The latter is also going to become mandatory for all new cars in the USA and EU in 2022.
Ugh, really? I am concerned about all these things that just add failure modes (will the car only operate in limp-home mode if there is a problem with the auto-brake system?), raise barriers to entry for new vehicle companies, and remove incentives for people to have situational awareness (couple with failure modes - if people are used to auto brake, so don't pay attention, then there is a problem with auto brake, what happens?).
That last one is probably the most concerning to me - people these days have so little situational awareness as it is...
This kind of thing - at the very least the finger pointing surrounding it - is why until now nobody put "beta" heavy machinery in the hands of the general public.
The general public should never be assumed to use things as designed - not all product liability lawsuits are as frivolous as they are sometimes portrayed.
I would probably find Tesla negligent just on the grounds that they are assuming people won't abuse (or even simply misuse!) the feature. Waivers notwithstanding - it would be interesting to see those in court, because I guarantee just about everyone who signed one would have to say "I just signed it to get the shiny, I don't know what it said" if they were being honest. This means there is no evidence of expectation in the general public that these things are "beta".
Put another way: you can call something "beta" all you want in theory, but if you're selling it to the general public, it ain't in practice.
I read it more as "Google to train 2M programmers in India to program Android", not "Google to train 2M people in India who don't know how to program at all, to program Android."
Very important difference - and training a programmer in another language / library / toolkit should be pretty trivial.
The question I have is - why 2M? What's special about that number? Is there a task (or set of tasks) that really needs 2M people to achieve, or is this just "let's pick a big number"?
Eh, I think the GP has it correct: the important thing is knowing when the mathematical solution is physical and when it isn't. Take for instance the simple problem of computing the area of a walkway around a pool with a certain perimeter. One of the formulations is a quadratic equation with one positive and one negative solution for the are. Clearly "negative area" has no physical interpretation (you can't build it), so you ignore that result.
The whole "use a complex number for AC circuit analysis, and then throw out the imaginary part" has always bugged me - where does it "go"?
So for some problems, you do ignore a mathematical possibility. In others, they are acceptable. Your example of QM showing "ridiculous predictions...to be correct" is interesting because it shows how there are indeed places where we don't know what mathematical results are sensible and which aren't. But I agree that simply assuming that all mathematical results are physical is naive.
Less than that, even. You forgot the 7+% for employer's part of FICA and Medicare. You forgot whatever unemployment insurance costs in the states of operation. You forgot whatever the employer's portion of various health insurance or other benefits. Taking the general rule of thumb that employer costs are 1.25 to 1.5 times the salary, That $6.3M is only 120 to 144 jobs at $35k- so significantly fewer than your estimate even.
I think most of the complaints about executive salaries aren't really because that money could be used to pay employees more or pay more employees, because those numbers don't really add up; I think the complaint is more just in order of magnitude - 10 times might be palatable, but someone making 100 times the salary of another means that person earns effectively an entire lifetime of the lower salary in a single year.
I guess I left out a few words there - "individuals are no longer owners of their transportation capital." So maybe yeah their financial costs are lower and they have invested in financial instruments or whatever, but they are now at the mercy of the vehicle pool - that is the "something important" which was exchanged for the lower cost.
People either are never taught or choose to ignore the idea that you should never borrow to pay for a depreciating asset unless that asset helps you be more productive than you would be without it by a margin greater than the financing cost.
So financing a car that doesn't make you more productive than a car that doesn't require financing is a silly financial move.
I don't count "autonomous cars as a subscription" as a meaningful improvement either; it means individuals are no longer owners of capital and are therefore giving up something important. Unless we make it so that somehow nobody owns those things, the ownership situation is worse. And saying everyone owns something (e.g., through a co-op or something) has other economic and logistic issues, the least of which is the base-load / peak-load issue.
Consider - if your system has no unused ("wasted") capacity, it is extraordinarily susceptible to shocks. Unused capacity is sometimes worth the extra material costs.
One thing I don't get about small drone delivery: even with a capacity of "two bags of groceries" that means for N packages you are going to need N or N/2 trips for each drone. This is opposed to, what, N/100 trips for traditional delivery services?
