Well in any case all of this seems to be saying is that it would be hard to actually measure WABR in practice. However I don't think that detracts from my main point, because when giving advice you're almost never going to have the opportunity split people into large test groups and actually measure the WABR of two different pieces of advice. You just have to consider if two different pieces of advice were given to exactly identical groups of people (which can't happen in real life, but the concept is well-defined), which piece of advice would probably do better.
Honestly, we all know that "advice" like "Dump Windows, install Linux" or "Use RAID for your personal home backups" or "Eat less and exercise more" would score badly under that system, which is what makes it bad advice.
The reason I wrote the original article at http://news.slashdot.org/story...
was because so much of the advice I got from veteran burners was unnecessarily difficult. Not a single one of them told me I could just bring a storebought tent and stake it to the ground, they all said that tents didn't work and you had to build a structure from PVC and tarp, or a hexayurt, and sleep in that. (The true answer is that tents don't work with the stakes provided, which are not secure enough to anchor them to the ground, but you can combine a storebought tent with rebar to anchor it down, for the drop-dead easiest way to build yourself a place to sleep, which was the question I was trying to get answered.)
I think if you want people to be prepared, you have to give them a trustworthy information source, and the way for the information source to earn that trust is to actually give the reader/listener the information they're looking for, and not steer them down a path that is more difficult than necessary. When I realized that the veterans were telling me about "hexayurts" to puff themselves up and not to be helpful, I stopped listening to whatever else they had to say.
To this day, do you know of any information source that gives complete and useful information about preparing for Burning Man (at minimum, for example, telling the reader to pre-reserve a rented bicycle, instead of telling them to use "community bikes") and also provides a checklist of things to bring? Then maybe that's the checklist we should point people to.
If Burning Man was easy to get to, and attending was as simple as using money to purchase things rather than organizing/creating things on your own/contributing, it would become Coachella. The fact that Burning Man is so difficult to get to, has a harsh desert climate, requires you to bring your own water etc helps ensure that more committed/invested people attend rather than tourists. Making it easier for people to get to the playa cheaply and without connections to Burning Man culture would increase the number of tourists on the playa, and change the festival drastically. It would also surely increase the amount of people in the medical tents and put a strain on emergency services, since they are unlikely to read the guides/adequately prepare.
I've met first-years who merely skimmed through the guide and thought it was "exaggerated" and came with no warm clothes and summarily froze every night. They later ended up in the medical tent for dehydration and a broken ankle. How could you think going to a web site and easily booking a flight and then bus ride to Burning Man would not increase the number of clueless tourists, who are potentially a safety hazard to themselves/others?
OK, you're right, my goal was to make it easy to get to Burning Man, and if you're assuming that in itself creates problems, then yes that changes everything.
But maybe a way to stop people from coming unprepared is to give them preparation advice which is really, actually correct. The official Burning Man website still tells people that there are community bikes lying around that you can use to get from place to place: http://www.burningman.com/on_t...
NO YOU CAN'T!! There are never any community bikes available!! What they should have said is that if you're not driving in with your own bike, you can pre-pay to rent a bike from a bicycle camp. But despite three different pages on the site about bikes: http://www.burningman.com/on_t... http://www.burningman.com/prep... http://www.burningman.com/prep...
they never mention the on-playa bike rental option once.
Some people did show up unprepared -- they found the bicycle rental camp and hoped that they could rent a bike, only to be told that the bike rentals were all booked up long ago -- and the reason they showed up unprepared is because the Burning Man website lied to them.
If you want people to show up prepared, let's at least try the approach of giving people correct and complete information on how to be prepared.
I should have been more clear that I didn't presume to think this was a new idea. (Perhaps people took it the wrong way that if I genuinely thought this was a new idea, I was so naïve that they didn't need to read the rest of it.)
Rather, I think this is something that a lot of people kinda know, but keep forgetting. I think it's something we should all keep foremost in our minds when giving advice ourselves, or in critiquing advice given by other people. I mean, surely you agree that there is a lot of "advice" out there (even excluding the pages put up to grab Google traffic) which doesn't even attempt to do well under this metric. Let's do more to have less of it.
