It seems to me that "Legend of the Fall" was not really a myth. An object only falls about 2 inches during the first 1/10 second of its fall from rest. A vehicle moving at 70mph can travel about 10 feet in that amount of time. With a rear wheel drive vehicle such as a motorcycle the front wheel can actually lift off the ground when rapidly accelerating, in which case there would just be a 2 inch impact on the rear wheel as it lands on the far side of a 10 foot gap.
Now I do believe that one should pay for what one receives, and I've no problem buying music, but I'd just rather buy from a source that is as close as possible to the creator. That way, as much of my money as possible goes to the creative elements of our society rather than the parasitic ones. Take iTunes: yes, Apple gets very little of the vast stream of raw dollars pouring into it's DRM-constricted throat, but the people actually producing that music don't get even that much. The rest is picked off by businesses that have litte raison d'être in the Internet age. Independent artists can easily get their music onto iTunes and keep almost all of the revenue. For example, anyone selling their CD's (even home made ones) through CDBaby can have their music appear on iTunes, Rhapsody, Yahoo, etc. for no extra charge. The independent artist gets 91% of all revenue that comes to CDBaby from digital sales of their music.
Someone mentioned Parallels, but there's also the Mac version of Wine (Crossover Mac). This has the advantage of not needing a copy of Windows to run Windows apps. Perhaps XP compatibility will be good enough soon to help with enterprise penetration? Or are there special difficulties associated with running custom enterprise apps that makes this scenario unrealistic?
I don't think that will be very difficult. Can you think of anything truly innovative in Windows that wasn't already done somewhere else? Remember that MS has a lot of smart people working in both research and development and that they're encouraged to produce patents, regardless of whether their work finds its way into Windows. And also remember that MS has acquired a lot of patents along with the many startups that it's gobbled up.
I think your nuclear war analogy is very apt: if there weren't substantial risks of retaliation, MS would have litigated already.
The hope of entirely avoiding this issue by avoiding all MS patents seems forlorn. Microsoft holds over 6300 issued US patents directly as the assignee (based on a quick search at uspto.gov), and about 10,000 patents worldwide (see this article). Many of them cover obvious ideas that are hard to avoid using. Proving that a patent claim is invalid based on obviousness is not easy. From the US Patent Office point of view, if an idea has clear benefits and was not in use at the time of the invention (i.e., no prior art), then it wasn't obvious. So in practice you're usually reduced to trying to find prior art. And of course you have to prove all relevant claims invalid.
Having said this, it's probably still important to avoid any truly innovative ideas that MS owns. Litigating based on these seems like less of a risk for MS.
The article didn't make one advantage very clear: since all colors of light are reflected identically, reflective optics can completely avoid chromatic aberration. This is precisely why Isaac Newton invented the reflective telescope. One reason that a good camera lens is heavy and expensive is because it combines elements with different refractive indices to try to minimize chromatic aberration.
The RAZR is the most popular cell phone in the US, and the newest version (available in the US through Cingular) interfaces directly with iTunes. This is a much nicer phone than the ROKR and comes with a 512MB microsd card (see review). Although it has the restriction of only holding 100 songs, this is about what will fit on the included memory card. I have one and I find it a compelling alternative to carrying around a separate nano. I'm not sure why people are so dismissive of this.
I agree that "pruning cruft" based solely on whether the article is important is a terrible idea. If they want to rank articles in some manner so that you can filter out what you don't want to see, that's fine. But Wikipedia is the perfect place for all sorts of information that matters to specialized communities. The function that is needed is review not pruning.
I helped write an article in Wikipedia about the software program DVArchive, which is widely used in the ReplayTV community. It was a short article but it was useful for the intended audience, and several other people also edited the article and added references. The article came up as one of the top two or three google hits for "DVArchive", out of 60,000. Then the article suddenly disappeared because DVArchive wasn't notable enough.
This seems like a bad process. Articles should be judged on whether they are useful, well written, correct, etc., but they should not dissapear simply because someone doesn't think they are important enough. I would favor something more like the slashdot moderation/ranking system, where you set some thresholds (e.g., recency, maturity, notability, etc.) and articles that fall below the thresholds are hidden.
