Thanks for that analysis. I bow to your superior knowledge.
Humans for safety: "We're talking about miles of track that cross public roadways with children on bikes." The human on board can't do anything useful if a kid crosses in front of their train (it takes hundreds of meters to stop), so this one is a non-issue.
So ideally for a freight rail system we want high throughput, short delivery times, cheap, and running to/from convenient nearby locations.
Breaking this down further, it suggests we want * Small trains (lowers latency - less time to wait for a train going to your destination. Removes/reduces need for transferring cargo between trains by allowing point-to-point service, so long as the 'point's are train stations.) * Autonomous (required by 'small trains' and 'cheap') * Handles congestion well (for high throughput with lots of small trains) * Fast * Moderately priced infrastructure. * High density of train stations around the country I.e. something like an internet for shipping containers.
Hyperloop gives us 'fast', but fails on infrastructure price, fails at least initially on density of stations, and congestion may be problematic. Starting with the existing rail network and moving to more automation and smaller trains and solving some congestion problems (perhaps the hardest bit) gives everything but 'fast', but for many purposes is 'fast enough'.
It still needs to be competitive compared to autonomous trucks.
With a bit more research: Compared to 1080, 1050 has 2/5 as many cores and about 5/8 the clock speed. 1080 has design thermal power of 180W. I don't remember if power is proportional to clock, but if it is, 1050 should have about 1/4 the power draw of a 1080, which puts it at 45W which won't require a power connector and is easy to passively cool, but possibly passive cards won't be available at launch. 1040 would be about 28W (expect fanless to be the norm), and 1060 about 62W (where a power connector might be required, and we might eventually get passive cooling but not real soon.)
If power is a higher power of clock (eg. clock speed squared) then the numbers get even better.
I like my computers very quiet, so my rule of thumb (sometimes violated) is buy the best GPU available which is passively cooled and needs no extra power connector.
I only found one page about the GTX 1050 or GTX 1040. This gives expected release date 2016Q3. However they don't give power consumption (critical for my purposes - I'd be looking for a maximum of about 60W) nor do the numbers they quote give me much idea of how much faster it will be than (say) a GTX 750, which so far as I know is the current best quiet GPU.
The landing is at about 9:10 here but there isn't very much to see: it was a night landing from a nearby camera - the moment of landing is invisible in glare. You get to see the glare, then it fades to reveal the landed rocket.
This mission seems very hard to justify from a commercial view point.
Wikipedia says "As of May 2012, SpaceX had operated on total funding of approximately $1 billion in its first ten years of operation. Of this, private equity provided about $200M, with Musk investing approximately $100M and other investors having put in about $100M."
So (as of four years ago) Musk only owns about 50% of SpaceX, so it isn't his plaything to do with as he wishes. How is this squared with the other investors?
Oops, some 'less than' signs got eaten by HTML. LLO to Earth intercept: < 1.3km/s (then you can aerobrake and re-enter) Total delta-v from LEO: < 6.4 km/s
I've speculated before on/. about how much effort/cost it would be for SpaceX to do a manned moon mission.
If they can do this Mars mission (landing on Mars but not returning), exactly the same hardware can do a lunar landing and return. From here we see for Mars mission: Delta-v LEO to Mars transfer typical value 4.3 km/s Transfer orbit to Mars capture orbit 0.9 km/s Capture orbit to low orbit 1.4 km/s Low orbit to surface 4.1 km/s Total delta-v from LEO: 10.7 km/s
Lunar mission: LEO to low lunar orbit: 1.3 km/s Low lunar orbit to surface: 1.9 km/s Surface to LLO: 1.9 km/s LLO to Earth intercept: 1.3km/s (then you can aerobrake and re-enter) Total delta-v from LEO: 6.4 km/s
Dragon 2 will have life support, and carry people. Red Dragon presumably won't. So to make it a manned mission they would need to reduce Red Dragon's delta-v capability by about 4-5km/s and use the saved weight to put in Dragon 2's manned capabilities (possibly for fewer people than Dragon 2.) Possibly (probably?) Dragon 2's life support requires a service module, and we haven't budgeted for that, so maybe it is a bit more complicated.