Does anyone working in the logistics industry have any comment on how this can possibly be more efficient than traditional delivery in any way? Maybe there is something subtle and I (and many others) just can't see it...
Even better would be to make the patent period relative rather than a fixed nominal time, to make it future proof. I would suggest that a patent be valid for three median product half-cycle times for participants in the industry, up to a maximum of 20 years.
So, for (say) smart phones, when the product cycle is typically about one year, you get a patent protection of 1.5 years. For automotive where the product cycle is 5 years, you get 7.5, etc. As industries get faster or slow down, the patent period adjusts accordingly. But if your industry is terribly slow, like, say, power plant construction, patents still cap out at 20 years.
I should clarify: I would say I have a tighter definition of 'decision', which includes "the decision maker has to have the ability to choose one path or the other given the same inputs". A computer program typically does not have the option to chose either path - a computer can only choose a single path. It is not a decision, it is an inevitable result of the inputs.
Now, if your computer program was probabilistic and given the same condition multiple times, gave different results, then I would say it is effectively making a decision. For instance, sometimes taking a lower probability of success path with a higher desirability outcome. If the computer is programmed to always take the highest-probability path, that's not a decision at all.
...a computer can make decisions far quicker and more rationally...
This is mostly an incorrect understanding of what a computer can do. A computer cannot "make decisions" or "act rationally"; all a computer can do is evaluate criteria and attempt to optimize a set of values to some goal function. All decision-making was done in the selection of the particular inputs, outputs, and optimization function. Maybe you could say this is "rational" because it can never be distracted from its task, but I doubt you can call it "making a decision".
After all, does a die "make a decision" when it comes up 5 instead of 3?
Now I think it's silly to have made this a simple majority decision and that any action should have required a 2/3 vote...
I've been thinking a lot about modern political voting schemes and and kind of majority (simple, 2/3, or otherwise) wins.
Think about the adage "the needs of the many outweigh the needs of the few." That sounds well and good, when the many are a couple orders of magnitude larger than the few. But what a lot of modern society has is where the "many" and the "few" are on the same order of magnitude - unless issues are couched in compromise - and big ones (things like this referendum, national leader elections) are never that way. They are "a small fraction of the population has a massive effect on a large fraction of the population." Even if you have a 2/3 majority - say you've a country with 60 million people voting. So 40 million people get their way, but twenty million have to live in a situation with which they may not agree. That is hardly "a few".
So what do you do in this situation? I've got to believe our systems need to stop being all-or-nothing, and instead be more fully proportional in everything. I'm almost beginning to think that democracy simply doesn't scale - once you get too many people involved, it's just untenable because of the large number effect: a small percentage of the population is still a large absolute number of people.
Consumption taxes are probably the worst type of tax if you're trying to be progressive, even if you include some kind of "necessity" exemption:
Taxes for purchases of "basic necessities" would be probated...
Who defines "basic necessity"? Is the amount based on prices in rural AR or San Francisco? "Prebates" are a terrible idea, just as are any fixed nominal money amount for, say, standard deductions or personal exemptions or UBI or "Fair Tax" style necessity exemptions. That type of exemption/rebate does not reflect actual market forces, but is instead simply fiat and so will never be without adverse wealth shifting effects.
You're forgetting that with the extra props across the wing, you need a much much smaller wing, and can have a wing with much higher aspect ratio*. The reduction in wing area and increased aspect ratio more than offsets any drag from the multiple (folded) props during cruise.
*Those props aren't there for thrust, they are for increasing flow velocity over the wings.
It's more surprising that people think C needs "fixing". C is just a toolbox with almost no restrictions - so you have to be careful using it. Once you start adding restrictions to your tools, you end up with other problems.
Put another way, C is almost the ultimate in coding freedom (probably only assembler is more free, but is not quite as portable as C) so requires lots of responsibility. Adding in checks to a language does free the programmer from responsibility but necessarily comes at the cost of some freedom (or performance, those checks aren't free).
I like how the summary also suggests that battery costs will always decrease. I would say that battery costs will decrease until battery production is inevitably consolidated, then battery prices will increase as demand increases.
