Hacker News new | past | comments | ask | show | jobs | submit login
Researchers Map The Sexual Network Of An Entire High School (osu.edu)
143 points by chaostheory on Sept 25, 2010 | hide | past | favorite | 56 comments



This article is from several years ago. When I read it then the thing that stood out to me is that the graph doesn't display the timing of the sexual relationships. So to say that someone who had sex with just one person is exposed to everyone in the graph is silly, because that relationship might have happened when both were virgins, and hence no such exposure happened.

I sent an email to James Moody pointing out this obvious flaw in the conclusions, and he responded saying that he agreed. His original work was not intended to show the chronology, and the article draws conclusions that he didn't claim.


I think the problem is that he is describing the graph as if it is undirected, when for the purposed of STDs, it should be directed.


I don't see how direction solves the problem. Wouldn't you need a weighted graph to encode the timeline?


Does it matter? Wherever there is a link, (IE persons A and B both slept with person C) there was an exposure. Unless they were simultaneous, in which case I went to the wrong high school.


Are you saying you can get infected retroactively, when one of your past lovers gets infected? That's harsh.


In your example: if person A slept with person C before B did then B has an exposure to A, but A has no exposure to B.


Right, so there's one exposure regardless. You don't know who it is I suppose, but you know it's there.


AB then CD then BC: the graph looks like A-B-C-D but there's no chain A->D or D->A that actually permits infection.


Obligatory:

http://sexchart.org/sexchart.9.15

(This charts the relationships of the informal collection of people from #hack, #phreak, and god knows what other EFNet "underground" channels).

A smart guy we hired wrote a traceroute for this (including an FSM parser for the ASCII art); in an attempt to one-up that, I wrote one that did all-points shortest path and could answer questions like "how many hops am I from X if you take Y out of the picture". We used it to test our routing code at Sonicity; it was handy.


Ahh, too bad I can't upvote you a 100 points. Words fail me when confronted with this chart. I'm taking up the challenge to parse the ASCII art to convert to Graphviz as my next project.


Just scan it looking for the first in a run of alnum characters, build a table of nodes, and then scan the perimeter of each node looking for a character that starts an edge (`'|- etc). Then write a little state machine for following each edge, and build an adjacency list.

I really ought to use this thing for C class.


Oh man, I think I see the nick of a high school girlfriend on there with a disreputable number. Thankfully that relationship predates the chart.


where's tqbf?


Some pretty interesting patterns there. One of the more interesting was that people who have had lots (5 or more) partners had tended towards 'unpopular' partners (zero or 1 other partners).

Also I wonder if 2 homosexual relationships (that I can see) is in any way accurate or due to embarrassment to admit such a thing even on an anonymous survey.


Though I do not have any statistics and I am not familiar with American high school culture, I believe that not many homosexuals are romantically/sexually active in high school. People in this age are still discovering these things, and I don't think that most of the homosexuals fully realize what their orientation is like. Even if they do, actively seeking partner is looking for trouble in high school reality -- children and teenagers are so cruel towards people different from them. I did not know of any of my school acquaintances to be homosexual, and the possibility of all of them being straight is kind of odd, considering the percentage of homosexuals in society.


"Though I do not have any statistics and I am not familiar with American high school culture, I believe that not many homosexuals are romantically/sexually active in high school. People in this age are still discovering these things, and I don't think that most of the homosexuals fully realize what their orientation is like."

Two things:

1. Yes they are sexually active. They just have to be more discrete.

2. You don't "discover" that you're gay any more than you "discover" that you're straight, and if it does happen it happens before high school. I can think of things going back to third grade that pointed towards me being gay. Now, whether people choose to accept their sexual orientation is a different matter.


> You don't "discover" that you're gay any more than you "discover" that you're straight (...) In perfect society it would be like that. However, the culture I live in revolts around straight relationships, and there's just no place for homosexuality, except of maybe designating some people as "them", in contrast to "us". It sucks, but it is the reality. Children growing in this kind of society come to think that being straight is "default". That's why I do not think that situation is symmetrical -- it takes more effort to acknowledge that you are gay than that you are straight in this kind of culture.