I understand, but all of that sounds like it's supporting the approach I was proposing: Divide your volunteers randomly into two groups, give Program 1 to the first group and Program 2 to the second group, and then check at the end of some time period which group has better results.
Would you suggest modifying that approach in any way based on your knowledge of statistics?
I understand but I still don't see why Dropbox isn't good enough for the scenario you're describing. It gives you "quick and easy access" to the stuff you want restored from backup, after all.
Unless you're talking about a system image or something so that you can get back up and running quickly without having to re-install all of your needed programs? In which case Dropbox wouldn't be appropriate.
I think this is true, just that it doesn't invalidate WABR, it just means WABR is impossible to use in this case. For energy usage, clearly the long-term consequences are what we care about, we just can't measure them in the short term. So we have to actually think, like you said:)
WABR might be better for helping people to reduce their own personal carbon footprint. If someone tells me to reduce my carbon footprint by taking the bus, I'm going to point out that it takes twice as long to get anywhere using the bus, and that's the end of it for me. If someone tells me that if I buy an electric car, it will cost $X more money up front, but that expense will pay for itself after N years because of gas savings, well now I'm listening.
Most of the people who wrote this in response to the first article, were unaware of the fact that Dropbox maintains a version history of your file for you.
Some people posted responses to my original article (in which I told people to use Dropbox) saying that Dropbox actually does maintain a version history, and if you delete a file (or make an update that you regret), you can get the old one back. I haven't tried it.
Are you saying the main difference is that Crashplan's versioning goes back further? (Or that you think it probably does, if you're not sure how it works on Dropbox?)
I think it works better as a comparison between two different pieces of advice, than as an absolute measurement.
For example, if one person tells their friends "Quit smoking!", and the other person tells their friends about a specific smoking cessation program or e-cig product that is known to produce good results, then the second person will extend more of their friends' lives on average. Isn't that a good thing? Why would anybody want to be the first person instead of the second?
You're right, this could have been made shorter, and an alternative shorter version of the article probably would have scored better under WABR. I just put this out there because right now there is nothing else out there making this point, that I'm aware of.
Thanks for the pointer to the "intention to treat" term -- I knew vaguely that there was a term for this in medicine specifically, but I couldn't find it.
I knew this wasn't an entirely new idea. I just think there isn't enough general awareness of the importance of judging advice by this metric, even casually given advice.
There must be doctors working with Michelle Obama who are aware of the "intention to treat" rule, for example. And yet they still signed off on the "Let's Move" campaign, which probably won't work very well by that criterion. (Most people are unaware of how long you'd have to run around to burn off the calories from one deceptively healthy-looking white bread sandwich, and how much easier it would be to replace that with a fruit snack.)
Do you have a better alternative, and an argument as to why the alternative is better? (I will assume that all answers of the form, "Yes I have a better alternative, but I'm not going to say it" are equivalent to "I don't have a better alternative.")
I think these are both excellent points -- that if you are trying to increase utility for a single person, then WABR advice should be based on estimates for the population that you think that person is part of. And if you're giving advice to a group, then the goal should depend on whether you're trying to achieve best-possible outcomes for a subset of the group or only pretty-good outcomes for most people in the group. Thanks.
Exactly, WABR would be the right test for all of these and I do think your alternatives would work better.
Although, virtually no dating "advice" for men can ever work really well, because studies have shown that when you ask women under lie-detector conditions, the thing they care overwhelmingly about is physical looks: http://psycnet.apa.org/index.c...
and this corresponds pretty well with actual dating preferences as revealed by which men get the most inquiries on dating sites, etc. This might actually be the one question on which there is the greatest gap between what professional researchers have found with their studies, and what the public thinks that the studies have found -- almost all of the public thinks that "studies show" women care more about personality. (By contrast, with global warming, yes there is a near-universal consensus among scientists, but at least a large part of the public is on board with that consensus as well.)
Thus the dating advice that tends to get passed around the most, probably has little to do with CBR versus WABR. The dating advice which endures is the kind that is unfalsifiable, so that even if it doesn't work, you can never show the advice-giver any evidence will change their mind. If someone tells you to be "confident", and it doesn't work, they can just say that "deep down", you weren't confident "enough". They're never going to say, "Oh, I guess my advice didn't work. Sorry about that."
