Re:Education decaying into retold legends of glory
on
Why We Fight
·
· Score: 1
Clear from the article? The only thing that's clear from the article is that the individual who wrote the article in question is lacking in clue. I've never been much one for the "Things would be better if I was dictator" meme, and I've seen plenty of shows on The History Channel that did not have anything to do with war, although the history of cheese is something Alton Brown would have on his show.
To the players, the appeal behind war games isn't because we all secretly yearn for some dictator to come and take over our lives, it's because games are about competition and war is as about as competitive as an activity gets. From the perspective of the publisher, a war game's appeal is similar to the reason that war novels are popular with writers: You get lots of conflict to drive the plot without needing a lot of justification or exposition.
Among other things, IPv6 is intended to be used for network auto-configuration. To that end, every machine is capable of coming up on a LAN and talking to all the other machines on that LAN without any configuration at all. It does this by using so-called "link-local" addresses. The thing is, since those link local addresses are automagically configured, there's no way to guarantee their uniqueness on the Internet as a whole, so they aren't supposed to be routed. In fact, they can't be routed because there's no way for a router to tell which network to forward an incoming packet to because all of the outbounds have addresses in fe00::/8
Routeable addresses are supposed to be assigned to TLA's (Top-Level Aggregators) and they portion them out to their customers and they portion them out to their customers and so on until you get your allocation. As far as I am aware, with the exception of the 3ffe::/16 pTLA (psuedo Top Level Aggregator) assignments that were made for the 6bone, and which are now well on their way to being phased out so they're now almost entirely of historical interest, all of the routeable IPv6 addresses are in 2000::/2, so if the first digit in your IPv6 address is not a 2, then you don't have IPv6 connectivity.
As an example, I've gone to the trouble of setting my network as 2001:5c0:8305::/48, but I don't (yet) allow incoming connections to my network. However, all of the interefaces on all of my machines have addresses in the fe80::/16 block even though I've got three physically separate LAN's at the house.
What DSL routers or CableModems work with IPv6? It doesn't matter if I can buy a OC-512 with IPv6 if "Joe Sixpack" can't get it through his cablemodem. Sure I can get (and, in fact, have gotten) an IPv6 tunnel for my network, but that means that my IPv6 throughput sucks.
Since demand for addresses necessarily comes from the leaf nodes of the network (where the bulk of them are consumed) rather than the backbones, I think it is disingenuous (to say the least) to claim that IPv6 is already "rolled out" because it is available from various backbone providers when the reality is that it is not available directly to the end users.
he wants to know which method is faster in general.
That's like asking "What's the sound of yellow". "In general" the question is not answerable. In my opinion, the question is also misguided. This sort of thing is what Hoare was talking about when he said "Premature optimization is the root of all evil". Making the program understandable and maintainable is far more important than trying to squeeze out every last ounce of performance.
Oh, and Mr. Macphisto, I still write C code. I write a lot of C code. I write in quite a number of other languages, too, some of my own divising, but C is a goodly chunk of it.
No one really knows how fast the SR-71 went, unless they've declassified more information in the last few years. Mach 3 was probably a lowball estimate by the government.
Even classified aircraft follow the laws of physics.
It's pretty easy to tell how fast a SR-71 is flying in cruise, providing you have a photograph of the aircraft in cruise. You just look at the engine inlets. They will be arranged such that the shock wave from the leading edge of the "gimmick" (the cone in the inlet) passes just outside the lip of the inlet proper. You measure the distance between the nose of the cone and the lip of the inlet and, knowing the size of the inlet, you know the angle. Then, you look up the mach number that gives that angle for the angle of the cone in something like Anderson's Modern Compressible Flow.
From the whole aerodynamics of the thing, it's pretty obvious that an SR-71 only goes about Mach 3.5.
The example you used was misleading as it would only work in a stored procedure that executes dynamic SQL. If the parameters are instead parameters, bind variables, or whatever else you want to call them, they are resolved only as variables and never as code.
While that may be true, it is also true that the call to the stored procedure is itself SQL, and that SQL is going to be dynamically interpreted because you won't know what the arguments. So, somewhere in the interface to the database, you're going to have a line that looks like this:
That works great, as long as both $first_arg and $second_arg are numbers. However, depending on where they get their values, they could be anything. What that means is that unless you escape everything, the call to the stored procedure that is supposed to look like this:
select * getnumber(1, 32);
Might actually look like this:
select * getnumber(1, 32); create user foo with god_access;
In both cases, the call to the stored procedure is fine, it's the generation of the call to the stored procedure that causes the problem in the second case.
