There's an old curse that seems relevant: "May you live in interesting times." Times are certainly interesting. At this point, it seems like some sort of full-scale war between NATO and Russia is more likely now than it has been any time since the 1980s (granted then it would have been NATO against the USSR but the basic point is the same). Worse, at least historically the military and diplomats spent much of their time making sure that things didn't spiral out of control. Without the Cold War feeling, people may feel less of a need to guard against such issues. Worse, Russian military doctrine currently describes a limited nuclear strike on conventional military targets as a de-escalation http://thebulletin.org/why-russia-calls-limited-nuclear-strike-de-escalation . While in official documents they reserve that terminology for using nuclear weapons to handle direct conventional military attacks on Russia itself, one finds very worrying the level of doublethink where one describes being the first to use nukes as de-escalating a situation.
During the Cold War, one popular explanation for the Fermi paradox https://en.wikipedia.org/wiki/Fermi_paradox, the apparent lack of highly advanced civilizations in the universe, was that species end up blowing themselves up. For most of my life, this belief looked almost quaint but it is not looking disturbingly likely. At this point, the evidence for some sort of serious barrier to civilizations emerging substantially is much stronger than it was a few decades ago. The apparent lack of K3 or K2.5 civilizations is at this point substantially robust https://en.wikipedia.org/wiki/Kardashev_scale with around 100,000 galaxies searched and almost no sign of any civilization using a substantial fraction of its galactic energy output http://www.scientificamerican.com/article/alien-supercivilizations-absent-from-100-000-nearby-galaxies/. With this return to Cold War norms, it looks like we need to not only take seriously that there's a Great Filter, but that the Filter might be nuclear war. That's especially the case because a nuclear war does not need to kill every member of the civilization to completely destroy any hope of a technologically advanced civilization. If not enough natural resources have been consumed by the civilization (e.g. the easily accessible coal and oil) then even if the species survives it may not have the ability to reboot itself to a high tech level since getting to a high tech level may actually require access to these resources (in which case one gets essentially one chance to get to be a high tech civilization).
The real worry should be Kessler Syndrome
on
Satellite Wars (ft.com)
·
· Score: 4, Interesting
There's a serious risk that in low-Earth orbit if one has enough debris it could cause a cascade of destruction where debris hits satellites breaking them up into more debris which hits more satellites and so on. Such a cascade is called Kessler Syndrome https://en.wikipedia.org/wiki/... . If this happens it could render many orbits unusable for years. In that context, deliberately destroying satellites should maybe be considered a war crime since the potential for collateral damage impacting all of humanity is so severe.
In the case of natural gas power plants. . . For now, they're much better than coal. For the future, solar power and nuclear fusion will eventually kill them off.
If we ever get fusion to work effectively it might end up killing everything off. But in the moderate term, solar isn't going to be the only one. A combination of nuclear fission, solar, wind, geothermal and hydroelectric is much more viable and solves many of the problems. Fusion is a very long way off and it is likely that we'll stop having susbstantial fossil fuel use well before fusion is a common power source.
Agree with your last two paragraphs (and the point about how right now storage is mainly going to be used for fossil plants is certainly important) . I think however you may be underestimating how often wind power gets wasted in some areas, although it may be an issue of where exactly one is looking. For example, in the US there's a lot of wind power in Texas but it often gets wasted. And if one is a smaller, isolated grid, such as many islands, this problem is even more severe. But you are right that "vast majority" was probably too strong especially in the context of Western and Central Europe which has done a good job integrating their grids.
It reduces it but not as much as a naive calculation suggests since a) conventional plants have startup times where you burn a lot more, b) due to slow startup time you need some fossil fuel running all the time if you don't have a lot of storage so if there's a sudden lull in wind or solar you can still keep the grid steady. There's an excellent book that covers all these issues for laypeople, "Before the Lights Go Out" about the history and future of the electric grids by Maggie Koerth-Baker. Her book focuses on the American grid (well grids really since there are three main grids, East, West and Texas) but most of the book applies to pretty much any large-scale grid.
This is great. People often underestimate how important energy storage is. Many of the sources of power that don't produce CO2 are intermittent. Wind and solar are the primary examples. Sometimes it is sunny but sometimes it isn't. Sometimes it is windy but it isn't. Thus for example you have headlines about how for one day or so you'll have some country or region produce more power than it needs using wind, but they miss that the vast majority of the time this extra power is wasted and the next day they need to go burn a lot of fossil fuels. The problem isn't as completely bad as one might guess since wind is generally strong at night when solar isn't an option, but the general need for cheap and efficient storage is definitely there. The best storage form in terms of being cheap and efficient is pumped hydroelectric https://en.wikipedia.org/wiki/... but it requires specific nice geology to work.