An unmanned Red Dragon to the moon seems a sensible step before Mars - you get to find out about mistakes in a few days, instead of nearly 2 years.
Pournelle: "New York Times best sellers [which his was] will get you through times of no Hugos better than Hugos will get you through times of no best sellers."
There was a quote from a counter-culture comic book The Fabulous Furry Freak Brothers "Dope will get you through times of no money better than money will get you through times of no dope."
Clearly these are related. Does anyone know the origin?
My first thought was "CO2 as a refrigerant - its kind of toxic isn't it? I wouldn't want to be around if a pipe broke."
Then I thought "Ammonia is also used as a commercial refrigerant, and that is also toxic. Which is worse?"
I haven't found any good answer online. Nobody seems to want to talk about toxic concentrations of ammonia in air, just in blood. Then there are all sorts of other complications - what quantities and pressures would be used for comparable CO2 and NH3 refrigeration plants? Does the lower density of NH3 mean it will disperse faster? Are you a whole lot worse off after being nearly killed by NH3 than after being nearly killed by CO2? Is CO2 more likely to take you by surprise, so you don't realize your danger until it is too late?
In summary - is it better to be near a catastrophically failed NH3 or CO2 refrigeration plant? What about other refrigerants used at a similar scale?
So if I'm a retiree with a budget of $350/week = $18200/year, $1600/year for both ears is nearly 10% of my budget just to be able to hear. Lots of people already can't afford cable and can't afford a car. Lets not add can't afford to hear to the list, when adequate solutions are available for 1/10th the price.
If your plan is something more like shared taxis than traditional buses (say 10-15 seats and door-to-door service, pickup on demand within 15 minutes) then there are many fewer passengers to share the cost of the driver, and I imagine it would be a significant part of the cost.
I don't know if this is what Musk has in mind, as the fine article is nearly contentless. (If so, it is hardly an exciting new idea.) I think that a service like this would get many people to go carless. It would do for me.
Case-as-a-heatsink does exist, but you need a clever system of heat pipes to get the heat from where it is generated (CPU, GPU) to the case. If the motherboard is not an integral part of the package, you need to be even more clever to have adjustablity in the heat pipes to accommodate different CPU, GPU locations. It makes for difficult installs, hard to maintain, and such cases cost at least about $500 over standard case costs. Generally such cases only get used in hostile environments (where the air will damage the computer) or super-noise-sensitive environments, such as a recording studio.
Also, I agree with other commenters - many large fans is good, not bad, for quiet. You can get the needed airflow by running them very slowly.
(This knowledge comes from being a long-time reader of http://www.silentpcreview.com/ . I've never met such a case in real life.)
The problem is it is a very expensive pudding. Instead of something you can plonk in a vacuum chamber and attach a few wires and detectors to, you need something which can survive the rigours of launch, and autonomously power itself and communicate, plus you have to pay launch costs (even if they have spare payload capacity on a launch, there will be costs for integrating your package and its deployment system, and SpaceX probably want some profit on top of that.)
I'm guessing here, but I'd think the earthbound experiment cost on the order of $20k to $50k (mostly in engineer salaries) but the orbital version would cost at least $1M. Not only does that investment need to have expected return greater than $1M ((value of discovering it works) times (probability it works)), at least in the short term it needs better payback than doing more earthbound experiments (which potentially might tell you that the space probe is a waste of money.)
Someone may well make the probe, and it has the potential to produce results which would prove the effect to the doubters (of which I am one) but it is not cheap or easy. Indeed, this really is the definitive experiment. Either the effect will be disproved on the ground, or one day there will be such a probe (or just maybe people will lose interest prior to there being clear result one way or the other, but I doubt it.)
My problem was that my very slight knowledge of Japanese provided me with two translations for 'hon' - 'book' or 'origin'. Neither seemed to fit very well, so I didn't know if one was right (and there was some connection I was missing) or (as it turned out) there was some other source of 'hon'.
'SpaceX does not "already" have one. They definitely didn't when SLS started.'
I agree with the second point. However Falcon 9 Heavy is currently scheduled for first launch in November, so unless this slips* or fails, by the time the next administration comes in (the timing implied by TFA's quote) they will.