I loved the early Heinlein and I occassionally re-read Heinlein juveniles to try to recapture that feeling of excitement and wonder that he communicated to me when I was a kid, and that still inspires me. The later Heinlein books, starting with Stranger in a Strange Land, are too heavy handed and self indulgent for my tastes. My favorite early novels are probably Citizen of the Galaxy, Tunnel in the Sky, Door Into Summer and Starman Jones, in about that order. These stories communicate Heinlein's love of science and intellect, his take on the nature of courage and integrity, and his optimism about the future.
I loved Heinlein's juveniles growing up and share the wistful desire of many others to see just one more new one. Unfortunately this book captures neither the optimistic spirit nor the intelligence of the best of Heinlein's stories. The first few chapters, which Robinson wrote as samples in order to get the gig, were by far the most Heinlein-esque, but then he spiraled off into a lesser universe. I was eager to love this novel, but in the end I didn't even like it.
For me the things that were most appealing about Heinlein were his enthusiasm and knowledge about science and his insight into people and societies. He had something original to say about both. Robinson doesn't. Robinson's protagonist, Joel Johnston, spends much of the story drunk and trying to destroy himself. Joel starts off with a strong sense of self and what he wants in life, but this seems to fade as the story progresses. The starship is powered not by science and engineering, but by newage mumbo jumbo. The ending is illogical within the context of the story (given faster than light travel, relativistic time dilation provides plenty of time to warn the colonies and have one of them build a rescue ship).
I'm not particularly concerned with the logical inconsistencies, since it would be hard to enjoy very many science fiction stories without ignoring those. But in the end the Heinlein spirit is absent and the story does not inspire or inform.
I wonder if Apple can take a big chunk of the OS market without running on other people's machines. Windows and Linux are currently fighting to expand into all possible niches (including handheld devices and storage devices, etc.). If Apple would like to do the same they should probably run on other people's hardware.
I thought Steve made a big point at WWDC that their Mac Pro is significantly cheaper than the equivalent Dell configuration. Competing head to head on hardware value seems like it would make it possible to separate hardware and software sales and release OS X to run on generic x86 hardware.
Come on Apple, you have such great products and everyone who matters knows it. You won't convince the Microsoft fanbois in this way, anyways and it only leads to a "But I had it first" shouting-match. It almost felt they had to provide red meat for Apple zealots.
Steve Jobs said that 1 in 4 at the conference is an Apple engineer. Perhaps he was rallying the Apple troops by throwing some Vista red meat to them.
The main danger in relying on labor from a distorted economy is that, if the distortion ends suddenly, then our economy is disrupted. If we don't expect that there will be an end to low-cost Mexican labor anytime soon, then the US economy gains tremendously from employing Mexicans. It's as if someone provided us with some other necessary resource such as oil or steel at a very low price. There is always a challenge in educating the people that are freed up by cheap manual labor, but we should get better at this.
I don't think there's a serious danger that we will outsource all our CS jobs. Today, practically every technical field involves programming, and most non-technical fields need help from programmers. To get a good job you may need to know how to program and also have some other domain-specific knowledge.
I'm not a patent attorney but I've filed about 40 patents on technology I've developed, written mostly by myself with a patent attorney just doing a final pass over the claims. There is a real art to writing good patent claims and if you're new to it you should get some professional help with at least that part (in addition to reading up on writing patents). In some ways a patent is like a computer program and the claims are the actual code -- the rest is just comments that help make the claims understandable. A court battle revolves completely around the claims. The battle is over whether the defendant's technology "reads" against the plaintiff's claims, and whether the relevant claims are valid, in light of ealier technology ("prior art").
A good prior art search makes your patent much more valuable and you are the best person to do that search, since you understand the subject area. There seems to be a common misconception that other patents are the main place you need to look. In my experience, the earliest prior art is almost never in other patents. For example, the best prior art for hash-based object naming I've seen is in a program called FWKCS that was used with a few BBS systems in the late 1980's. The patents in this area were all filed starting in the mid-1990's. Online mailing lists are a good source of pointers to programs and also descriptions of ideas that themselves constitute prior art, since they are public. Do some searching in Google Groups (formerly usenet groups) and in the archives of specialized mailing lists that are relevant to your topic (e.g., www-talk for early Internet-related ideas).