We don't actually have a proof that group isomorphism is in P, and we don't also know whether this new work will help at all with that problem. One might hope that one could take a pair of groups, convert them to the relevant graphs and then run this, but that's unlikely to be helpful since we already have group isomorphism algorithms that work in quasipolynomial time- in fact the most obvious non-trivial one does so. Note also that in the case of the groups arising from Diffie-Hellman and similar procedures, we know exactly what the groups are isomorphic to. They are in fact easy to write down Abelian groups and knowing precisely what the groups are is necessary for the cryptographic algorithms to work. So there's no obvious way that this sort of thing helps there.
There are a lot of natural problems that involve graph theoretic aspects or graph isomorphism in particular. Chip design is one major example, where it is used in establishing that different designs for part of a chip really will do the same thing. However, it is not that likely this will end up having a substantial practical implication by itself because for most purposes graph isomorphism is an easy problem. In particular, for two random graphs it is easy to tell whether they are isomorphic or not (and for many practical applications NAUTY will work fine http://www3.cs.stonybrook.edu/... ). This is in contrast to factoring integers where factoring a random positive integer seems very tough. This is also why crypto uses factoring but not graph isomorphism: making a crypto scheme where random instances are easy isn't a great idea.
The summary is missing important information. As TFA discusses, the primary issue was (as already strongly suspected) the loss of the magnetic field around Mars. With only a very weak magnetic field nothing protected the planet's atmosphere from the solar wind which blasted the atmosphere away over a time span probably in the hundreds of millions of years. This last result, the slow loss of the atmosphere is a genuinely novel and important discovery because as TFA discusses this makes it more plausible that if there was life on Mars that it would have had time to evolve to survive the gradually harsher environment. The research also suggests that Mars will become completely airless in around 2 billion years.
This has nothing to do with being specifically liberals. People of all political stripes will advocate censoring views they don't like when they have the power to do so. Look at for example Ben Carson's recent proposal to have the Department of Education actively censor "extreme" political speech.
I agree. I saw the headline and thought I was going to see an interesting article about Go and computers, maybe something explaining in detail how Monte Carlo methods have made Go programs much more successful than the older brute force used to be. But nope. Maybe someone should write a program in Go to play Go for maximum confusion.
Yes and no. There is some advantage to getting close to real time data: there's a a Supernova Early Warning System http://snews.bnl.gov/. This isn't a safety issue, but rather an astronomy issue.
Detectors like IceCube can be used to actually detect the neutrinos from a supernova before the supernova's light reaches Earth. This isn't due to the erroneous claim from a few years ago that neutrinos travel faster than light, but rather because when a supernova occurs, the light from the core of the star takes multiple hours to get out of the core because of all the mass in the way, while the neutrinos aren't slowed down by this almost at all. This means that the neutrinos effectively get a few hours head start on the light- since they are traveling so close to the speed of light, they get to keep almost all this head start by the time they reach Earth. In the case of SN 1987A https://en.wikipedia.org/wiki/SN_1987A, a supernova in 1987 which was close enough that we could detect the flood of neutrinos, the neutrinos did as predicted arrive a few hours before the light. This means we can if we detect a neutrino burst and can get its directional data (which IceCube can approximately do) then we can point our telescopes at a supernova *before the light arrives at Earth* which means we'll get to see the very beginning of the supernova and hopefully get a much better understanding.
In order to do this you have to do at least some of your processing in at least close to real time as you can. This is especially important because it isn't actually easy to figure out from the neutrino burst what direction the supernova is coming from, and IceCube is one of the few detectors which gets any good directional data at all, so if this happens we want to process the data rapidly enough to get a good idea of where to look.
Note that there aren't any crypto systems which are proveably safe from quantum attacks, only systems where we strongly suspect them to be safe. At this point, other than one-time pads, we can't prove any system is even safe in a classicial setting. For example, RSA and Diffie-Hellman both rely on conjectures which are strictly stronger than P != NP.
