There're some questionable assumptions in this paper, notably this one:
"What we really want is to minimize the number of infections resulting from any given number of sexual encounters; the flip side of this observation is that it is desirable to maximize the number of (consensual) sexual encounters leading up to any given number of infections."
He's saying that there's this ratio, infections per sexual encounter, that they want to minimize. And the easiest way to minimize this is to increase the denominator. If you have a universe with a skewed (self-selected) distribution, then as you add more people to a category, you'll increasingly get more marginal people. In this case, it's people who have fewer sexual partners and hence are less likely to carry AIDS, resulting in a reduced chance of infection.
Whether this is the metric you care about is debatable. If you're a public health official, you probably care about the absolute number of infections, not the number per sexual encounter, and so this only helps if it brings the probability low enough that the disease can't sustain itself.
Similarly, if you're a married guy who just wants a faithful and devoted wife and kids, you don't care about the chance of infection per sexual encounter, you care about the absolute chance of infection within your life time. This is minimized by not having sex with anyone, or, failing that, having sex with someone who only has sex with you.
Only people who regularly have sex with lots of partners would benefit from this - for them, the metric is chance of infection per sexual encounter (since they'll be having lots of sexual encounters anyway), so why shouldn't they?
I read his book of the same title. It is entertaining, but I agree that he makes some questionable interesting assumptions. I get the feeling that he starts with a conclusion that he knows will get some attention or likes -- and then works backwards, even when one must use a suspension of disbelief like when reading fiction.
Even if the sexual contact network is not a hub oriented "small-world network" and more like our highway system -- that doesn't seem to mean that diseases will stay isolated in one area, just it might not spread quite as quickly or pervasively.
"If you're a public health official, you probably care about the absolute number of infections, not the number per sexual encounter"
Perhaps they made the assumption that the total number of sexual encounters over the entire population is roughly constant. With this assumption the distinction would not matter.
They're arguing for "increased sexual activity by conservatives" which obviously means they're arguing for increase in total number of sexual encounters over the entire population.
No, they argue that this would lead to less demand for prostitutes, which would balance it out.
In effect they say that disease would spread more slowly in a graph with less highly connected hubs and more balanced number of connections per node. This is quite unsurprising.
Sure, this is quite obvious though I did not understand how they could balance out the increase, so I did not even mention this option.
The point about prostitutes was just a thought experiment, and it looks rather far from reality. The second thought experiment about Joan hooking up with Maxwell seems much more realistic, but I don't see why Maxwell wouldn't hook up with someone else the same day. Actually, the more promiscuous side usually takes the initiative, so I'd say it is quite likely that if not for Joan, some other poor girl would get AIDS anyway.
That may be the case; but it doesn't affect the theory. What's important is reducing the standard deviation from the average number of sexual partners.
Or to put it another way; to reduce the outliers who take a disproportionately high number of lovers. It is logical that this would hamper the "global" spread of disease.
This particular theory is saying that the way to do that is to reduce the opposite form of outlier - the sexual conservative.
nostrademons, I'm confused why you would wonder about the viewpoint of a public health official. Are you in government yourself? Do you often view things from the point of view of someone working in government? Is there anything about the opinion of a public health official that normally interests you?
Why would you assume these people are government employees? Even if they are, what difference would that make?
On that note, I know a lot of people in the public health field and not one works for a government. Virtually all of them are indirectly paid by the Bill and Melinda Gates Foundation, through one circuitous path or another.
HIV infections per encounter will go down if you have a lot of sex, this is true.
And it will to zero forever once you get infected.
But its the likelihood of infection overall that I am concerned with. The more sex you have, the likelier it is that you will have an infective encounter earlier.
the fact remains that if people never have sex, they will never get infected. If they only have sex with one person in their life, and that person follows the same rule, no one will be infected either.
But that's not reality. so we should follow some simple rules, which can be found at http://evensafersex.com
Although tfa is slim on details, it seems that the economist is assuming that the total number of sexual encounters per unit time is constant, and that increasing the activity of some people would only change the distribution in the number of sexual contacts.
Fair enough-- if you could change the distribution, that would probably slow down the epidemic. But, counter intuitively, it would also increase the number ultimately infected.
The article made the mistake of analyzing only a tiny piece of the system and attempting to draw conclusions for the whole system.
One of the mistakes the article makes is in the statement that, if "conservative" girls went out twice as often, the chances of an HIV transfer are decreased because the guy has twice as many potential partners who are HIV-negative. What's missing is the recognition that, if the guy is already HIV positive, he now has twice as many potential partners who he can pass the infection on to.
The changed network could either slow down or speed up the spread of HIV, depending on various specific details (how many men have HIV, how many women have HIV, how many connections to non-carriers do the men have vs the women, and so on.)
As you correctly note, even if this particular change slows down the rate of infection in the present, it may ultimately increase the total number of infections. It may even slow the infection rate for a few years and then suddenly speed it up as some particular critical threshold is reached.