Academic papers are often an excellent source of prior art. Many papers are available online for free though citseer
(http://citeseer.ist.psu.edu/). You may also want to join the IEEE and ACM to gain access to their rather complete databases of all their published journal articles. Finally, you should also search in some patent databases. The USPTO has full text search online at their site
(https://sportal.uspto.gov/secure/portal/efs-unreg istered). Subscription to a search database such as Delphion
(http://www.delphion.com/) is relatively cheap if you only subscribe for a month or two. Note that the important date for a patent is its priority date, which is when it was filed or when an earlier application it is based on was filed. In the US, applicants are allowed to present proof in a court case that they actually had the idea up to a year before their priority date, so you may need to find art a year older than a filing date to be able to argue that a claim is invalid.
Finally, remember that you're searching for art relevant to claims (usually the broadest claims in a patent). Patents are not invalidated, specific sets of claims are. The more you understand what the most essential differences are between what you've done and what others have done before you, the stronger a set of claims you and an experienced patent writer will be able to put together. Good luck!
In the interview, Eckert mentions two concepts, still important in modern computers, which first appeared in the ENIAC. He says, "The idea of a subroutine was original with ENIAC." He also says, "The idea of using internal memory was also original with ENIAC." In fact, Ada Byron (who became the Countess Lady Lovelace) is usually credited with inventing the subroutine. She wrote programs for Babbage's Analytical Engine, which was never completed. The Analytical Engine design also had internal memory. This was about a century before the ENIAC.
19th century engineering wasn't really up to building the Analytical Engine. Babbage famously said, late in his life, that he would gladly give up the rest of his time if he could spend just three days 500 years in the future. Of course a man who was really out of this time was Leonardo DaVinci, who sketched a 13 digit cogwheel digital adder in the 15th century.
A DNA-fingerprint isn't like a finger-fingerprint, since once databases get a good sample of the population DNA, they can get close to identifying anyone from DNA evidence alone. The fact that a kid tracked someone down who wasn't in the database suggests that, as these databases mature and become cheap, individuals will be able to track each other down from their DNA. Didn't get her phone number? Look for a strand of hair. And better not use the restroom at the car dealership if you don't want to start receiving targeted advertising...
I'm one of those kids who always got 100% on tests in math/science classes. I never skipped any grades and am happy about that: the best thing about primary school was that it took up so little of my time that I had the freedom to actually develop interests and learn things (on my own). At university, I foolishly got myself into a program that was sufficiently time consuming that it became very difficult for me to learn anything at all! For me, learning answers without having the time and freedom to come up with and play with questions simply spoils the fun and kills interest in a subject. The hardest thing in science is not learning facts or techniques, but maintaining excitement and motivation and figuring out what is worth doing. I don't see how putting a young kid in university is going to help with this.
This paper is speculative, but in some ways it is also conservative. Their suggestion is that the new force between dark matter particles that is needed to explain anomolous astrophysical observations is... gravity.
The idea of thin extra dimensions has been current in physics for decades, and is the best way that theorists have found to reconcile General Relativity with Quantum Mechanics (not just in string theory, but long before it). It's analogous to taking a 2D bitmap and giving it a few pixels of depth in the 3rd dimension.
The ADD model that the authors cite was proposed to explain a non-astrophysical mystery: why the force of gravity is so much weaker than the electromagnetic force. The ADD idea was that, if you already have a theory with extra dimensions, some of the extra dimensions might be much thicker than the fundamental Planck size scale, but still very thin. The size of the thicker dimensions compared to the Planck scale determines the ratio of the EM and gravitational forces.
The authors of this paper realized that the ADD model could be used to explain anomolous observations about dark matter -- the observed behavior is explained by gravity being stronger at small distances. They used astrophysical observations to set the number of dimensions in the ADD model and the mass of the fundamental dark matter particle. Now that they've made new predictions about the force of gravity at small distances, lab experiments can be designed to test them.
A key point here is that you can scale both the clock and the voltage. The power dissipated each time you get rid of a stored charge is proportional to the voltage squared. How often you do this per second depends on the frequency. Thus if you can run at half the voltage and one third of the frequency, then the processor dissipates 1/12 of the power!
A fair comparison of processors would be to calculate how much processing power you get in some benchmark per Watt of power dissipated: Fan noise for getting rid of CPU heat should be lowest when the least heat needs to be gotten rid of. Probably the processor with the best performance per Watt at full speed also has the lowest disspation per Watt at lower speeds.
Infostor did an article in February about how healthcare and financial services companies are now required to keep a lot of stuff, and how they are handling it.