* SpaceX (and Tesla) are known for missing their release schedules, so by past performance a slip is quite likely.
From TFA: "The next president, or some in Congress, may begin asking why NASA is spending billions to develop its own heavy-lift rocket when SpaceX already has one."
As I recall, within about a year of taking office, Obama tried to kill the SLS (Nasa's heavy rocket) on this reasoning (that private companies could do the job better, given the chance, and secondarily that the funding NASA was getting for SLS was insufficient to achieve anything in a reasonable time frame) but Congress resurrected it.
Can anyone who has followed this more closely comment on the political history of COTS, and in particular the attitude of Bush and then Obama, and Congress/Senate, to COTS and SLS?
I have been wondering lately: How much extra effort would it take SpaceX to do a manned moon mission?
They will soon have most of the pieces. Two Falcon 9 Heavies (first flight planned for November) have nearly the payload of a Saturn V, the manned Dragon (first flight planned early-mid next year) would work as a command module*. Missing is a lander, and possibly a stage to transfer from low Earth orbit to lunar orbit, which could probably be a lightly modified stage 2 of the existing Falcon 9.
The 'only' expensive bit still to do then is the lander. Not coincidentally, this is also the only bit which has no other use, so money spent on design has no future pay-back (except for more moon missions.) I don't know how much this would cost, but I doubt it would be under half a billion dollars, likely much more.
*The Dragon is about the same diameter as the Apollo module, but about twice as long. I don't know how it compares by mass - it is bigger, but modern materials are lighter.
They'll have people on the barge within perhaps 30 minutes of landing. They can hose stuff down with fresh water, then put a tarpaulin over the engines. All the delicate stuff already has to be protected from hypersonic travel through atmosphere, so is probably pretty well protected.
For the Dragon capsule, they're still intending propulsive landing. To me this seems a less clear cut decision than for stage I.
When manned, the Dragon needs both parachutes and rockets whatever landing scenario you have in mind, because you need to allow for abort on launch (rockets to get away from misbehaving lower stages, parachutes to land safely.) A parachute landing is likely to have a sideways velocity component, but this is much less critical for the short Dragon than the very tall stage I.
I'd have imagined a landing scenario where you use the chutes, and just a little bit of rocket right at the end to cushion the impact. (Chutes get released either shortly before or immediately after landing, so as to not get dragged sideways by the wind.)
Perhaps it is a matter of precision: my scenario works fine if you have lots of empty tundra to land in, not so well if you have a hundred metre landing pad to aim for.
The payload cost of reusability is much more that you are saying. From wikipedia:
In order to make the Falcon 9 reusable and return to the launch site, extra propellant and landing gear must be carried on the first stage, requiring around a 30 percent reduction of the maximum payload to orbit in comparison with the expendable Falcon 9."
(That is a return-to-launch-site cost, not a land-on-barge cost, which I didn't find a value for and must be a fair bit lower.)
SpaceX has indicated that the Falcon Heavy payload performance to geosynchronous transfer orbit (GTO) will be reduced due to the addition of the reusable technology, but would fly at much lower launch price. With full reusability on all three booster cores, GTO payload will be 7,000 kg (15,000 lb). If only the two outside cores fly as reusable cores while the center core is expendable, GTO payload would be approximately 14,000 kg (31,000 lb).[39] "Falcon 9 will do satellites up to roughly 3.5 tonnes, with full reusability of the boost stage, and Falcon Heavy will do satellites up to 7 tonnes with full reusability of the all three boost stages," [Musk] said, referring to the three Falcon 9 booster cores that will comprise the Falcon Heavy's first stage. He also said Falcon Heavy could double its payload performance to GTO "if, for example, we went expendable on the center core."
(Not clear if this is return to launch site or land on barge.)
Note that I'm not saying this makes it uneconomic. If you have a source to show that contingency fuel (e.g. to accommodate a failure of one engine during launch) is of the same order as fuel requirement to land the stage, please share it. (Also, I agree with your comments about salt water and horizontal loading.)
