, but with a decent drawing API which is built around the capabilities of a modern GPU. I would do likewise in the main machine, having a small ARM chip handling IO tasks, and bringing up the main processors. Move certain parts of security onto these devices, and you can have a setup where rogue software running on the main processors can do little harm. In addition, the main processors can be used totally for running user programs, and can be powered down if necessary, leaving something with the power of a raspberry pi for basic IO and device management. If the GUI is running in the external display, this lightweight io processor needs only send basic instructions to the display to power a reasonable gui interface, and do stuff like the ILO features you find on rack servers. Given how cheap these small chips are, it is IMHO quite stupid to have the main processors doing the housekeeping tasks they do on current machines.
The basic rule should be that interfaces are not copyrightable, but implementations are. If the contents of /usr/include can be treated as not subject to the GPL, that is no problem. If the 'commonly available' bit applies as the op-ed mentions, then would this not apply at least to all 'freeware'? And to the space filler that comes on a new laptop's hard drive? Copyright needs fixing anyway.
To put my grammar nazi hat on, it is not incorrect to refer to a female using 'he', but it is incorrect to refer to a male using 'she'. Women are special in the language because they can have babies whilst males cannot, and this little fact, sexist or otherwise, is critical to keeping the human race going. Something like that.
Have a facebook account for business stuff, don't use it for personal stuff. Basically when you're getti g f@#$ed by facebook, make sure to wear a computational condom.
If you do, on an older machine, you may well get driver issues and software incompatibilities. I have a number of laptops from circa 2010-2012, and Windows 10 is basically unusable on them. I am gradually purchasing SSDs and reinstalling Windows 7 on them (may as well do the SSD upgrade while I'm at it). I have a friend who's laptop failed to do the update properly and was left in an unbootable state, at which point he was talking about taking it to a repair man to get the data off and buying a new laptop. That is the kind of nightmare the non-tech-savvy are put into. As for security problems, if Microsoft really cared about security, they would do a better job of designing Windows. They intentionally put in may hidden cubbyholes for proprietary software and oem customisations to take advantage of, and much malware thrives on this sort of thing. If companies like MS made things as simple as possible, security would be much more straightforward.
1. Port Windows and Mac OS X to run on top of a Linux kernel (and merging the best bits of each) so that all have a shared foundation.
2. Allowing apps for each platform (and you may as well do likewise with Android etc.) to send objects via simple IPC and shared memory (data structures like those you see in clojure, are a good idea here) basically an improvement upon the CLR idea MS has.
3. Open sourcing much of those foundations.
4. Putting ARM cores like those on smartphones onto the GPU, and running most of the GUI on the GPU. This is a throwback to X11, but based around modern GPUs with the pcie backplane being the network.
5. Having the GUI frontend be a separate small OS running on said ARM cores, which both runs the GUI, sound and such, and brings up the main processors, which are then freed up for the general purpose processing tasks they are best at.
6. Moving away from binary code to higher level code (android runtime sort of illustrates this) which can be comipiled either AOT or JIT when loaded onto a system.
7. Using dynamic compilation for both performance and security purposes (this entail rethinking the syscall interface, so that a process can only access the syscalls it needs: something akin to capability security, which can be achieved via the AOT and JIT compilation so that a process is limited in what syscalls it can make: do not allow processes to create executable binary code without explicit permission, and so on. This would make reverse engineering much easier, which is why things probably aren't heading this way, but Free Software would not suffer in the same way. (The thesis on the Synthesis OS, from quite a few years back, is worth a quick perusal.)
8. Do likewise as the coprocessed GUI for sound, and synchronosed sound and graphics, and IO (rather than taxing the main processors with the overhead of USB, having a small ARM core or similar doing this would get us back some of the advantages Firewire traditionally had over USB.
The thing is, small ARM cores (or similar) as we find in mobile phones, raspberry pis and so on, can be cheaply added to e.g. a GPU, and since main processors (intel and amd) are hitting a wall with single core performance, it is sensible to start offloading to coprocessors as we had to do in the old days. But these days a small ARM core together with a specialised processor would be the way to go. Making it ubiquitous would lead to economies of scale (provided patent nightmares don't rear their ugly head as they tend to).