A problem with 3D processors is that they contain a lot of active circuitry, so it's hard to get the heat out (surface to volume ratio and all that). Memory doesn't inherently have the same problem because you only need to activate a small part of it at a time to get particular bits out. ROMs are particularly nice because they don't need to dissipate power just to remember a bit.
Another reason that big 3D memory is much easier than big 3D processors is because, with memory, areas with manufacturing defects can simply not be used.
If Moore's Law continues for another 20 years, memory chips will reach a density of about one bit for every atom on the surface of the active area of the chip. That sounds like a hard limit, but it isn't. Even this density still lets us use millions of atoms per bit if the bits aren't on the surface, but are piled up inside: there's a lot more volume than surface! Thus the notion that we reach an absolute limit in 20 years is a misconception.
Exactly the same technique (running backwards by restarting from a checkpoint and running forwards) was used in a reversible debugger about 30 years ago at MIT. I think that Ed Fredkin wrote it.
Slashdot Mirror
It seems to me that "Legend of the Fall" was not really a myth. An object only falls about 2 inches during the first 1/10 second of its fall from rest. A vehicle moving at 70mph can travel about 10 feet in that amount of time. With a rear wheel drive vehicle such as a motorcycle the front wheel can actually lift off the ground when rapidly accelerating, in which case there would just be a 2 inch impact on the rear wheel as it lands on the far side of a 10 foot gap.
Someone mentioned Parallels, but there's also the Mac version of Wine (Crossover Mac). This has the advantage of not needing a copy of Windows to run Windows apps. Perhaps XP compatibility will be good enough soon to help with enterprise penetration? Or are there special difficulties associated with running custom enterprise apps that makes this scenario unrealistic?
I think your nuclear war analogy is very apt: if there weren't substantial risks of retaliation, MS would have litigated already.
The hope of entirely avoiding this issue by avoiding all MS patents seems forlorn. Microsoft holds over 6300 issued US patents directly as the assignee (based on a quick search at uspto.gov), and about 10,000 patents worldwide (see this article). Many of them cover obvious ideas that are hard to avoid using. Proving that a patent claim is invalid based on obviousness is not easy. From the US Patent Office point of view, if an idea has clear benefits and was not in use at the time of the invention (i.e., no prior art), then it wasn't obvious. So in practice you're usually reduced to trying to find prior art. And of course you have to prove all relevant claims invalid.
Having said this, it's probably still important to avoid any truly innovative ideas that MS owns. Litigating based on these seems like less of a risk for MS.
The article about Vista vulnerabilities that you're referring to is here.
The article didn't make one advantage very clear: since all colors of light are reflected identically, reflective optics can completely avoid chromatic aberration. This is precisely why Isaac Newton invented the reflective telescope. One reason that a good camera lens is heavy and expensive is because it combines elements with different refractive indices to try to minimize chromatic aberration.
The RAZR is the most popular cell phone in the US, and the newest version (available in the US through Cingular) interfaces directly with iTunes. This is a much nicer phone than the ROKR and comes with a 512MB microsd card (see review). Although it has the restriction of only holding 100 songs, this is about what will fit on the included memory card. I have one and I find it a compelling alternative to carrying around a separate nano. I'm not sure why people are so dismissive of this.
I agree that "pruning cruft" based solely on whether the article is important is a terrible idea. If they want to rank articles in some manner so that you can filter out what you don't want to see, that's fine. But Wikipedia is the perfect place for all sorts of information that matters to specialized communities. The function that is needed is review not pruning.
I helped write an article in Wikipedia about the software program DVArchive, which is widely used in the ReplayTV community. It was a short article but it was useful for the intended audience, and several other people also edited the article and added references. The article came up as one of the top two or three google hits for "DVArchive", out of 60,000. Then the article suddenly disappeared because DVArchive wasn't notable enough.
This seems like a bad process. Articles should be judged on whether they are useful, well written, correct, etc., but they should not dissapear simply because someone doesn't think they are important enough. I would favor something more like the slashdot moderation/ranking system, where you set some thresholds (e.g., recency, maturity, notability, etc.) and articles that fall below the thresholds are hidden.