Of the 2 homosexual relationships, one of them is in a triangle with a heterosexual partner. That may be more of a threesome than a purely homosexual relationship.

One interesting pattern that I saw was that, other than the threesome, there only seem to be four cycles--one is extremely long and the other three all involve the same guy who is of degree 9 and serves as a cut vertex to no less than seven subgraphs.


That first observation is really interesting!

I think the homosexual numbers are accurate, at least for the guys. The gay high school students I knew seemed to be more actively engaged with older men.


One of the more interesting was that people who have had lots (5 or more) partners had tended towards 'unpopular' partners (zero or 1 other partners).

It looks like most of the vertices here have degree 1 or 2. Do those adjacent to vertices with degree 5+ actually tend more towards low degree than the rest of the population? (I haven't gone through and counted, but I'm skeptical)


I agree, there doesn't seem to be a degree bias when you compare partners of high-degree vertices to the rest of the population.


What a pity they didn't add the question 'do you practice safe sex' and colored the dots accordingly.

It would be of far more educational value to the group that was studied and all their peers.

Side note: In the US there are about 750,000 teen pregnancies annually.


That was not the point of the study. The point of the study was to examine the relational map within the high school. It was not to examine sexual hygiene of students in the school.

> "It would be of far more educational value to the group that was studied and all their peers."

How is that so?


> That was not the point of the study.

I realize that.

> The point of the study was to examine the relational map within the high school. It was not to examine sexual hygiene of students in the school.

No disagreement there.

>> "It would be of far more educational value to the group that was studied and all their peers." > How is that so?

Because it would show them in a fairly easy to understand graph how STDs spread through such a network.

Having the 'outside world' connected wouldn't hurt either.


> 750,000 teen pregnancies annually

These days, getting pregnant is one of the nicest things that can happen. At least it's neither fatal nor contagious...


~80% of teenage pregnancies are 'unwanted'.


I bet 100% of the STDs are unwanted too.

Don't want to minimize the seriousness of teen pregnancy - it's a really huge problem, but, at least, the baby doesn't kill its host and a couple friends ;-)

And that's why I am a very careful parent. I tell my kids I wish them an uneventful life and death from old age.


Re: Baby Doesn't kill its host

Not to minimize the seriousness of STDs, but I'm pretty certain there is a very long laundry list of STDs that have a lower mortality rate than pregnancy.

The best data I could track from the CDC is from 2007:

http://www.cdc.gov/nchs/fastats/womens_health.htm

"In 2007, a total of 548 women were reported to have died of maternal causes (Tables 33 and 34). As in previous years, the number of maternal deaths does not include all deaths occurring to pregnant women, but only those deaths reported on the death certificate that were assigned to causes related to or aggravated by pregnancy or pregnancy management"

"The maternal mortality rate for 2007 was 12.7 deaths per 100,000 live births. Black women have a substantially higher risk of maternal death than white women. The maternal mortality rate for black women was 26.5, roughly 2.7 times the rate for white women (10.0 deaths per 100,000 live births)."

Of course, there are varying rates at different ages, so...

From: http://www.cdc.gov/NCHS/data/nvsr/nvsr58/nvsr58_19.pdf, Table 11, pg 37+ (unfortunately they don't have a breakout for women at various ages, but I eyeballed Table 12, to see if there was a gender bias and noted)

All Causes for ages 15-24 (all causes 79.9 Deaths per 100,000)

  Syphillis: none
  HIV: 0.4 (2.5x greater chance for men)
  Pregnancy: 0.4 (Always female, so 0.8 per 100,000 for a woman)
  Motor Vehicle Accidents: 24.9 (2x Greater chance for men)
  Suicide: 9.7 (4x greater chance for men)
  Homicide: 13.1 (4x Greater chance for men)
If we can reasonably double the rate of death by pregnancy for women, 15-24, to be 0.8 deaths per 100,000 then a girl, 15-24, is probably more likely to be kill by pregnancy than an STD, but is dwarfed by a Car Accidents (And is actually much smaller than Suicide and Homicides as well)

Admittedly, this is just a statistical survey, and actions like proper medical attention can sway these statistics (though that is the case for STDs as well as pregnancy)


> I bet 100% of the STDs are unwanted too.