I don't understand why analysts always forget that; battery production is an industry which has a fairly high barrier to entry (environmental regulation, high capital requirements, etc.) so it's not an industry in which competition can keep increasing to keep prices low as battery demand skyrockets.
It's very difficult to evaluate fit & finish and other qualitative aspects of a product when shopping online. (E.g., is the text on the box in grammatically correct language, or some botched translation?) You can't evaluate fit and feel and color of clothing online.
There are many interesting articles about how online retail actually suffers huge returns costs because of those things.
VR isn't going to help with those aspects either, because a virtual good isn't going to be the same as a real physical good.
I agree - as a consumer it would be great to see computers (and electronics in general, including software) have to provide the same support lifetimes as, say, automobiles or aircraft. (I admit I don't know how much of this is regulation and how much is de-facto in those industries; e.g., you wouldn't last long as an auto manufacturer if you repair parts for your car were unavailable after 5 years.)
It's kind of a shame that other industries have product support regulations, but software / computers don't seem to.
From a developer standpoint though, I can see this being a bit of a pain, because the trend now is so much for "disposable" short-term development cycles, rather than developing for the long-term. Part of the tradeoff between fast dev cycles and robustness. I think we've swung a bit too far to the "rapid" side of things, and need to go a little ways back to robust.
When you're stuck in traffic you look up into the sky and see all that unused space. Being limited to 2D feels silly in that situation.
You might want to read up on what happens at Oshkosh every year. That's what commuting by air would look like when everyone wants to go to/from the same place. Couple that with electric aircraft with extremely limited flight durations and the tendency of people to not refuel their cars / aircraft with the idea of a contingency situation...
Computer control isn't even going to help you there, unless the computer control allows zero deviation from a programmed start and destination so it can guarantee a no-take-off if there is insufficient fuel/charge to make that flight plus required reserves due to unforeseen traffic, weather, etc.
Slashdot Mirror
Ugh, really? I am concerned about all these things that just add failure modes (will the car only operate in limp-home mode if there is a problem with the auto-brake system?), raise barriers to entry for new vehicle companies, and remove incentives for people to have situational awareness (couple with failure modes - if people are used to auto brake, so don't pay attention, then there is a problem with auto brake, what happens?).
That last one is probably the most concerning to me - people these days have so little situational awareness as it is...
This kind of thing - at the very least the finger pointing surrounding it - is why until now nobody put "beta" heavy machinery in the hands of the general public.
The general public should never be assumed to use things as designed - not all product liability lawsuits are as frivolous as they are sometimes portrayed.
I would probably find Tesla negligent just on the grounds that they are assuming people won't abuse (or even simply misuse!) the feature. Waivers notwithstanding - it would be interesting to see those in court, because I guarantee just about everyone who signed one would have to say "I just signed it to get the shiny, I don't know what it said" if they were being honest. This means there is no evidence of expectation in the general public that these things are "beta".
Put another way: you can call something "beta" all you want in theory, but if you're selling it to the general public, it ain't in practice.
I read it more as "Google to train 2M programmers in India to program Android", not "Google to train 2M people in India who don't know how to program at all, to program Android."
Very important difference - and training a programmer in another language / library / toolkit should be pretty trivial.
The question I have is - why 2M? What's special about that number? Is there a task (or set of tasks) that really needs 2M people to achieve, or is this just "let's pick a big number"?
Eh, I think the GP has it correct: the important thing is knowing when the mathematical solution is physical and when it isn't. Take for instance the simple problem of computing the area of a walkway around a pool with a certain perimeter. One of the formulations is a quadratic equation with one positive and one negative solution for the are. Clearly "negative area" has no physical interpretation (you can't build it), so you ignore that result.
The whole "use a complex number for AC circuit analysis, and then throw out the imaginary part" has always bugged me - where does it "go"?
So for some problems, you do ignore a mathematical possibility. In others, they are acceptable. Your example of QM showing "ridiculous predictions...to be correct" is interesting because it shows how there are indeed places where we don't know what mathematical results are sensible and which aren't. But I agree that simply assuming that all mathematical results are physical is naive.