Scott Aaronson has an excellent summary of this research on his blog: http://www.scottaaronson.com/b... One point that Scott makes that is easy to lose track of is how much working this out required people on both the theoretical crypto end and the practical crypto end to work together. This is a combination of multiple vulnerabilities and some clever number theory.
D-Wave's claimed quantum computers depend very much on what you call a quantum computer. D-Waves machines use a form of quantum annealing https://en.wikipedia.org/wiki/Quantum_annealing but they are not a universal quantum computer in the traditional sense, and even for quantum annealing they are very limited in what they can do and it isn't even clear that the problems that the D-Wave machine can do are any problems where we should expect any actual speedup from a quantum computer, and certainly the D-Wave machines have no capability for doing many of the problems we do want to use quantum computers for like factoring large integers.
SETH is a very big assumption, much stronger than even assuming P != NP. Essentially, the exponential time hypothesis (ETH) says that any algorithm which solves instances of 3-SAT https://en.wikipedia.org/wiki/3-SAT will have worst case running time that is exponential. However, it is conceivable that one could have a whole series of algorithms which each solves 3-SAT better and better. The idea is that there's an algorithm a_i which solves 3-SAT in time O(x_i^n) and as i goes to infinity, x_i goes to 1. SETH is the hypothesis that says essentially that that doesn't happen. Many people are not convinced that ETH is true although it certainly looks plausible. I think most people who have thought about this consider the possibility that ETH holds and SETH doesn't hold to be a deeply weird and generally unlikely option.
This is incredibly confused. 103 was the *maximum recorded lifespan.* What matters is the *average lifespan.* It is true that if one takes into account improvements in infant mortality the jump in life expectancy hasn't been as large http://www.ancient-origins.net/news-evolution-human-origins/life-expectancy-myth-and-why-many-ancient-humans-lived-long-077889. But even given that, life expectancy on average has gone up by about a decade in the US in the last 200 years even if one only works at people surviving past infancy.
No. Commercial speech is in general protected- the line it crosses to being non-protected is when money changes hands under false pretenses. At that point, it isn't a speech issue, but is fraud.
No. Commercial issues are distinct. But in general saying "This Foo is from the Ming dynasty" when it isn't should not be and is not illegal by itself. Selling it under false pretenses is correctly illegal. Notice that what is happening here involves not just speech but money changing hands.
*Raises hand* I don't know what hate speech is, but I'm not in favor of laws censoring speech, and that includes Holocaust denial. My grandmother had a number on her arm. Many people on my mother's side of the family were killed in the Holocaust. Holocaust denial is disgusting and these people spouting this denial are in the very best case reprehensible fucktards. That doesn't mean they don't have a right to spout their reprehensible fucktardery.
The Sierra Club which is in many ways more moderate than Greenpeace weakly opposes such fusion also http://www.sierraclub.org/policy/energy/nuclear-power, and while their main argument is that it is too expensive compared to more conventional renewables, they also cite "The dangers posed by the probable releases of tritium used by fusion plants, the problems with decommissioning these plants" which only makes sense if you both don't fully understand how little tritium is being used and how think that the plants will be highly radioactive like conventional fission plants.
The environmental movement has done a lot of good and continues to do a lot of good. But there is a definite anti-technology bent in some parts and general anti-nuclear bent which is very unfortunate. There are some environmentalists who understand the potential benefits of fusion and how it is different than fission power, but it is definitely not all of them and certainly doesn't include some of the most prominent organizations.
If you have 100 studies you are replicating, by sheer chance you are likely still going to have a few who you successfully replicate but aren't real. So the problem may be worse than that (slightly). Psychology isn't the only field with these issues. There have been a lot of problems in medicine also. See https://www.newscientist.com/article/mg21528826-000-is-medical-science-built-on-shaky-foundations/. Part of the problem is one of incentives: the incentives to do a study which simply replicates a pre-existing study is low, and many journals won't even publish them. This also combines in bad ways with post-hoc analysis where you look at your data and find a pattern in it that is worth publishing; the worst offender here is medicine where people use different statistical tests and different subgroup analysis until they get a positive result.