"The 68 Micro Journal", "Computer Language", "The C User's Journal", "The Journal of Pascal, Ada, and Modula-2", and "Micro Cornucopia".
The astute may notice that none of those are still in publication, although two have gone through name and, unfortunately, focus changes. ("Computer Language" became "Software Development" and "The C User's Journal" became "The C/C++ User's Journal" but either one is only a pale shadow of the magazines that they used to be.)
The problem with talking about how hard it is to do AI in games is the fact that game implementers don't need to do actual AI (like Douglas Hofstadter's group is trying to do) but to create NPC's that behave more realistically. I mean, if a cockroach, who has no more than six brain cells, can figure out how to hide if you're chasing it, then it should be no problem for a programmer to figure out how to make an NPC do it.
It occurs to me that, successful human armies don't let their soldiers wait until actual combat and rely on them to figure out what to do in any given circumstance. Instead, they attempt to enumerate the sorts of situations their soldiers are likely to find themselves in and attempt to train those soldiers to handle each of them. Emphasis is placed on recognizing the sort of situation they're in and reacting as they were taught in training. That sounds perfectly amenable to the sort of "rigged demo" approach commonly used in games.
In the reference you gave, the propellant weights are given in thousands of pounds, not pounds, so your total impulse is off by a factor of a thousand, low. Since each letter is a doubling, that puts your class off by about ten letters. That would mean that it's an "AD" or "AE", by your calculations. This makes more sense to me because I know of amateurs that have built "T" class engines.
You can't directly compare a "D" engine to an "F1" because the Estes engine includes an amount of propellant while the F1 is just the equipment used to turn propellant into thrust. The "D" specification gives the total impulse of the engine and the total impulse depends on the amount of propellant consumed. It's about how big the "gas tank" is rather than how hard it pushes. If you're going to directly compare one to another, you either have to compare the thrusts of the engines or specify how long the F1 engine burns.
The Estes D engines that I am aware of have an average thrust of 12 Newtons, burn for about 1.6 seconds and produce a total impulse of 20 Newton seconds. Each F1 engine has a thrust of about 6,675,000 Newtons. If that engine burns for 150 seconds, (the burn time of the first stage for the Apollo lunar missions) then you get a total impulse of 1,001,250,000 Newton-seconds.
Using the same scale that model rocket engines use, 1,001,250,000 NS would be about a "49% AC", (where the designation goes "Y", "Z", "AA", "AB" and so forth) if I've done the math correctly.
I buy the vast majority of my computer parts from a local mom-and-pop clone shop called "PC Parts and Service". To vet them, I simply walked in and evaluated how they responded to my requests. I chose that particular store because they are convenient to me.
I decided I wanted to buy from a local business because it dramatically simplifies the purchasing process, it usually dramatically reduces the wait, and eliminates the "he said, she said" if items paid for don't arrive intact. If I order a system from them, or even just a motherboard, CPU, and RAM, I can see it work before I leave the store.
I like to buy from a small vendor because I figure that Fry's or The Micro Center has plenty of other customers to give them money. They probably haven't even noticed that I don't shop at their stores any more. Not for computer parts, anyway.
I especially like the fact that I can do things like ask for a "3d accelerated video PCI card in the $50 range" and they'll tell me what to buy or I can tell them exactly what I want and if they don't have it stocked, they'll order it for me.
Not that I won't buy stuff from Internet vendors. I've bought stuff at eBay auction and I've wandered through pricewatch, but I don't like the risk in that process. The cost differential between the local store and an Internet store is simply not worth the worry for me. Your mileage may vary.
Historically speaking, hardly anyone has the expertise, the time, and the desire to contribute code to any open source project. So, the "number who would likely be interested" in modifying any project may be large, but the number who would actually contribute code is going to be very, very small compared to that. In other words, it's not obvious to me that being written in Haskell is necessarily a loss for darcs because it is excluding people from contributing.
In fact, I like seeing projects written in languages other than C or C++, even though they're mostly what I write in. There should be lots of widely-used languages with efficient compilers, and having major projects in different langauges promotes that. Monoculture is not just about operating systems, but about other ways of doing things.