I think you're right, someone could probably write an article making a similar point that would score better under WABR, one reason being that it could probably be made shorter. I just wanted to get the point out there because right now there was nothing else making this point, that I could find (at least with regard to tech advice).
Good point, perhaps the best advice is something along the lines of, "If security is a very high priority, switch to Linux, otherwise, install such-and-such." I mentioned in the article that you could segment your advice by different cases -- If your scenario is A, do B, but if your scenario is C, do D -- and still call the whole thing "one piece of advice", as long as it's still judged based on the average results achieved by everybody who hears it, without nit-picking whether they "followed' it or not.
The texting and driving is a tougher question, but I think WABR is still the right approach. I believe the point you're making is that if you tell people "Wear your seat belt while texting and driving", that will encourage people to wear their seat belt who otherwise would not have, but it might also encourage some people to text and drive who otherwise would not have texted while driving at all. However, that will still be reflected in the WABR test -- if your advice has harmful effects, then that will increase the number of car accidents in the group of people who receive the advice. (It's hard to think of something that would work well on people who text and drive, because they must have already heard all the exhortations not to. My suggestion: pull over at the next spot, text everybody you're in a "conversation" with and tell them you're about to be driving and can't talk. Then you won't be tempted to keep texting them because you know that if you do, they'll unload on you for texting and driving next time you see them.)
I think this sounds like great advice and is better than 90% of the suggested "improvements" that were posted on my "4 tips for your new laptop" article. In particular thanks for the pointer about Chrome. Although can you explain what's the benefit of Crashplan's backup if you're already saving documents to Google docs / Sky Drive / Dropbox?
If one piece of advice produces better long-term results than another piece of advice, you could use this using WABR as well. You just have to check in after three years to see how both groups are doing, while still following the WABR rule of collecting results from everybody and not just from the people that you think "did it right". I mentioned that in the weight-loss example -- what you really want to do is check in to see who's kept the weight off three years down the road.
Thanks. This is indeed a valid objection -- I just think it depends on the proportion of people who are in this situation.
After all, if everybody has no idea when they're leaving, then they'll just all use the non-priority queue, which will just become the regular queue, and we'll be no worse off than we already are.
On the other hand, I met a lot of people who drove there but were not running a major theme camp, they just drove there and parked because that's how they got there. I think they could probably pack up on a fairly predictable schedule. When you move people out of the old-style queue into the license-plate-priority queue, you reduce average wait times for everybody, and you at least don't make things worse for people in the old-style queue (at least, as long as the gatekeepers are still letting people through the old-style queue while merging them with the license-plate queue).
My suggestion was based on the information in the Burning Man FAQ: http://www.burningman.com/prep...
which says that the main reason they don't use a priority express-lane system is that registering, and verifying registrations on exit, would be a huge hassle. So I was trying to come up with a way that avoided a registration system or a cumbersome verification process. If there are other problems with a scheduled registration system, then yes, that changes the problem. If most people have no idea how long it will take them to pack up and leave, then the idea falls apart. It seemed to me though like a lot of people who came in cars wouldn't take that long to pack up.
Oh, when you said it was already a "near-perfect queue", I assumed you meant a queue where people had information about how long the line would take and only got in line when the wait was worth it. (Otherwise, what did you mean, that the queue is last-in-first-out? I would have thought that was too obvious to bother saying:) )
If everybody in the regular queue just "gets in line when they're ready" but it's no longer the only queue to get out, then that factor will tend to make that queue longer. On the other hand, with the license-plate-prioritized queue, the regular queue will have fewer people in it, which will tend to make it shorter. Meanwhile the queue time for people in the prioritized line is almost zero. Only the first factor increases the average queue time, the other two factors reduce the average wait time.
I think it is more complex and it does get interesting.
I was responding more to your specific statement: "The above statement ['the queue will still grow to the point where the convenience of getting out is just barely outweighed by the inconvenience of waiting in line'] is STILL true when the wait is 30 seconds."
My argument was that this is not true if you divide the population into small enough groups. Past a certain threshold, the wait time drops while the value of the prize at the end stays the same, and the difference is a free lunch, a pure benefit to people in the queue, at nobody else's expense.