Slashdot Mirror
There's an old curse that seems relevant: "May you live in interesting times." Times are certainly interesting. At this point, it seems like some sort of full-scale war between NATO and Russia is more likely now than it has been any time since the 1980s (granted then it would have been NATO against the USSR but the basic point is the same). Worse, at least historically the military and diplomats spent much of their time making sure that things didn't spiral out of control. Without the Cold War feeling, people may feel less of a need to guard against such issues. Worse, Russian military doctrine currently describes a limited nuclear strike on conventional military targets as a de-escalation http://thebulletin.org/why-russia-calls-limited-nuclear-strike-de-escalation . While in official documents they reserve that terminology for using nuclear weapons to handle direct conventional military attacks on Russia itself, one finds very worrying the level of doublethink where one describes being the first to use nukes as de-escalating a situation.
During the Cold War, one popular explanation for the Fermi paradox https://en.wikipedia.org/wiki/Fermi_paradox, the apparent lack of highly advanced civilizations in the universe, was that species end up blowing themselves up. For most of my life, this belief looked almost quaint but it is not looking disturbingly likely. At this point, the evidence for some sort of serious barrier to civilizations emerging substantially is much stronger than it was a few decades ago. The apparent lack of K3 or K2.5 civilizations is at this point substantially robust https://en.wikipedia.org/wiki/Kardashev_scale with around 100,000 galaxies searched and almost no sign of any civilization using a substantial fraction of its galactic energy output http://www.scientificamerican.com/article/alien-supercivilizations-absent-from-100-000-nearby-galaxies/. With this return to Cold War norms, it looks like we need to not only take seriously that there's a Great Filter, but that the Filter might be nuclear war. That's especially the case because a nuclear war does not need to kill every member of the civilization to completely destroy any hope of a technologically advanced civilization. If not enough natural resources have been consumed by the civilization (e.g. the easily accessible coal and oil) then even if the species survives it may not have the ability to reboot itself to a high tech level since getting to a high tech level may actually require access to these resources (in which case one gets essentially one chance to get to be a high tech civilization).
There's a serious risk that in low-Earth orbit if one has enough debris it could cause a cascade of destruction where debris hits satellites breaking them up into more debris which hits more satellites and so on. Such a cascade is called Kessler Syndrome https://en.wikipedia.org/wiki/... . If this happens it could render many orbits unusable for years. In that context, deliberately destroying satellites should maybe be considered a war crime since the potential for collateral damage impacting all of humanity is so severe.
In the case of natural gas power plants. . . For now, they're much better than coal. For the future, solar power and nuclear fusion will eventually kill them off.
If we ever get fusion to work effectively it might end up killing everything off. But in the moderate term, solar isn't going to be the only one. A combination of nuclear fission, solar, wind, geothermal and hydroelectric is much more viable and solves many of the problems. Fusion is a very long way off and it is likely that we'll stop having susbstantial fossil fuel use well before fusion is a common power source.
Agree with your last two paragraphs (and the point about how right now storage is mainly going to be used for fossil plants is certainly important) . I think however you may be underestimating how often wind power gets wasted in some areas, although it may be an issue of where exactly one is looking. For example, in the US there's a lot of wind power in Texas but it often gets wasted. And if one is a smaller, isolated grid, such as many islands, this problem is even more severe. But you are right that "vast majority" was probably too strong especially in the context of Western and Central Europe which has done a good job integrating their grids.
It reduces it but not as much as a naive calculation suggests since a) conventional plants have startup times where you burn a lot more, b) due to slow startup time you need some fossil fuel running all the time if you don't have a lot of storage so if there's a sudden lull in wind or solar you can still keep the grid steady. There's an excellent book that covers all these issues for laypeople, "Before the Lights Go Out" about the history and future of the electric grids by Maggie Koerth-Baker. Her book focuses on the American grid (well grids really since there are three main grids, East, West and Texas) but most of the book applies to pretty much any large-scale grid.
This is great. People often underestimate how important energy storage is. Many of the sources of power that don't produce CO2 are intermittent. Wind and solar are the primary examples. Sometimes it is sunny but sometimes it isn't. Sometimes it is windy but it isn't. Thus for example you have headlines about how for one day or so you'll have some country or region produce more power than it needs using wind, but they miss that the vast majority of the time this extra power is wasted and the next day they need to go burn a lot of fossil fuels. The problem isn't as completely bad as one might guess since wind is generally strong at night when solar isn't an option, but the general need for cheap and efficient storage is definitely there. The best storage form in terms of being cheap and efficient is pumped hydroelectric https://en.wikipedia.org/wiki/... but it requires specific nice geology to work.