I loved the early Heinlein and I occassionally re-read Heinlein juveniles to try to recapture that feeling of excitement and wonder that he communicated to me when I was a kid, and that still inspires me. The later Heinlein books, starting with Stranger in a Strange Land, are too heavy handed and self indulgent for my tastes. My favorite early novels are probably Citizen of the Galaxy, Tunnel in the Sky, Door Into Summer and Starman Jones, in about that order. These stories communicate Heinlein's love of science and intellect, his take on the nature of courage and integrity, and his optimism about the future.
I loved Heinlein's juveniles growing up and share the wistful desire of many others to see just one more new one. Unfortunately this book captures neither the optimistic spirit nor the intelligence of the best of Heinlein's stories. The first few chapters, which Robinson wrote as samples in order to get the gig, were by far the most Heinlein-esque, but then he spiraled off into a lesser universe. I was eager to love this novel, but in the end I didn't even like it.
For me the things that were most appealing about Heinlein were his enthusiasm and knowledge about science and his insight into people and societies. He had something original to say about both. Robinson doesn't. Robinson's protagonist, Joel Johnston, spends much of the story drunk and trying to destroy himself. Joel starts off with a strong sense of self and what he wants in life, but this seems to fade as the story progresses. The starship is powered not by science and engineering, but by newage mumbo jumbo. The ending is illogical within the context of the story (given faster than light travel, relativistic time dilation provides plenty of time to warn the colonies and have one of them build a rescue ship).
I'm not particularly concerned with the logical inconsistencies, since it would be hard to enjoy very many science fiction stories without ignoring those. But in the end the Heinlein spirit is absent and the story does not inspire or inform.
I wonder if Apple can take a big chunk of the OS market without running on other people's machines. Windows and Linux are currently fighting to expand into all possible niches (including handheld devices and storage devices, etc.). If Apple would like to do the same they should probably run on other people's hardware.
I thought Steve made a big point at WWDC that their Mac Pro is significantly cheaper than the equivalent Dell configuration. Competing head to head on hardware value seems like it would make it possible to separate hardware and software sales and release OS X to run on generic x86 hardware.
Steve Jobs said that 1 in 4 at the conference is an Apple engineer. Perhaps he was rallying the Apple troops by throwing some Vista red meat to them.
The main danger in relying on labor from a distorted economy is that, if the distortion ends suddenly, then our economy is disrupted. If we don't expect that there will be an end to low-cost Mexican labor anytime soon, then the US economy gains tremendously from employing Mexicans. It's as if someone provided us with some other necessary resource such as oil or steel at a very low price. There is always a challenge in educating the people that are freed up by cheap manual labor, but we should get better at this.
I don't think there's a serious danger that we will outsource all our CS jobs. Today, practically every technical field involves programming, and most non-technical fields need help from programmers. To get a good job you may need to know how to program and also have some other domain-specific knowledge.
I'm not a patent attorney but I've filed about 40 patents on technology I've developed, written mostly by myself with a patent attorney just doing a final pass over the claims. There is a real art to writing good patent claims and if you're new to it you should get some professional help with at least that part (in addition to reading up on writing patents). In some ways a patent is like a computer program and the claims are the actual code -- the rest is just comments that help make the claims understandable. A court battle revolves completely around the claims. The battle is over whether the defendant's technology "reads" against the plaintiff's claims, and whether the relevant claims are valid, in light of ealier technology ("prior art").
A good prior art search makes your patent much more valuable and you are the best person to do that search, since you understand the subject area. There seems to be a common misconception that other patents are the main place you need to look. In my experience, the earliest prior art is almost never in other patents. For example, the best prior art for hash-based object naming I've seen is in a program called FWKCS that was used with a few BBS systems in the late 1980's. The patents in this area were all filed starting in the mid-1990's. Online mailing lists are a good source of pointers to programs and also descriptions of ideas that themselves constitute prior art, since they are public. Do some searching in Google Groups (formerly usenet groups) and in the archives of specialized mailing lists that are relevant to your topic (e.g., www-talk for early Internet-related ideas).
Academic papers are often an excellent source of prior art. Many papers are available online for free though citseer (http://citeseer.ist.psu.edu/). You may also want to join the IEEE and ACM to gain access to their rather complete databases of all their published journal articles. Finally, you should also search in some patent databases. The USPTO has full text search online at their site (https://sportal.uspto.gov/secure/portal/efs-unreg istered). Subscription to a search database such as Delphion
(http://www.delphion.com/) is relatively cheap if you only subscribe for a month or two. Note that the important date for a patent is its priority date, which is when it was filed or when an earlier application it is based on was filed. In the US, applicants are allowed to present proof in a court case that they actually had the idea up to a year before their priority date, so you may need to find art a year older than a filing date to be able to argue that a claim is invalid.