Only the cars in the priority line would have reduced wait times. Because the bottleneck is actually the highway, that reduction comes at the expense of raising the wait times for everyone not in the priority line - unless the lure of the priority queue is so strong that there's enough people waiting for the priority queue to clear the highway bottleneck.
It depends on what you mean by "wait time". If you mean the elapsed time from midnight on Saturday until the time when a car gets off of the playa, then yes this algorithm does not reduce the wait time, because as long as cars can only leave at 1,000 cars per hour, then no matter how you re-order cars, the same number of people have to get out later rather than earlier.
On the other hand, if you define "wait time" to mean "time waiting in the actual physical car queue", then this algorithm does reduce the amount of time that people have to spend sitting in their cars, for people in the priority line. And that extra time saved doesn't come at the expense of the people sitting in the non-prioritized line (because the non-prioritized line will always grow to the point where the convenience of getting out just barely outweighs the convenience of waiting in line -- but no further). It doesn't come at anybody's expense at all. It's a genuine free lunch.
The only thing that will reduce the wait time in a fair way is to stagger departures more, so that there's lower peak load. Micromananging the current exodus process, a process which is already a nearly perfect queue, will only serve to make things worse on average, in addition to requiring more manpower to manage.
It depends on what you mean by "a nearly perfect queue". If you mean "a queue that follows the theoretical algorithmic properties of a queue", then obviously yes. If you mean "a queue that minimizes the amount of time that people have to spend physically sitting in their cars", then no.
Frankly, even though you've been to Burning Man, I don't feel you're competent to be suggesting improvements to the exodus traffic problem. You've been there once, and you came in and left on a fast-track bus which was able to bypass the entire process you're trying to critique. Had you been through it a few times and seen how traffic exodus works for people who are actually in the thick of it, I doubt you'd have bothered trying to micromanage the queues.
This is not an answer to the question posed in the article, which was, "Why wouldn't this work?"
Your previous paragraphs were presumably an attempt to answer that question. However I clarified that I think it was based on a misunderstanding of what I meant by "reduced wait time".
Slashdot Mirror
Well in any case all of this seems to be saying is that it would be hard to actually measure WABR in practice. However I don't think that detracts from my main point, because when giving advice you're almost never going to have the opportunity split people into large test groups and actually measure the WABR of two different pieces of advice. You just have to consider if two different pieces of advice were given to exactly identical groups of people (which can't happen in real life, but the concept is well-defined), which piece of advice would probably do better.
Honestly, we all know that "advice" like "Dump Windows, install Linux" or "Use RAID for your personal home backups" or "Eat less and exercise more" would score badly under that system, which is what makes it bad advice.
The reason I wrote the original article at
http://news.slashdot.org/story...
was because so much of the advice I got from veteran burners was unnecessarily difficult. Not a single one of them told me I could just bring a storebought tent and stake it to the ground, they all said that tents didn't work and you had to build a structure from PVC and tarp, or a hexayurt, and sleep in that. (The true answer is that tents don't work with the stakes provided, which are not secure enough to anchor them to the ground, but you can combine a storebought tent with rebar to anchor it down, for the drop-dead easiest way to build yourself a place to sleep, which was the question I was trying to get answered.)
I think if you want people to be prepared, you have to give them a trustworthy information source, and the way for the information source to earn that trust is to actually give the reader/listener the information they're looking for, and not steer them down a path that is more difficult than necessary. When I realized that the veterans were telling me about "hexayurts" to puff themselves up and not to be helpful, I stopped listening to whatever else they had to say.
To this day, do you know of any information source that gives complete and useful information about preparing for Burning Man (at minimum, for example, telling the reader to pre-reserve a rented bicycle, instead of telling them to use "community bikes") and also provides a checklist of things to bring? Then maybe that's the checklist we should point people to.
If Burning Man was easy to get to, and attending was as simple as using money to purchase things rather than organizing/creating things on your own/contributing, it would become Coachella. The fact that Burning Man is so difficult to get to, has a harsh desert climate, requires you to bring your own water etc helps ensure that more committed/invested people attend rather than tourists. Making it easier for people to get to the playa cheaply and without connections to Burning Man culture would increase the number of tourists on the playa, and change the festival drastically. It would also surely increase the amount of people in the medical tents and put a strain on emergency services, since they are unlikely to read the guides/adequately prepare.