We don't actually have a proof that group isomorphism is in P, and we don't also know whether this new work will help at all with that problem. One might hope that one could take a pair of groups, convert them to the relevant graphs and then run this, but that's unlikely to be helpful since we already have group isomorphism algorithms that work in quasipolynomial time- in fact the most obvious non-trivial one does so. Note also that in the case of the groups arising from Diffie-Hellman and similar procedures, we know exactly what the groups are isomorphic to. They are in fact easy to write down Abelian groups and knowing precisely what the groups are is necessary for the cryptographic algorithms to work. So there's no obvious way that this sort of thing helps there.
Yes, and in the title isomorphism is spelled wrong also. Not my best submission.
There are a lot of natural problems that involve graph theoretic aspects or graph isomorphism in particular. Chip design is one major example, where it is used in establishing that different designs for part of a chip really will do the same thing. However, it is not that likely this will end up having a substantial practical implication by itself because for most purposes graph isomorphism is an easy problem. In particular, for two random graphs it is easy to tell whether they are isomorphic or not (and for many practical applications NAUTY will work fine http://www3.cs.stonybrook.edu/... ). This is in contrast to factoring integers where factoring a random positive integer seems very tough. This is also why crypto uses factoring but not graph isomorphism: making a crypto scheme where random instances are easy isn't a great idea.
The summary is missing important information. As TFA discusses, the primary issue was (as already strongly suspected) the loss of the magnetic field around Mars. With only a very weak magnetic field nothing protected the planet's atmosphere from the solar wind which blasted the atmosphere away over a time span probably in the hundreds of millions of years. This last result, the slow loss of the atmosphere is a genuinely novel and important discovery because as TFA discusses this makes it more plausible that if there was life on Mars that it would have had time to evolve to survive the gradually harsher environment. The research also suggests that Mars will become completely airless in around 2 billion years.
This has nothing to do with being specifically liberals. People of all political stripes will advocate censoring views they don't like when they have the power to do so. Look at for example Ben Carson's recent proposal to have the Department of Education actively censor "extreme" political speech.
I agree. I saw the headline and thought I was going to see an interesting article about Go and computers, maybe something explaining in detail how Monte Carlo methods have made Go programs much more successful than the older brute force used to be. But nope. Maybe someone should write a program in Go to play Go for maximum confusion.
Yes and no. There is some advantage to getting close to real time data: there's a a Supernova Early Warning System http://snews.bnl.gov/. This isn't a safety issue, but rather an astronomy issue.
Detectors like IceCube can be used to actually detect the neutrinos from a supernova before the supernova's light reaches Earth. This isn't due to the erroneous claim from a few years ago that neutrinos travel faster than light, but rather because when a supernova occurs, the light from the core of the star takes multiple hours to get out of the core because of all the mass in the way, while the neutrinos aren't slowed down by this almost at all. This means that the neutrinos effectively get a few hours head start on the light- since they are traveling so close to the speed of light, they get to keep almost all this head start by the time they reach Earth. In the case of SN 1987A https://en.wikipedia.org/wiki/SN_1987A, a supernova in 1987 which was close enough that we could detect the flood of neutrinos, the neutrinos did as predicted arrive a few hours before the light. This means we can if we detect a neutrino burst and can get its directional data (which IceCube can approximately do) then we can point our telescopes at a supernova *before the light arrives at Earth* which means we'll get to see the very beginning of the supernova and hopefully get a much better understanding.
In order to do this you have to do at least some of your processing in at least close to real time as you can. This is especially important because it isn't actually easy to figure out from the neutrino burst what direction the supernova is coming from, and IceCube is one of the few detectors which gets any good directional data at all, so if this happens we want to process the data rapidly enough to get a good idea of where to look.
Note that there aren't any crypto systems which are proveably safe from quantum attacks, only systems where we strongly suspect them to be safe. At this point, other than one-time pads, we can't prove any system is even safe in a classicial setting. For example, RSA and Diffie-Hellman both rely on conjectures which are strictly stronger than P != NP.
Scott Aaronson has an excellent summary of this research on his blog: http://www.scottaaronson.com/b... One point that Scott makes that is easy to lose track of is how much working this out required people on both the theoretical crypto end and the practical crypto end to work together. This is a combination of multiple vulnerabilities and some clever number theory.
What probability do you estimate that Biden will be the nominee?