Picking the only named engineer in the article (actually the in the Sugarvolt video), Marty Bradley has a PhD in aerospace engineering and a high profile career at one of the worlds best aerospace companies, with a whole bunch of other super smart engineers looking over his shoulder to ensure he doesn't waste company money. He has studied the idea in detail, and thinks it might work. You've read a pop-engineering article, and on that basis you call Dr Bradley an idiot.
It looks like a hard problem. I don't see how it could work. But my degrees in maths and physics are nothing to his education and experience in this field. If you happen to be an experienced aerospace engineer who has looked in detail at the proposals, you've earned the right to call him an idiot (if the proposals are, in fact, idiotic.) Otherwise acknowledge your own ignorance, and be a whole lot more humble.
After being very anti-nuclear in my youth, I've moved rather in response to global climate change.
I think that nuclear power (and waste disposal) can be done safely and with much lower environmental impact than burning fossil fuels.
However 'can be done' is not the same as 'will be done'. The knowledge and technology to avoid Fukushima was there, but wasn't used. The sea wall could and should have been higher (another power station up the coast survived because their safety engineer fought upper management to build a higher wall.) The backup generators could and should have been above the flood level. Provision could and should have been made to safely vent hydrogen, which would have greatly reduced the impact of the accident. Avoiding very rare high impact events is hard, because there is always the temptation to do less because what you've been doing seems to have been working fine so far. There are many cases of airlines which cut back on maintenance and it seemed to be working fine until the crash happened.
Also 'can be done' is not the same as 'can be done economically'. Nuclear power seems to have a big problem here, but I don't know enough to judge how much is real economics and how much is politics.
I think the way of the future will be intermittent renewable sources (wind, solar, perhaps tides) combined with large scale power storage. It is not clear whether this will happen soon enough that we don't need nuclear to bridge the gap.
Slashdot Mirror
Thanks for that analysis. I bow to your superior knowledge.
Humans for safety: "We're talking about miles of track that cross public roadways with children on bikes." The human on board can't do anything useful if a kid crosses in front of their train (it takes hundreds of meters to stop), so this one is a non-issue.
So ideally for a freight rail system we want high throughput, short delivery times, cheap, and running to/from convenient nearby locations.
Breaking this down further, it suggests we want
* Small trains (lowers latency - less time to wait for a train going to your destination. Removes/reduces need for transferring cargo between trains by allowing point-to-point service, so long as the 'point's are train stations.)
* Autonomous (required by 'small trains' and 'cheap')
* Handles congestion well (for high throughput with lots of small trains)
* Fast
* Moderately priced infrastructure.
* High density of train stations around the country
I.e. something like an internet for shipping containers.
Hyperloop gives us 'fast', but fails on infrastructure price, fails at least initially on density of stations, and congestion may be problematic. Starting with the existing rail network and moving to more automation and smaller trains and solving some congestion problems (perhaps the hardest bit) gives everything but 'fast', but for many purposes is 'fast enough'.
It still needs to be competitive compared to autonomous trucks.
TL;DR: I agree.
With a bit more research:
Compared to 1080, 1050 has 2/5 as many cores and about 5/8 the clock speed. 1080 has design thermal power of 180W. I don't remember if power is proportional to clock, but if it is, 1050 should have about 1/4 the power draw of a 1080, which puts it at 45W which won't require a power connector and is easy to passively cool, but possibly passive cards won't be available at launch. 1040 would be about 28W (expect fanless to be the norm), and 1060 about 62W (where a power connector might be required, and we might eventually get passive cooling but not real soon.)
If power is a higher power of clock (eg. clock speed squared) then the numbers get even better.
I like my computers very quiet, so my rule of thumb (sometimes violated) is buy the best GPU available which is passively cooled and needs no extra power connector.
I only found one page about the GTX 1050 or GTX 1040. This gives expected release date 2016Q3. However they don't give power consumption (critical for my purposes - I'd be looking for a maximum of about 60W) nor do the numbers they quote give me much idea of how much faster it will be than (say) a GTX 750, which so far as I know is the current best quiet GPU.
The landing is at about 9:10 here but there isn't very much to see: it was a night landing from a nearby camera - the moment of landing is invisible in glare. You get to see the glare, then it fades to reveal the landed rocket.