And, as a matter of fact, I use darcs at home. I like the fact that it has some sort of theoretical basis.
According to the Xen FAQ, (question 1.4, the one about Microsoft Windows) "virtualisation features in next-generation CPUs should make it much easier to support unmodified OSes". Does anyone know what virtualisation features they're talking about?
Can anybody buy a rocket and launch it from his backyard? Well, it depends on where you live. If you live in the United States, then the answer is a qualified "yes". It's "qualified" because you're not supposed to launch any rockets near airports.
The air traffic is not really an issue, even near a airports. It's hard to hit an airplane with a rocket. That's why military missles are guided and typically use some sort of proximity fuse and explosive warhead. Those guidance systems and warheads with proximity fuses are not generally available to the general public.
Anyway, the regulations recognize three different categories of rockets. If you have a rocket that masses no more than 453 grams and has no more than 113g of grams of propellant, with no more than 62.5 grams in any given engine, then you have what is known as a "model rocket" and you can launch from anywhere that isn't near an airport it without telling anybody you're going to do it.
If you have a rocket that masses no more than 1500 grams and has no more than 125 grams of propellant, with no more than 62.5 grams of propellant in each engine, and a total impulse of no more than 160 NS and an average thrust of no more than 80 Newtons, then you have what's known as a "large model rocket", and you need to notify the FAA before you launch it.
If you have a rocket that masses more than 1500 grams, or has more than 125 grams of propellant or has more than 62.5 grams of propellant in any one engine, has a total impulse of more than 160 NS or an average impulse of more than 80 Newtons, then you have a "high power rocket" which requires that you get the FAA's permission to launch it.
Whether or not you can legally buy the propellant (premanufactured "single use" engines are recognized by everyone as legal to buy, sell, and use) is a matter of opinion, and there are lawsuits pending on that subject right now. If you have an interest in these matters, please donate the NAR legal fund or the TRA legal fund.
The guys I know who do high-power (I'm starting to do LMR's now) made a great deal of fun of these Camui rockets because the high-power guys build rockets that go a great deal higher for a lot less money. A $300 investment can get you a completely reusable rocket that goes up a couple of kilometers as well as the propellant for one flight.
It's never been adequately explained to me why it is legitimate to use the model that says you can't have a particular type of reaction to describe the characteristics of that reaction, should it exist. Perhaps you could enlighten me.
1) A Feature: multimaster asynchronous merge replication. This is Very Hard(tm) (and bogged down by patents in the corporate reich of america), but would make postgres a contender for distributed database applications.
You know, generally the complaint about software patents being unreasonable has to do with how easy it is to do patented stuff. Have you considered that the only reason you know how to do multimaster asynchronous merge replication may be because someone figured out how to do it and told everyone else in a patent?
So, what patents apply and when do they expire?
If you're security-minded at all, you might want to re-think using Debian's testing. Security patches are made to Stable and Unstable, but their integration into testing depends on what other defects may be outstanding for the patched packages.
Actually, the science behind the art can be very important to the artist. A true artist will understand the materials he uses in order to better achieve the effect he wants. So, in that sense, "arts is science as well", just as you say.
That, by the way, illustrates why the "cooking is art, so Alton Brown is a fool" folks completely miss the point. Art is about what you're trying to accomplish, the science (actually, the technology, but that's another discussion) is about how to accomplish it.
If it's so easy to demonstrate that swap will always improve the performance of a system, then why has no one made that demonstration? In fact, I haven't seen any demonstration that suggests that it is better to have 256M of semiconductor memory plus 256M of swap than 512M of semiconductor memory with no swap. Why is it so much harder to grep the kernel tree if there is an application resident in RAM that would have long since been swapped out to disk on a system with a lesser amount of RAM?
Could someone please clarify what is meant by "roll-out" in the article? It wasn't clear to me what they were talking about. My observation is that demand for addresses comes from the leaf nodes, and there is little or no end user access equipment available that does IPv6, so does this mean that they expect some to be widely available by 2005?
I happened to sit in the audience on a panel at ConDFW whose topic was "Man on Mars by 2030?" or some such. At that session, it was pointed out that NASA is a political beast by its very nature, (and how can it be otherwise as NASA is an arm of the US government, itself one of the most political of beasts,) and does what it does for political reasons. That's why they haven't managed to get anyone out of low earth orbit in 32 years. There's been no goal that a politician has had that requires NASA to do that.