D-Wave's claimed quantum computers depend very much on what you call a quantum computer. D-Waves machines use a form of quantum annealing https://en.wikipedia.org/wiki/Quantum_annealing but they are not a universal quantum computer in the traditional sense, and even for quantum annealing they are very limited in what they can do and it isn't even clear that the problems that the D-Wave machine can do are any problems where we should expect any actual speedup from a quantum computer, and certainly the D-Wave machines have no capability for doing many of the problems we do want to use quantum computers for like factoring large integers.
SETH is a very big assumption, much stronger than even assuming P != NP. Essentially, the exponential time hypothesis (ETH) says that any algorithm which solves instances of 3-SAT https://en.wikipedia.org/wiki/3-SAT will have worst case running time that is exponential. However, it is conceivable that one could have a whole series of algorithms which each solves 3-SAT better and better. The idea is that there's an algorithm a_i which solves 3-SAT in time O(x_i^n) and as i goes to infinity, x_i goes to 1. SETH is the hypothesis that says essentially that that doesn't happen. Many people are not convinced that ETH is true although it certainly looks plausible. I think most people who have thought about this consider the possibility that ETH holds and SETH doesn't hold to be a deeply weird and generally unlikely option.
This is incredibly confused. 103 was the *maximum recorded lifespan.* What matters is the *average lifespan.* It is true that if one takes into account improvements in infant mortality the jump in life expectancy hasn't been as large http://www.ancient-origins.net/news-evolution-human-origins/life-expectancy-myth-and-why-many-ancient-humans-lived-long-077889. But even given that, life expectancy on average has gone up by about a decade in the US in the last 200 years even if one only works at people surviving past infancy.
No. Commercial speech is in general protected- the line it crosses to being non-protected is when money changes hands under false pretenses. At that point, it isn't a speech issue, but is fraud.
No. Commercial issues are distinct. But in general saying "This Foo is from the Ming dynasty" when it isn't should not be and is not illegal by itself. Selling it under false pretenses is correctly illegal. Notice that what is happening here involves not just speech but money changing hands.
*Raises hand* I don't know what hate speech is, but I'm not in favor of laws censoring speech, and that includes Holocaust denial. My grandmother had a number on her arm. Many people on my mother's side of the family were killed in the Holocaust. Holocaust denial is disgusting and these people spouting this denial are in the very best case reprehensible fucktards. That doesn't mean they don't have a right to spout their reprehensible fucktardery.
Unfortunately, while some do, many loud and prominent ones do not. Greenpeace is the most obvious example. See especially their opposition to ITER: http://www.greenpeace.org/international/en/press/releases/ITERprojectFrance/, http://www.greenpeace.org/international/en/news/Blogs/nuclear-reaction/lockheed-martins-compact-nuclear-reactor-yet-/blog/51074/, http://www.theregister.co.uk/2008/10/22/fusion_greenpeace_no/.
The Sierra Club which is in many ways more moderate than Greenpeace weakly opposes such fusion also http://www.sierraclub.org/policy/energy/nuclear-power, and while their main argument is that it is too expensive compared to more conventional renewables, they also cite "The dangers posed by the probable releases of tritium used by fusion plants, the problems with decommissioning these plants" which only makes sense if you both don't fully understand how little tritium is being used and how think that the plants will be highly radioactive like conventional fission plants.
Sortir du nucléaire, one of the major French anti-nuclear groups are basically treating ITER and fusion in general very close to how they treat fission power. See e.g. http://www.dw.com/en/france-wins-nuclear-fusion-plant/a-1631650
The environmental movement has done a lot of good and continues to do a lot of good. But there is a definite anti-technology bent in some parts and general anti-nuclear bent which is very unfortunate. There are some environmentalists who understand the potential benefits of fusion and how it is different than fission power, but it is definitely not all of them and certainly doesn't include some of the most prominent organizations.
If you have 100 studies you are replicating, by sheer chance you are likely still going to have a few who you successfully replicate but aren't real. So the problem may be worse than that (slightly). Psychology isn't the only field with these issues. There have been a lot of problems in medicine also. See https://www.newscientist.com/article/mg21528826-000-is-medical-science-built-on-shaky-foundations/. Part of the problem is one of incentives: the incentives to do a study which simply replicates a pre-existing study is low, and many journals won't even publish them. This also combines in bad ways with post-hoc analysis where you look at your data and find a pattern in it that is worth publishing; the worst offender here is medicine where people use different statistical tests and different subgroup analysis until they get a positive result.
Good points. I was aware of them,but forgot about them when writing my comment.