This mission seems very hard to justify from a commercial view point.
Wikipedia says
"As of May 2012, SpaceX had operated on total funding of approximately $1 billion in its first ten years of operation. Of this, private equity provided about $200M, with Musk investing approximately $100M and other investors having put in about $100M."
So (as of four years ago) Musk only owns about 50% of SpaceX, so it isn't his plaything to do with as he wishes. How is this squared with the other investors?
Oops, some 'less than' signs got eaten by HTML.
LLO to Earth intercept: < 1.3km/s (then you can aerobrake and re-enter)
Total delta-v from LEO: < 6.4 km/s
I've speculated before on /. about how much effort/cost it would be for SpaceX to do a manned moon mission.
If they can do this Mars mission (landing on Mars but not returning), exactly the same hardware can do a lunar landing and return. From here we see for Mars mission:
Delta-v LEO to Mars transfer typical value 4.3 km/s
Transfer orbit to Mars capture orbit 0.9 km/s
Capture orbit to low orbit 1.4 km/s
Low orbit to surface 4.1 km/s
Total delta-v from LEO: 10.7 km/s
Lunar mission:
LEO to low lunar orbit: 1.3 km/s
Low lunar orbit to surface: 1.9 km/s
Surface to LLO: 1.9 km/s
LLO to Earth intercept: 1.3km/s (then you can aerobrake and re-enter)
Total delta-v from LEO: 6.4 km/s
Dragon 2 will have life support, and carry people. Red Dragon presumably won't. So to make it a manned mission they would need to reduce Red Dragon's delta-v capability by about 4-5km/s and use the saved weight to put in Dragon 2's manned capabilities (possibly for fewer people than Dragon 2.) Possibly (probably?) Dragon 2's life support requires a service module, and we haven't budgeted for that, so maybe it is a bit more complicated.
An unmanned Red Dragon to the moon seems a sensible step before Mars - you get to find out about mistakes in a few days, instead of nearly 2 years.
Pournelle: "New York Times best sellers [which his was] will get you through times of no Hugos better than Hugos will get you through times of no best sellers."
There was a quote from a counter-culture comic book The Fabulous Furry Freak Brothers
"Dope will get you through times of no money better than money will get you through times of no dope."
Clearly these are related. Does anyone know the origin?
My first thought was "CO2 as a refrigerant - its kind of toxic isn't it? I wouldn't want to be around if a pipe broke."
Then I thought "Ammonia is also used as a commercial refrigerant, and that is also toxic. Which is worse?"
I haven't found any good answer online. Nobody seems to want to talk about toxic concentrations of ammonia in air, just in blood. Then there are all sorts of other complications - what quantities and pressures would be used for comparable CO2 and NH3 refrigeration plants? Does the lower density of NH3 mean it will disperse faster? Are you a whole lot worse off after being nearly killed by NH3 than after being nearly killed by CO2? Is CO2 more likely to take you by surprise, so you don't realize your danger until it is too late?
In summary - is it better to be near a catastrophically failed NH3 or CO2 refrigeration plant? What about other refrigerants used at a similar scale?
Here's an NH3 refrigerant accident: http://www.reuters.com/article...
And here's a CO2 one: http://www.fluorocarbons.org/m...
Interestingly this last link refers to CO2's 'low toxicity'.
So if I'm a retiree with a budget of $350/week = $18200/year, $1600/year for both ears is nearly 10% of my budget just to be able to hear. Lots of people already can't afford cable and can't afford a car. Lets not add can't afford to hear to the list, when adequate solutions are available for 1/10th the price.
For hearing, the pain threshold is well above the damage threshold. Painfully loud is way too loud and then a bunch more.
If your plan is something more like shared taxis than traditional buses (say 10-15 seats and door-to-door service, pickup on demand within 15 minutes) then there are many fewer passengers to share the cost of the driver, and I imagine it would be a significant part of the cost.
I don't know if this is what Musk has in mind, as the fine article is nearly contentless. (If so, it is hardly an exciting new idea.) I think that a service like this would get many people to go carless. It would do for me.