In fact, my opinion is that essentially no progress has been made in spaceflight in those 32 years. After all, it doesn't matter to me if a very select few gets to occasionally ride into space because I want to go, and I think that there are lots of people like me. Our interest in space is derived, not from a desire to read about or watch the exploits of a Glenn or an Armstrong, but to go ourselves. However, it appears as if the folks at NASA don't want that. They still view flying in space as being something only for the, well, few that they've selected. I'd like to see that change. Establishing a lunar base gives us the possibility of seeing that change.
There are a number of companies that have been established to exploit space commercially. However, none have really been successful so far. The primary reason is that the income from that exploitation has been uncertain at best. NASA now has the opportunity to change that. If they were to call for a request for bid on, say, five contracts: For providing transfer of personnel from the earth's surface to low earth orbit, for providing transfer of cargo from the earth's surface to low earth orbit, for providing transfer of personnel from low earth orbit to the lunar surface, for providing transfer of cargo from low earth orbit to the lunar surface, and for the construction of a lunar base, this would be the sort of guaranteed income needed to get commercial space ventures really going.
And once those contractors become established, they're going to look around for other ways to make money. One of those ways will be tourism.
In fact, in order to do business those contractors will have to build just the infrastructure you need to send human explorers off to the other planets. It is the establishment of the infrastructure that makes the cost of launching a Mars mission from a lunar base larger than going the Mars direct route. If NASA can get others to build the infrastructure instead, then the numbers look a lot better for launching from the moon or from a space station than for Mars direct.
Slashdot Mirror
To the players, the appeal behind war games isn't because we all secretly yearn for some dictator to come and take over our lives, it's because games are about competition and war is as about as competitive as an activity gets. From the perspective of the publisher, a war game's appeal is similar to the reason that war novels are popular with writers: You get lots of conflict to drive the plot without needing a lot of justification or exposition.
What would happen if, say, the Chinese went up to MIT and said "We'll give you a billion dollars for your class-A address"?
Routeable addresses are supposed to be assigned to TLA's (Top-Level Aggregators) and they portion them out to their customers and they portion them out to their customers and so on until you get your allocation. As far as I am aware, with the exception of the 3ffe::/16 pTLA (psuedo Top Level Aggregator) assignments that were made for the 6bone, and which are now well on their way to being phased out so they're now almost entirely of historical interest, all of the routeable IPv6 addresses are in 2000::/2, so if the first digit in your IPv6 address is not a 2, then you don't have IPv6 connectivity.
As an example, I've gone to the trouble of setting my network as 2001:5c0:8305::/48, but I don't (yet) allow incoming connections to my network. However, all of the interefaces on all of my machines have addresses in the fe80::/16 block even though I've got three physically separate LAN's at the house.
Since demand for addresses necessarily comes from the leaf nodes of the network (where the bulk of them are consumed) rather than the backbones, I think it is disingenuous (to say the least) to claim that IPv6 is already "rolled out" because it is available from various backbone providers when the reality is that it is not available directly to the end users.
That's like asking "What's the sound of yellow". "In general" the question is not answerable. In my opinion, the question is also misguided. This sort of thing is what Hoare was talking about when he said "Premature optimization is the root of all evil". Making the program understandable and maintainable is far more important than trying to squeeze out every last ounce of performance.
Oh, and Mr. Macphisto, I still write C code. I write a lot of C code. I write in quite a number of other languages, too, some of my own divising, but C is a goodly chunk of it.
Even classified aircraft follow the laws of physics.
It's pretty easy to tell how fast a SR-71 is flying in cruise, providing you have a photograph of the aircraft in cruise. You just look at the engine inlets. They will be arranged such that the shock wave from the leading edge of the "gimmick" (the cone in the inlet) passes just outside the lip of the inlet proper. You measure the distance between the nose of the cone and the lip of the inlet and, knowing the size of the inlet, you know the angle. Then, you look up the mach number that gives that angle for the angle of the cone in something like Anderson's Modern Compressible Flow.
From the whole aerodynamics of the thing, it's pretty obvious that an SR-71 only goes about Mach 3.5.