Less than that, even. You forgot the 7+% for employer's part of FICA and Medicare. You forgot whatever unemployment insurance costs in the states of operation. You forgot whatever the employer's portion of various health insurance or other benefits. Taking the general rule of thumb that employer costs are 1.25 to 1.5 times the salary, That $6.3M is only 120 to 144 jobs at $35k- so significantly fewer than your estimate even.
I think most of the complaints about executive salaries aren't really because that money could be used to pay employees more or pay more employees, because those numbers don't really add up; I think the complaint is more just in order of magnitude - 10 times might be palatable, but someone making 100 times the salary of another means that person earns effectively an entire lifetime of the lower salary in a single year.
I guess the world is rediscovering that special-purpose chips will always be faster at their special purpose than a general-purpose chip will be.
I guess I left out a few words there - "individuals are no longer owners of their transportation capital." So maybe yeah their financial costs are lower and they have invested in financial instruments or whatever, but they are now at the mercy of the vehicle pool - that is the "something important" which was exchanged for the lower cost.
People either are never taught or choose to ignore the idea that you should never borrow to pay for a depreciating asset unless that asset helps you be more productive than you would be without it by a margin greater than the financing cost.
So financing a car that doesn't make you more productive than a car that doesn't require financing is a silly financial move.
I don't count "autonomous cars as a subscription" as a meaningful improvement either; it means individuals are no longer owners of capital and are therefore giving up something important. Unless we make it so that somehow nobody owns those things, the ownership situation is worse. And saying everyone owns something (e.g., through a co-op or something) has other economic and logistic issues, the least of which is the base-load / peak-load issue.
Consider - if your system has no unused ("wasted") capacity, it is extraordinarily susceptible to shocks. Unused capacity is sometimes worth the extra material costs.
One thing I don't get about small drone delivery: even with a capacity of "two bags of groceries" that means for N packages you are going to need N or N/2 trips for each drone. This is opposed to, what, N/100 trips for traditional delivery services?
Does anyone working in the logistics industry have any comment on how this can possibly be more efficient than traditional delivery in any way? Maybe there is something subtle and I (and many others) just can't see it...
Even better would be to make the patent period relative rather than a fixed nominal time, to make it future proof. I would suggest that a patent be valid for three median product half-cycle times for participants in the industry, up to a maximum of 20 years.
So, for (say) smart phones, when the product cycle is typically about one year, you get a patent protection of 1.5 years. For automotive where the product cycle is 5 years, you get 7.5, etc. As industries get faster or slow down, the patent period adjusts accordingly. But if your industry is terribly slow, like, say, power plant construction, patents still cap out at 20 years.
I should clarify: I would say I have a tighter definition of 'decision', which includes "the decision maker has to have the ability to choose one path or the other given the same inputs". A computer program typically does not have the option to chose either path - a computer can only choose a single path. It is not a decision, it is an inevitable result of the inputs.
Now, if your computer program was probabilistic and given the same condition multiple times, gave different results, then I would say it is effectively making a decision. For instance, sometimes taking a lower probability of success path with a higher desirability outcome. If the computer is programmed to always take the highest-probability path, that's not a decision at all.
This is mostly an incorrect understanding of what a computer can do. A computer cannot "make decisions" or "act rationally"; all a computer can do is evaluate criteria and attempt to optimize a set of values to some goal function. All decision-making was done in the selection of the particular inputs, outputs, and optimization function. Maybe you could say this is "rational" because it can never be distracted from its task, but I doubt you can call it "making a decision".
After all, does a die "make a decision" when it comes up 5 instead of 3?
I've been thinking a lot about modern political voting schemes and and kind of majority (simple, 2/3, or otherwise) wins.
Think about the adage "the needs of the many outweigh the needs of the few." That sounds well and good, when the many are a couple orders of magnitude larger than the few. But what a lot of modern society has is where the "many" and the "few" are on the same order of magnitude - unless issues are couched in compromise - and big ones (things like this referendum, national leader elections) are never that way. They are "a small fraction of the population has a massive effect on a large fraction of the population." Even if you have a 2/3 majority - say you've a country with 60 million people voting. So 40 million people get their way, but twenty million have to live in a situation with which they may not agree. That is hardly "a few".