Having done the above, OS architecture would need a bit of chanigng.
The question I want to answer, for a given movie, is whether I will like it, not some abstract average viewer.
What IMDB should do is:
1. Require users to have accounts to vote
2. Analyse correlations between votes for various movies, for a given user
3. When rating movies for me, produce a number of ratings based on groupings from the analysis
4. Allow me to select a few movies I like and don't like to seed the process.
The data gathered would be valuable (so the effort could pay for itself).
Further, if there is a large male/female discrepancy, for example, the rating system can point this out.
Firstly, that a digital computer contained within this universe can accurately replicate the behaviour of a brain in real time, let alone the behaviour of the brain coupled with its body.
Secondly, even if one comes up with a passable approximation, I remain to be convinced that my conscious experience will be transferred into the digital version so that the 'digital me' will not be a simulated prediction of who I am. The 'experts' tend to handwave around the difficult parts of matters like this, saying that it's just a matter of having enough instructions per second without any justification that there is an 'enough ips' or that such an 'enough ips' is physically feasible. (For where I am coming from with the latter bit, the horrors of Ramsey Theory come to mind.)
I only use Windows for running Windows only software, like VSTs and DAWs; only use my mac for running music software. All else I use Linux. Having to use the mouse to launch my main apps, rather than a two-key combo is a real annoyance when going back to them. Likewise, not being able to write short scripts with ease is again an annoyance. Not being able to install to a USB stick, taking it from machine to machine without activation problems is another annoyance.
With xfce, for common apps, the procedure:
1. Think of what key combination is most obvious
2. If it is not used, bind the app to that key combo
3. If it is used, add a modifier key, else choose next most obvious
Super-T for terminal, Super-B for browser, Super-E for file manager (inherited from Win95), Super-hash for virt-manager or x2goclient, and so on.
1. Offline backups
2. Never trust the cloud: use it, but do not rely upon it
3. Never trust Apple: use them, but do not rely upon them
In the US, you have weakly regulated free market capitalism where companies are charged with making profits for their shareholders, and some praise this system... then you complain about Apple exploiting tax regulations so as to improve its profitability to its shareholders.
Capitalism stands on self-organisation and enlightened self-interest. When there was the threat of the USSR, this kept greed and short-sightedness in check. In the aftermath of the Cold War, however, that is giving way to unrestrained self-interest. Competition is good to a point, but expending resources to compete rather than produce, and to alter one's strategy away from the overall common good is what is happening. I imagine game theorists saw it coming to a certain degree.
A major problem is one of semantics. Does the answer to 'do we live in a computer simulation?' depend on what exactly we mean by 'computer simulation'? If not, why not (has anybody made any kind of coherent case for this?), and if so, who is bothering to give sufficiently precise detail about what they mean that one can give the question a meaningful answer. I tend to lean towards the 'yes' position, but I tend to leave what is meant by 'computer simulation' sufficiently flexible as to be able to fudge things later. One can always fudge a 'no' answer by taking the 'that's not a computer simulation' objection and repeating it ad nauseam like a broken record.
Thats what 100+ magic pills a day does to your brain.
A universal tradeable resource token is what money is. In earlier parts of civilisation things were simpler, and a money-trade system of resource organisation probably worked reasonably. The modern world has grown complex to a degree that organising resources using money no longer works well. There are too many middlemen, and too many ways for 'value' to leak out in the money system. With regards to the basic resources needed for living, there are surely better ways of organising the human and material resources we have.
In the war between facts and dogma, facts have a habit of coming second. Facts are hard to think through and analyse properly, and proper analyses are detailed and tough to understand. Dogma doesn't have any of these drawbacks.
They don't. The anti sex lobby probably want to use the uncertainty as a weapon against all remotely sexual material.
But it is well known that the momentum of a photon depends on its frequency. If the device spits out microwaves in two directions, the same number of photons per second, but the microwaves come out in one direction are of a higher frequency than the other, then more momentum will be emitted in photons in one direction than the other. I would expect the quantities involved to be miniscule, but that's the first guess I had in mind when I read about this.