I've met first-years who merely skimmed through the guide and thought it was "exaggerated" and came with no warm clothes and summarily froze every night. They later ended up in the medical tent for dehydration and a broken ankle. How could you think going to a web site and easily booking a flight and then bus ride to Burning Man would not increase the number of clueless tourists, who are potentially a safety hazard to themselves/others?
OK, you're right, my goal was to make it easy to get to Burning Man, and if you're assuming that in itself creates problems, then yes that changes everything.
But maybe a way to stop people from coming unprepared is to give them preparation advice which is really, actually correct. The official Burning Man website still tells people that there are community bikes lying around that you can use to get from place to place:
http://www.burningman.com/on_t...
NO YOU CAN'T!! There are never any community bikes available!! What they should have said is that if you're not driving in with your own bike, you can pre-pay to rent a bike from a bicycle camp. But despite three different pages on the site about bikes:
http://www.burningman.com/on_t...
http://www.burningman.com/prep...
http://www.burningman.com/prep...
they never mention the on-playa bike rental option once.
Some people did show up unprepared -- they found the bicycle rental camp and hoped that they could rent a bike, only to be told that the bike rentals were all booked up long ago -- and the reason they showed up unprepared is because the Burning Man website lied to them.
If you want people to show up prepared, let's at least try the approach of giving people correct and complete information on how to be prepared.
I should have been more clear that I didn't presume to think this was a new idea. (Perhaps people took it the wrong way that if I genuinely thought this was a new idea, I was so naïve that they didn't need to read the rest of it.)
Rather, I think this is something that a lot of people kinda know, but keep forgetting. I think it's something we should all keep foremost in our minds when giving advice ourselves, or in critiquing advice given by other people. I mean, surely you agree that there is a lot of "advice" out there (even excluding the pages put up to grab Google traffic) which doesn't even attempt to do well under this metric. Let's do more to have less of it.
Sorry, I didn't mean "official cleanup". I meant things like helping to clean up around your own camp.
I understand, but all of that sounds like it's supporting the approach I was proposing: Divide your volunteers randomly into two groups, give Program 1 to the first group and Program 2 to the second group, and then check at the end of some time period which group has better results.
Would you suggest modifying that approach in any way based on your knowledge of statistics?
I understand but I still don't see why Dropbox isn't good enough for the scenario you're describing. It gives you "quick and easy access" to the stuff you want restored from backup, after all.
Unless you're talking about a system image or something so that you can get back up and running quickly without having to re-install all of your needed programs? In which case Dropbox wouldn't be appropriate.
I think this is true, just that it doesn't invalidate WABR, it just means WABR is impossible to use in this case. For energy usage, clearly the long-term consequences are what we care about, we just can't measure them in the short term. So we have to actually think, like you said :)
WABR might be better for helping people to reduce their own personal carbon footprint. If someone tells me to reduce my carbon footprint by taking the bus, I'm going to point out that it takes twice as long to get anywhere using the bus, and that's the end of it for me. If someone tells me that if I buy an electric car, it will cost $X more money up front, but that expense will pay for itself after N years because of gas savings, well now I'm listening.
Most of the people who wrote this in response to the first article, were unaware of the fact that Dropbox maintains a version history of your file for you.
Some people posted responses to my original article (in which I told people to use Dropbox) saying that Dropbox actually does maintain a version history, and if you delete a file (or make an update that you regret), you can get the old one back. I haven't tried it.
Are you saying the main difference is that Crashplan's versioning goes back further? (Or that you think it probably does, if you're not sure how it works on Dropbox?)
I think it works better as a comparison between two different pieces of advice, than as an absolute measurement.
For example, if one person tells their friends "Quit smoking!", and the other person tells their friends about a specific smoking cessation program or e-cig product that is known to produce good results, then the second person will extend more of their friends' lives on average. Isn't that a good thing? Why would anybody want to be the first person instead of the second?
You're right, this could have been made shorter, and an alternative shorter version of the article probably would have scored better under WABR. I just put this out there because right now there is nothing else out there making this point, that I'm aware of.
Thanks for the pointer to the "intention to treat" term -- I knew vaguely that there was a term for this in medicine specifically, but I couldn't find it.