So what do you do in this situation? I've got to believe our systems need to stop being all-or-nothing, and instead be more fully proportional in everything. I'm almost beginning to think that democracy simply doesn't scale - once you get too many people involved, it's just untenable because of the large number effect: a small percentage of the population is still a large absolute number of people.
Well not quite - that $5B gigafactory gives you an idea of how much battery factory you can get, not how much research...
Back-of-the-envelope says $10B should give you about a hundred-thousand man-years of research.
Seconded.
Seriously, whatever happened to even pretending to be professional?
Consumption taxes are probably the worst type of tax if you're trying to be progressive, even if you include some kind of "necessity" exemption:
Who defines "basic necessity"? Is the amount based on prices in rural AR or San Francisco? "Prebates" are a terrible idea, just as are any fixed nominal money amount for, say, standard deductions or personal exemptions or UBI or "Fair Tax" style necessity exemptions. That type of exemption/rebate does not reflect actual market forces, but is instead simply fiat and so will never be without adverse wealth shifting effects.
You're forgetting that with the extra props across the wing, you need a much much smaller wing, and can have a wing with much higher aspect ratio*. The reduction in wing area and increased aspect ratio more than offsets any drag from the multiple (folded) props during cruise.
*Those props aren't there for thrust, they are for increasing flow velocity over the wings.
It's more surprising that people think C needs "fixing". C is just a toolbox with almost no restrictions - so you have to be careful using it. Once you start adding restrictions to your tools, you end up with other problems.
Put another way, C is almost the ultimate in coding freedom (probably only assembler is more free, but is not quite as portable as C) so requires lots of responsibility. Adding in checks to a language does free the programmer from responsibility but necessarily comes at the cost of some freedom (or performance, those checks aren't free).
I like how the summary also suggests that battery costs will always decrease. I would say that battery costs will decrease until battery production is inevitably consolidated, then battery prices will increase as demand increases.
I don't understand why analysts always forget that; battery production is an industry which has a fairly high barrier to entry (environmental regulation, high capital requirements, etc.) so it's not an industry in which competition can keep increasing to keep prices low as battery demand skyrockets.
It's very difficult to evaluate fit & finish and other qualitative aspects of a product when shopping online. (E.g., is the text on the box in grammatically correct language, or some botched translation?) You can't evaluate fit and feel and color of clothing online.
There are many interesting articles about how online retail actually suffers huge returns costs because of those things.
VR isn't going to help with those aspects either, because a virtual good isn't going to be the same as a real physical good.
I almost wonder if we'd have readily-available cures for polio, measles, etc. if they hadn't already been available in the early 20th century.
What happened to creating helpful drugs for the purpose of helping people, rather than making money?
I agree - as a consumer it would be great to see computers (and electronics in general, including software) have to provide the same support lifetimes as, say, automobiles or aircraft. (I admit I don't know how much of this is regulation and how much is de-facto in those industries; e.g., you wouldn't last long as an auto manufacturer if you repair parts for your car were unavailable after 5 years.)
It's kind of a shame that other industries have product support regulations, but software / computers don't seem to.
From a developer standpoint though, I can see this being a bit of a pain, because the trend now is so much for "disposable" short-term development cycles, rather than developing for the long-term. Part of the tradeoff between fast dev cycles and robustness. I think we've swung a bit too far to the "rapid" side of things, and need to go a little ways back to robust.
You might want to read up on what happens at Oshkosh every year. That's what commuting by air would look like when everyone wants to go to/from the same place. Couple that with electric aircraft with extremely limited flight durations and the tendency of people to not refuel their cars / aircraft with the idea of a contingency situation...
Computer control isn't even going to help you there, unless the computer control allows zero deviation from a programmed start and destination so it can guarantee a no-take-off if there is insufficient fuel/charge to make that flight plus required reserves due to unforeseen traffic, weather, etc.
This "X McXface" thing is an interesting example of the homogenization of childhood and adulthood.
This makes for an interesting idea - mandate that legal fees/penalties must always be the same percentage of each party*'s net worth.
Not sure it's a good idea, but it would avoid that kind of nonsense...
*Define 'party' as 'all owners' to avoid hiding things in a shell company with zero assets.