Case-as-a-heatsink does exist, but you need a clever system of heat pipes to get the heat from where it is generated (CPU, GPU) to the case. If the motherboard is not an integral part of the package, you need to be even more clever to have adjustablity in the heat pipes to accommodate different CPU, GPU locations. It makes for difficult installs, hard to maintain, and such cases cost at least about $500 over standard case costs. Generally such cases only get used in hostile environments (where the air will damage the computer) or super-noise-sensitive environments, such as a recording studio.
Also, I agree with other commenters - many large fans is good, not bad, for quiet. You can get the needed airflow by running them very slowly.
(This knowledge comes from being a long-time reader of http://www.silentpcreview.com/ . I've never met such a case in real life.)
The problem is it is a very expensive pudding. Instead of something you can plonk in a vacuum chamber and attach a few wires and detectors to, you need something which can survive the rigours of launch, and autonomously power itself and communicate, plus you have to pay launch costs (even if they have spare payload capacity on a launch, there will be costs for integrating your package and its deployment system, and SpaceX probably want some profit on top of that.)
I'm guessing here, but I'd think the earthbound experiment cost on the order of $20k to $50k (mostly in engineer salaries) but the orbital version would cost at least $1M. Not only does that investment need to have expected return greater than $1M ((value of discovering it works) times (probability it works)), at least in the short term it needs better payback than doing more earthbound experiments (which potentially might tell you that the space probe is a waste of money.)
Someone may well make the probe, and it has the potential to produce results which would prove the effect to the doubters (of which I am one) but it is not cheap or easy. Indeed, this really is the definitive experiment. Either the effect will be disproved on the ground, or one day there will be such a probe (or just maybe people will lose interest prior to there being clear result one way or the other, but I doubt it.)
My problem was that my very slight knowledge of Japanese provided me with two translations for 'hon' - 'book' or 'origin'. Neither seemed to fit very well, so I didn't know if one was right (and there was some connection I was missing) or (as it turned out) there was some other source of 'hon'.
TFA doesn't say what the name means, so after a few minutes on this internet thingy I find:
The robot is named RoboHon, which amalgamates the Japanese words for "robot" and "phone" together.
and an online dictionary gives "terehon", "terefon", "denwa" as translations of "telephone".
So it is named "robophone", slightly mangled by transliteration.
'SpaceX does not "already" have one. They definitely didn't when SLS started.'
I agree with the second point. However Falcon 9 Heavy is currently scheduled for first launch in November, so unless this slips* or fails, by the time the next administration comes in (the timing implied by TFA's quote) they will.
* SpaceX (and Tesla) are known for missing their release schedules, so by past performance a slip is quite likely.
From TFA: "The next president, or some in Congress, may begin asking why NASA is spending billions to develop its own heavy-lift rocket when SpaceX already has one."
As I recall, within about a year of taking office, Obama tried to kill the SLS (Nasa's heavy rocket) on this reasoning (that private companies could do the job better, given the chance, and secondarily that the funding NASA was getting for SLS was insufficient to achieve anything in a reasonable time frame) but Congress resurrected it.
Can anyone who has followed this more closely comment on the political history of COTS, and in particular the attitude of Bush and then Obama, and Congress/Senate, to COTS and SLS?
I have been wondering lately: How much extra effort would it take SpaceX to do a manned moon mission?
They will soon have most of the pieces. Two Falcon 9 Heavies (first flight planned for November) have nearly the payload of a Saturn V, the manned Dragon (first flight planned early-mid next year) would work as a command module*. Missing is a lander, and possibly a stage to transfer from low Earth orbit to lunar orbit, which could probably be a lightly modified stage 2 of the existing Falcon 9.
The 'only' expensive bit still to do then is the lander. Not coincidentally, this is also the only bit which has no other use, so money spent on design has no future pay-back (except for more moon missions.) I don't know how much this would cost, but I doubt it would be under half a billion dollars, likely much more.
*The Dragon is about the same diameter as the Apollo module, but about twice as long. I don't know how it compares by mass - it is bigger, but modern materials are lighter.
All the following is speculation.
They'll have people on the barge within perhaps 30 minutes of landing. They can hose stuff down with fresh water, then put a tarpaulin over the engines. All the delicate stuff already has to be protected from hypersonic travel through atmosphere, so is probably pretty well protected.