I knew this wasn't an entirely new idea. I just think there isn't enough general awareness of the importance of judging advice by this metric, even casually given advice.
There must be doctors working with Michelle Obama who are aware of the "intention to treat" rule, for example. And yet they still signed off on the "Let's Move" campaign, which probably won't work very well by that criterion. (Most people are unaware of how long you'd have to run around to burn off the calories from one deceptively healthy-looking white bread sandwich, and how much easier it would be to replace that with a fruit snack.)
Do you have a better alternative, and an argument as to why the alternative is better? (I will assume that all answers of the form, "Yes I have a better alternative, but I'm not going to say it" are equivalent to "I don't have a better alternative.")
I think these are both excellent points -- that if you are trying to increase utility for a single person, then WABR advice should be based on estimates for the population that you think that person is part of. And if you're giving advice to a group, then the goal should depend on whether you're trying to achieve best-possible outcomes for a subset of the group or only pretty-good outcomes for most people in the group. Thanks.
Exactly, WABR would be the right test for all of these and I do think your alternatives would work better.
Although, virtually no dating "advice" for men can ever work really well, because studies have shown that when you ask women under lie-detector conditions, the thing they care overwhelmingly about is physical looks:
http://psycnet.apa.org/index.c...
and this corresponds pretty well with actual dating preferences as revealed by which men get the most inquiries on dating sites, etc. This might actually be the one question on which there is the greatest gap between what professional researchers have found with their studies, and what the public thinks that the studies have found -- almost all of the public thinks that "studies show" women care more about personality. (By contrast, with global warming, yes there is a near-universal consensus among scientists, but at least a large part of the public is on board with that consensus as well.)
Thus the dating advice that tends to get passed around the most, probably has little to do with CBR versus WABR. The dating advice which endures is the kind that is unfalsifiable, so that even if it doesn't work, you can never show the advice-giver any evidence will change their mind. If someone tells you to be "confident", and it doesn't work, they can just say that "deep down", you weren't confident "enough". They're never going to say, "Oh, I guess my advice didn't work. Sorry about that."
I think you're right, someone could probably write an article making a similar point that would score better under WABR, one reason being that it could probably be made shorter. I just wanted to get the point out there because right now there was nothing else making this point, that I could find (at least with regard to tech advice).
Good point, perhaps the best advice is something along the lines of, "If security is a very high priority, switch to Linux, otherwise, install such-and-such." I mentioned in the article that you could segment your advice by different cases -- If your scenario is A, do B, but if your scenario is C, do D -- and still call the whole thing "one piece of advice", as long as it's still judged based on the average results achieved by everybody who hears it, without nit-picking whether they "followed' it or not.
The texting and driving is a tougher question, but I think WABR is still the right approach. I believe the point you're making is that if you tell people "Wear your seat belt while texting and driving", that will encourage people to wear their seat belt who otherwise would not have, but it might also encourage some people to text and drive who otherwise would not have texted while driving at all. However, that will still be reflected in the WABR test -- if your advice has harmful effects, then that will increase the number of car accidents in the group of people who receive the advice. (It's hard to think of something that would work well on people who text and drive, because they must have already heard all the exhortations not to. My suggestion: pull over at the next spot, text everybody you're in a "conversation" with and tell them you're about to be driving and can't talk. Then you won't be tempted to keep texting them because you know that if you do, they'll unload on you for texting and driving next time you see them.)
I think this sounds like great advice and is better than 90% of the suggested "improvements" that were posted on my "4 tips for your new laptop" article. In particular thanks for the pointer about Chrome. Although can you explain what's the benefit of Crashplan's backup if you're already saving documents to Google docs / Sky Drive / Dropbox?
If one piece of advice produces better long-term results than another piece of advice, you could use this using WABR as well. You just have to check in after three years to see how both groups are doing, while still following the WABR rule of collecting results from everybody and not just from the people that you think "did it right". I mentioned that in the weight-loss example -- what you really want to do is check in to see who's kept the weight off three years down the road.
Thanks. This is indeed a valid objection -- I just think it depends on the proportion of people who are in this situation.
After all, if everybody has no idea when they're leaving, then they'll just all use the non-priority queue, which will just become the regular queue, and we'll be no worse off than we already are.