While that may be true, it is also true that the call to the stored procedure is itself SQL, and that SQL is going to be dynamically interpreted because you won't know what the arguments. So, somewhere in the interface to the database, you're going to have a line that looks like this:
$query = "select * getnumber(".$first_arg.",".$second_arg.");";
That works great, as long as both $first_arg and $second_arg are numbers. However, depending on where they get their values, they could be anything. What that means is that unless you escape everything, the call to the stored procedure that is supposed to look like this:
select * getnumber(1, 32);
Might actually look like this:
select * getnumber(1, 32); create user foo with god_access;
In both cases, the call to the stored procedure is fine, it's the generation of the call to the stored procedure that causes the problem in the second case.
The astute may notice that none of those are still in publication, although two have gone through name and, unfortunately, focus changes. ("Computer Language" became "Software Development" and "The C User's Journal" became "The C/C++ User's Journal" but either one is only a pale shadow of the magazines that they used to be.)
How do you determine which version of a given program is installed on a Slackware system?
It occurs to me that, successful human armies don't let their soldiers wait until actual combat and rely on them to figure out what to do in any given circumstance. Instead, they attempt to enumerate the sorts of situations their soldiers are likely to find themselves in and attempt to train those soldiers to handle each of them. Emphasis is placed on recognizing the sort of situation they're in and reacting as they were taught in training. That sounds perfectly amenable to the sort of "rigged demo" approach commonly used in games.
The device in question looks neat, but I need analog I/O, and this doesn't appear to do it.
In the reference you gave, the propellant weights are given in thousands of pounds, not pounds, so your total impulse is off by a factor of a thousand, low. Since each letter is a doubling, that puts your class off by about ten letters. That would mean that it's an "AD" or "AE", by your calculations. This makes more sense to me because I know of amateurs that have built "T" class engines.
The Estes D engines that I am aware of have an average thrust of 12 Newtons, burn for about 1.6 seconds and produce a total impulse of 20 Newton seconds. Each F1 engine has a thrust of about 6,675,000 Newtons. If that engine burns for 150 seconds, (the burn time of the first stage for the Apollo lunar missions) then you get a total impulse of 1,001,250,000 Newton-seconds.
Using the same scale that model rocket engines use, 1,001,250,000 NS would be about a "49% AC", (where the designation goes "Y", "Z", "AA", "AB" and so forth) if I've done the math correctly.
I decided I wanted to buy from a local business because it dramatically simplifies the purchasing process, it usually dramatically reduces the wait, and eliminates the "he said, she said" if items paid for don't arrive intact. If I order a system from them, or even just a motherboard, CPU, and RAM, I can see it work before I leave the store.
I like to buy from a small vendor because I figure that Fry's or The Micro Center has plenty of other customers to give them money. They probably haven't even noticed that I don't shop at their stores any more. Not for computer parts, anyway.
I especially like the fact that I can do things like ask for a "3d accelerated video PCI card in the $50 range" and they'll tell me what to buy or I can tell them exactly what I want and if they don't have it stocked, they'll order it for me.
Not that I won't buy stuff from Internet vendors. I've bought stuff at eBay auction and I've wandered through pricewatch, but I don't like the risk in that process. The cost differential between the local store and an Internet store is simply not worth the worry for me. Your mileage may vary.
In fact, I like seeing projects written in languages other than C or C++, even though they're mostly what I write in. There should be lots of widely-used languages with efficient compilers, and having major projects in different langauges promotes that. Monoculture is not just about operating systems, but about other ways of doing things.
And, as a matter of fact, I use darcs at home. I like the fact that it has some sort of theoretical basis.
According to the Xen FAQ, (question 1.4, the one about Microsoft Windows) "virtualisation features in next-generation CPUs should make it much easier to support unmodified OSes". Does anyone know what virtualisation features they're talking about?
The air traffic is not really an issue, even near a airports. It's hard to hit an airplane with a rocket. That's why military missles are guided and typically use some sort of proximity fuse and explosive warhead. Those guidance systems and warheads with proximity fuses are not generally available to the general public.
Anyway, the regulations recognize three different categories of rockets. If you have a rocket that masses no more than 453 grams and has no more than 113g of grams of propellant, with no more than 62.5 grams in any given engine, then you have what is known as a "model rocket" and you can launch from anywhere that isn't near an airport it without telling anybody you're going to do it.