This article sounds a lot like Russian roulette. Increasing the number of chambers making it safer to play. It's not clear to me that the virus would just die out. I'll probably end up talking out of my hat but with graph of all the connections (sexual encounters). Would it not be possible to show that you'd need a very connected graph where most people have a small number of encounters and that almost everyone sticks to a low number. One hyper connected infected person (or multitple), could end up infecting a large part of the graph. I would assume that if he affects a large enough number of people, they (before discovering their infection) could infect their connection and ... snowball. In their model, they seem to say that even if their are very promiscuous people, getting non promiscuous people to get more connected would improve the odds. I don't see how that would reduce the number of infected people and kill the virus.
Like I said, I have no clue if this is correct or complete crap, never done anything related to disease and how they spread.
I think what the article argues is exactly this: there would be less hyper-connected nodes if the nodes would be more locally connected. The spreading of the disease would happen much more slowly in this case. This fits in with the theory of scale-free networks.
Its a dynamic function. At the beginning, infections go down but only because there was a pool of uninfected people. As time goes by, this pool diminishes (they are having more sex, right?) Eventually everyone is infected (markov chain). How is that better? Its just bad math.
mmm, that what I think would happen, at least it's what seems more intuitive. If you could remove the very promiscuous infected people from the pool of available partners quickly (better detection) that could solve the issues I think. The article does point out that increasing the encounters of monogamous people would reduce the amount of encounters from the most promiscuous people. Not sure this would occur, what says the promiscuous people would have more encounters now that the pool of available partners increased.
1. It would be pretty simple to write a computer simulation to test this, no? Would anyone care to do so?
2. I don't know enough about the AIDS situation in Africa to say whether it is empirical evidence for or against this paper. Is anyone else more familiar with this?
Preliminary results: holding sex drive and promiscuity constant, decreasing societal shyness (i.e. making single people more likely to go out and meet someone to fuck) seems to increase population infection rates rather than decreasing them. This is true if shyness is independent from promiscuity. (My simulation tends to send more promiscuous people out to meet someone to fuck more often by increasing their partner-switching rate, so shyness is indeed independent.)
I took game theory back in college. I noticed something wrong about that article: a complete lack of math. It's not that I don't believe there are counter-intuitive things that can happen. But I sure don't believe them without proof.
And your simulation is a lot more interesting than anything found in that article.
It was a fun exercise. It was great for remembering a handful of things about statistics and experiment design--for instance, it was impossible to get good results until I started doing controlled runs on the same randomly generated population rather than regenerating, because the effect of changing social shyness (while there) is way, way smaller than the effect of the random variations in promiscuity and sex drive.
I may want to play with adding condoms to the simulation, though. My transmission rate is based upon the HIV transmission rate for unprotected anal sex, which is pretty clearly an upper bound--that's why especially slutty or outgoing populations get >50% infection rates after 20 years seeding from 1% infection rate. You also see the odd run where you just have a 1% infection rate after 20 years--probably cases where the randomly seeded HIV+ individual is especially shy and has a low sex drive.
Also, it would have been a lot more interesting if I could have generated random values along different distributions, or...well, lots of things would have helped.
Articles like this are what we should find on Digg. I don't want to be that snooty bastard, even though I am.... I'm not sure how this is relevant to Hacker News? However, I'm impressed by the mature comments. Can we not turn into Digg?
No, check out philwelch's awesome simulation a few comments up testing the hypothesis (and finding it wanting). This is what happens to this kind of story on HN. Not only is it awesome, its the proof that HN is still different and not slipping any time soon.
It seems a lot of people are concerned about HN turning into other communities. How is this avoidable? Require everyone to debug a snippet of code before registering or logging in? I personally wouldn't mind this.
Don't worry about it and it's all good. HN is self-balancing. I've been here for 884 days according to my profile, and I've always been hearing these worries since day 1. Somehow, thankfully, HN works. It has not turned into Reddit, Digg, or MySpace regardless of what people say. Yes, it has changed over time, no it's not a worse place.
Yeah, kind of thought so--a little diversity can be refreshing imo. I noticed that HN redirects you to a previous submission if you use the same url. Perhaps the same could be done if you post something with the same url used on a previous reddit or digg submission.
It's clearly a hacker question. There's math, simulation, and girls involved. A true hacker can turn interesting articles like this into very hacker-ish engineering puzzles - read the other comments and see what I mean.
"What we really want is to minimize the number of infections resulting from any given number of sexual encounters; the flip side of this observation is that it is desirable to maximize the number of (consensual) sexual encounters leading up to any given number of infections."
He's saying that there's this ratio, infections per sexual encounter, that they want to minimize. And the easiest way to minimize this is to increase the denominator. If you have a universe with a skewed (self-selected) distribution, then as you add more people to a category, you'll increasingly get more marginal people. In this case, it's people who have fewer sexual partners and hence are less likely to carry AIDS, resulting in a reduced chance of infection.
Whether this is the metric you care about is debatable. If you're a public health official, you probably care about the absolute number of infections, not the number per sexual encounter, and so this only helps if it brings the probability low enough that the disease can't sustain itself.
Similarly, if you're a married guy who just wants a faithful and devoted wife and kids, you don't care about the chance of infection per sexual encounter, you care about the absolute chance of infection within your life time. This is minimized by not having sex with anyone, or, failing that, having sex with someone who only has sex with you.
Only people who regularly have sex with lots of partners would benefit from this - for them, the metric is chance of infection per sexual encounter (since they'll be having lots of sexual encounters anyway), so why shouldn't they?