For the Dragon capsule, they're still intending propulsive landing. To me this seems a less clear cut decision than for stage I.
When manned, the Dragon needs both parachutes and rockets whatever landing scenario you have in mind, because you need to allow for abort on launch (rockets to get away from misbehaving lower stages, parachutes to land safely.) A parachute landing is likely to have a sideways velocity component, but this is much less critical for the short Dragon than the very tall stage I.
I'd have imagined a landing scenario where you use the chutes, and just a little bit of rocket right at the end to cushion the impact. (Chutes get released either shortly before or immediately after landing, so as to not get dragged sideways by the wind.)
Perhaps it is a matter of precision: my scenario works fine if you have lots of empty tundra to land in, not so well if you have a hundred metre landing pad to aim for.
The payload cost of reusability is much more that you are saying. From wikipedia:
In order to make the Falcon 9 reusable and return to the launch site, extra propellant and landing gear must be carried on the first stage, requiring around a 30 percent reduction of the maximum payload to orbit in comparison with the expendable Falcon 9."
(That is a return-to-launch-site cost, not a land-on-barge cost, which I didn't find a value for and must be a fair bit lower.)
And from here:
SpaceX has indicated that the Falcon Heavy payload performance to geosynchronous transfer orbit (GTO) will be reduced due to the addition of the reusable technology, but would fly at much lower launch price. With full reusability on all three booster cores, GTO payload will be 7,000 kg (15,000 lb). If only the two outside cores fly as reusable cores while the center core is expendable, GTO payload would be approximately 14,000 kg (31,000 lb).[39] "Falcon 9 will do satellites up to roughly 3.5 tonnes, with full reusability of the boost stage, and Falcon Heavy will do satellites up to 7 tonnes with full reusability of the all three boost stages," [Musk] said, referring to the three Falcon 9 booster cores that will comprise the Falcon Heavy's first stage. He also said Falcon Heavy could double its payload performance to GTO "if, for example, we went expendable on the center core."
(Not clear if this is return to launch site or land on barge.)
Note that I'm not saying this makes it uneconomic. If you have a source to show that contingency fuel (e.g. to accommodate a failure of one engine during launch) is of the same order as fuel requirement to land the stage, please share it. (Also, I agree with your comments about salt water and horizontal loading.)
Picking the only named engineer in the article (actually the in the Sugarvolt video), Marty Bradley has a PhD in aerospace engineering and a high profile career at one of the worlds best aerospace companies, with a whole bunch of other super smart engineers looking over his shoulder to ensure he doesn't waste company money. He has studied the idea in detail, and thinks it might work. You've read a pop-engineering article, and on that basis you call Dr Bradley an idiot.
It looks like a hard problem. I don't see how it could work. But my degrees in maths and physics are nothing to his education and experience in this field. If you happen to be an experienced aerospace engineer who has looked in detail at the proposals, you've earned the right to call him an idiot (if the proposals are, in fact, idiotic.) Otherwise acknowledge your own ignorance, and be a whole lot more humble.
After being very anti-nuclear in my youth, I've moved rather in response to global climate change.
I think that nuclear power (and waste disposal) can be done safely and with much lower environmental impact than burning fossil fuels.
However 'can be done' is not the same as 'will be done'. The knowledge and technology to avoid Fukushima was there, but wasn't used. The sea wall could and should have been higher (another power station up the coast survived because their safety engineer fought upper management to build a higher wall.) The backup generators could and should have been above the flood level. Provision could and should have been made to safely vent hydrogen, which would have greatly reduced the impact of the accident. Avoiding very rare high impact events is hard, because there is always the temptation to do less because what you've been doing seems to have been working fine so far. There are many cases of airlines which cut back on maintenance and it seemed to be working fine until the crash happened.
Also 'can be done' is not the same as 'can be done economically'. Nuclear power seems to have a big problem here, but I don't know enough to judge how much is real economics and how much is politics.
I think the way of the future will be intermittent renewable sources (wind, solar, perhaps tides) combined with large scale power storage. It is not clear whether this will happen soon enough that we don't need nuclear to bridge the gap.