If you have a rocket that masses no more than 1500 grams and has no more than 125 grams of propellant, with no more than 62.5 grams of propellant in each engine, and a total impulse of no more than 160 NS and an average thrust of no more than 80 Newtons, then you have what's known as a "large model rocket", and you need to notify the FAA before you launch it.
If you have a rocket that masses more than 1500 grams, or has more than 125 grams of propellant or has more than 62.5 grams of propellant in any one engine, has a total impulse of more than 160 NS or an average impulse of more than 80 Newtons, then you have a "high power rocket" which requires that you get the FAA's permission to launch it.
There are two organizations supporting model and amateur rocketry in the USA. The National Association of Rocketry and the Tripoli Rocketry Association. Those web sites have links to local chapters all over the country.
Whether or not you can legally buy the propellant (premanufactured "single use" engines are recognized by everyone as legal to buy, sell, and use) is a matter of opinion, and there are lawsuits pending on that subject right now. If you have an interest in these matters, please donate the NAR legal fund or the TRA legal fund.
The guys I know who do high-power (I'm starting to do LMR's now) made a great deal of fun of these Camui rockets because the high-power guys build rockets that go a great deal higher for a lot less money. A $300 investment can get you a completely reusable rocket that goes up a couple of kilometers as well as the propellant for one flight.
It's never been adequately explained to me why it is legitimate to use the model that says you can't have a particular type of reaction to describe the characteristics of that reaction, should it exist. Perhaps you could enlighten me.
You know, generally the complaint about software patents being unreasonable has to do with how easy it is to do patented stuff. Have you considered that the only reason you know how to do multimaster asynchronous merge replication may be because someone figured out how to do it and told everyone else in a patent? So, what patents apply and when do they expire?
Walter Bright's version is but a pale shadow of the real "Empire". For more information, you can go to the source
If you're security-minded at all, you might want to re-think using Debian's testing. Security patches are made to Stable and Unstable, but their integration into testing depends on what other defects may be outstanding for the patched packages.
That, by the way, illustrates why the "cooking is art, so Alton Brown is a fool" folks completely miss the point. Art is about what you're trying to accomplish, the science (actually, the technology, but that's another discussion) is about how to accomplish it.
If it's so easy to demonstrate that swap will always improve the performance of a system, then why has no one made that demonstration? In fact, I haven't seen any demonstration that suggests that it is better to have 256M of semiconductor memory plus 256M of swap than 512M of semiconductor memory with no swap. Why is it so much harder to grep the kernel tree if there is an application resident in RAM that would have long since been swapped out to disk on a system with a lesser amount of RAM?
Could someone please clarify what is meant by "roll-out" in the article? It wasn't clear to me what they were talking about. My observation is that demand for addresses comes from the leaf nodes, and there is little or no end user access equipment available that does IPv6, so does this mean that they expect some to be widely available by 2005?
In fact, my opinion is that essentially no progress has been made in spaceflight in those 32 years. After all, it doesn't matter to me if a very select few gets to occasionally ride into space because I want to go, and I think that there are lots of people like me. Our interest in space is derived, not from a desire to read about or watch the exploits of a Glenn or an Armstrong, but to go ourselves. However, it appears as if the folks at NASA don't want that. They still view flying in space as being something only for the, well, few that they've selected. I'd like to see that change. Establishing a lunar base gives us the possibility of seeing that change.
There are a number of companies that have been established to exploit space commercially. However, none have really been successful so far. The primary reason is that the income from that exploitation has been uncertain at best. NASA now has the opportunity to change that. If they were to call for a request for bid on, say, five contracts: For providing transfer of personnel from the earth's surface to low earth orbit, for providing transfer of cargo from the earth's surface to low earth orbit, for providing transfer of personnel from low earth orbit to the lunar surface, for providing transfer of cargo from low earth orbit to the lunar surface, and for the construction of a lunar base, this would be the sort of guaranteed income needed to get commercial space ventures really going.
And once those contractors become established, they're going to look around for other ways to make money. One of those ways will be tourism.
In fact, in order to do business those contractors will have to build just the infrastructure you need to send human explorers off to the other planets. It is the establishment of the infrastructure that makes the cost of launching a Mars mission from a lunar base larger than going the Mars direct route. If NASA can get others to build the infrastructure instead, then the numbers look a lot better for launching from the moon or from a space station than for Mars direct.