Finally, remember that you're searching for art relevant to claims (usually the broadest claims in a patent). Patents are not invalidated, specific sets of claims are. The more you understand what the most essential differences are between what you've done and what others have done before you, the stronger a set of claims you and an experienced patent writer will be able to put together. Good luck!
In the interview, Eckert mentions two concepts, still important in modern computers, which first appeared in the ENIAC. He says, "The idea of a subroutine was original with ENIAC." He also says, "The idea of using internal memory was also original with ENIAC." In fact, Ada Byron (who became the Countess Lady Lovelace) is usually credited with inventing the subroutine. She wrote programs for Babbage's Analytical Engine, which was never completed. The Analytical Engine design also had internal memory. This was about a century before the ENIAC.
19th century engineering wasn't really up to building the Analytical Engine. Babbage famously said, late in his life, that he would gladly give up the rest of his time if he could spend just three days 500 years in the future. Of course a man who was really out of this time was Leonardo DaVinci, who sketched a 13 digit cogwheel digital adder in the 15th century.
A DNA-fingerprint isn't like a finger-fingerprint, since once databases get a good sample of the population DNA, they can get close to identifying anyone from DNA evidence alone. The fact that a kid tracked someone down who wasn't in the database suggests that, as these databases mature and become cheap, individuals will be able to track each other down from their DNA. Didn't get her phone number? Look for a strand of hair. And better not use the restroom at the car dealership if you don't want to start receiving targeted advertising...
I'm one of those kids who always got 100% on tests in math/science classes. I never skipped any grades and am happy about that: the best thing about primary school was that it took up so little of my time that I had the freedom to actually develop interests and learn things (on my own). At university, I foolishly got myself into a program that was sufficiently time consuming that it became very difficult for me to learn anything at all! For me, learning answers without having the time and freedom to come up with and play with questions simply spoils the fun and kills interest in a subject. The hardest thing in science is not learning facts or techniques, but maintaining excitement and motivation and figuring out what is worth doing. I don't see how putting a young kid in university is going to help with this.
The idea of thin extra dimensions has been current in physics for decades, and is the best way that theorists have found to reconcile General Relativity with Quantum Mechanics (not just in string theory, but long before it). It's analogous to taking a 2D bitmap and giving it a few pixels of depth in the 3rd dimension.
The ADD model that the authors cite was proposed to explain a non-astrophysical mystery: why the force of gravity is so much weaker than the electromagnetic force. The ADD idea was that, if you already have a theory with extra dimensions, some of the extra dimensions might be much thicker than the fundamental Planck size scale, but still very thin. The size of the thicker dimensions compared to the Planck scale determines the ratio of the EM and gravitational forces.
The authors of this paper realized that the ADD model could be used to explain anomolous observations about dark matter -- the observed behavior is explained by gravity being stronger at small distances. They used astrophysical observations to set the number of dimensions in the ADD model and the mass of the fundamental dark matter particle. Now that they've made new predictions about the force of gravity at small distances, lab experiments can be designed to test them.
A fair comparison of processors would be to calculate how much processing power you get in some benchmark per Watt of power dissipated: Fan noise for getting rid of CPU heat should be lowest when the least heat needs to be gotten rid of. Probably the processor with the best performance per Watt at full speed also has the lowest disspation per Watt at lower speeds.
Infostor did an article in February about how healthcare and financial services companies are now required to keep a lot of stuff, and how they are handling it.
Another reason that big 3D memory is much easier than big 3D processors is because, with memory, areas with manufacturing defects can simply not be used.
If Moore's Law continues for another 20 years, memory chips will reach a density of about one bit for every atom on the surface of the active area of the chip. That sounds like a hard limit, but it isn't. Even this density still lets us use millions of atoms per bit if the bits aren't on the surface, but are piled up inside: there's a lot more volume than surface! Thus the notion that we reach an absolute limit in 20 years is a misconception.
Exactly the same technique (running backwards by restarting from a checkpoint and running forwards) was used in a reversible debugger about 30 years ago at MIT. I think that Ed Fredkin wrote it.