On the other hand, I met a lot of people who drove there but were not running a major theme camp, they just drove there and parked because that's how they got there. I think they could probably pack up on a fairly predictable schedule. When you move people out of the old-style queue into the license-plate-priority queue, you reduce average wait times for everybody, and you at least don't make things worse for people in the old-style queue (at least, as long as the gatekeepers are still letting people through the old-style queue while merging them with the license-plate queue).
My suggestion was based on the information in the Burning Man FAQ:
http://www.burningman.com/prep...
which says that the main reason they don't use a priority express-lane system is that registering, and verifying registrations on exit, would be a huge hassle. So I was trying to come up with a way that avoided a registration system or a cumbersome verification process. If there are other problems with a scheduled registration system, then yes, that changes the problem. If most people have no idea how long it will take them to pack up and leave, then the idea falls apart. It seemed to me though like a lot of people who came in cars wouldn't take that long to pack up.
Oh, when you said it was already a "near-perfect queue", I assumed you meant a queue where people had information about how long the line would take and only got in line when the wait was worth it. (Otherwise, what did you mean, that the queue is last-in-first-out? I would have thought that was too obvious to bother saying :) )
If everybody in the regular queue just "gets in line when they're ready" but it's no longer the only queue to get out, then that factor will tend to make that queue longer. On the other hand, with the license-plate-prioritized queue, the regular queue will have fewer people in it, which will tend to make it shorter. Meanwhile the queue time for people in the prioritized line is almost zero. Only the first factor increases the average queue time, the other two factors reduce the average wait time.
I think it is more complex and it does get interesting.
I was responding more to your specific statement: "The above statement ['the queue will still grow to the point where the convenience of getting out is just barely outweighed by the inconvenience of waiting in line'] is STILL true when the wait is 30 seconds."
My argument was that this is not true if you divide the population into small enough groups. Past a certain threshold, the wait time drops while the value of the prize at the end stays the same, and the difference is a free lunch, a pure benefit to people in the queue, at nobody else's expense.
Only the cars in the priority line would have reduced wait times. Because the bottleneck is actually the highway, that reduction comes at the expense of raising the wait times for everyone not in the priority line - unless the lure of the priority queue is so strong that there's enough people waiting for the priority queue to clear the highway bottleneck.
It depends on what you mean by "wait time". If you mean the elapsed time from midnight on Saturday until the time when a car gets off of the playa, then yes this algorithm does not reduce the wait time, because as long as cars can only leave at 1,000 cars per hour, then no matter how you re-order cars, the same number of people have to get out later rather than earlier.
On the other hand, if you define "wait time" to mean "time waiting in the actual physical car queue", then this algorithm does reduce the amount of time that people have to spend sitting in their cars, for people in the priority line. And that extra time saved doesn't come at the expense of the people sitting in the non-prioritized line (because the non-prioritized line will always grow to the point where the convenience of getting out just barely outweighs the convenience of waiting in line -- but no further). It doesn't come at anybody's expense at all. It's a genuine free lunch.
The only thing that will reduce the wait time in a fair way is to stagger departures more, so that there's lower peak load. Micromananging the current exodus process, a process which is already a nearly perfect queue, will only serve to make things worse on average, in addition to requiring more manpower to manage.
It depends on what you mean by "a nearly perfect queue". If you mean "a queue that follows the theoretical algorithmic properties of a queue", then obviously yes. If you mean "a queue that minimizes the amount of time that people have to spend physically sitting in their cars", then no.
Frankly, even though you've been to Burning Man, I don't feel you're competent to be suggesting improvements to the exodus traffic problem. You've been there once, and you came in and left on a fast-track bus which was able to bypass the entire process you're trying to critique. Had you been through it a few times and seen how traffic exodus works for people who are actually in the thick of it, I doubt you'd have bothered trying to micromanage the queues.
This is not an answer to the question posed in the article, which was, "Why wouldn't this work?"
Your previous paragraphs were presumably an attempt to answer that question. However I clarified that I think it was based on a misunderstanding of what I meant by "reduced wait time".
The whole article was posed as a question: "If this wouldn't work, why wouldn't it work?"
Very few people have made much of an effort to answer the actual question.