Yes, hence my suggestion to ask the question about safe-sex and to add two more sets of dots.

> Don't want to minimize the seriousness of teen pregnancy - it's a really huge problem,

ok

> but, at least, the baby doesn't kill its host

Teen pregnancies are a serious risk to both mother and child.

In developing countries the risks are much higher still.

It is much less rare for a teen-age pregnancy to have complications which may even result in fatalities.

> And that's why I am a very careful parent.

Good :)

> I tell my kids I wish them an uneventful life and death from old age.

Not sure about uneventful, but if by that you mean unscheduled hospital visits then I'm all for it.


Can someone explain to me why I do not see isolated vertices in this graph? When I was in (Central Europe) high school, I recall that apart from me, there was considerable amount of people not involved in any kind of relationships, and even not counting them, many people used to have partners from other schools. Is it really like this in American high schools, or did the researchers just choose this way of presenting the results?


"many people used to have partners from other schools."

One thing the researchers noted is that they chose a large-ish but relatively isolated community with a single high school that's about an hour from a major city. This means that geography dictates that students are relatively unlikely to date people from other high schools.


I believe they omitted students who did not have sexual relationships.


The article states that just over half were sexually active. So they must not be including celibate students.


I meant the graph linked in the adjacent comment: http://researchnews.osu.edu/archive/chainspix.htm The description states that it presents romantic connections, not the sexual ones. However, rereading the article makes me think that they did not take single students into account when creating this graph.


The article said it was in a small town where the second nearest high school would be more than an hour away. Therefore the only 'legal' relationship they could have with people close enough were in Jefferson High School.


Oh, thank you, I must have missed it.


As I studied in a male only institution, we used to rush to the the girls high school, it was a five minutes walk. Weekends were devoted to establishing new connections (trying to), we were struggling to find partners, specially when approaching Christmas. Good old time. I imagine studying in a mixed sex high school should be better to get more relations, but is not so clear. We strive to do it our way.


The actual article mentioned in the link: http://www.soc.washington.edu/users/stovel/Chains.pdf

See also: "The web of human sexual contacts" http://www.nature.com/nature/journal/v411/n6840/full/411907a...

(Non pay-wall version: http://arxiv.org/abs/cond-mat/0106507 )


I don't see a better way to do it, but what would make them think that results based on asking teenagers who they are having sex with would be reliable/actionable? This isn't even addressed in the article?


That's probably because this is a press release/internal news item, not the research paper itself.

That said, I have no idea if they discuss that in the paper either.


I wonder how they handled the case where X claims relationship with Y, but Y doesn't agree.

I once amused myself for a few minutes by dreaming up an app for Facebook that would let you claim a relationship with someone and send that person a message allowing them to confirm. It would then collect statistics like false claims, and potentially produce graphs like the one in this article about your social circle.


I made a sex graph for my social group several years ago (names blocked out). http://bit.ly/bHPjN4

It was enlightening for everyone.


Damn, that square with seven edges is no "square" at all. He/she slept with half your friends of the opposite sex!



"breaking one link in the chain – any link - will stop that part of the network from spreading any further"

Interesting. I wonder whether the "save sex till marriage" movement is having much of an impact in slowing the spread of STDs.

My impression is that they don't change many teenagers' behavior in the long run. But for this specific effect they wouldn't have to change more than a few to have a big impact.


A handful of condoms would have the same effect, and even more people might use them.


I'd be careful with inferences like that. It's not a directed graph, so talking about flow in time is very speculative.


And it does not document promiscuity either (in time), for all we know these are nicely serialized or it could be one big orgy.

It also keeps the relationships within the school, which is not a realistic picture whatsoever, and once you add in the outside world (especially to that larger chunk of the graphic) all bets are off.


are there high rates of STD infection for teens? or is it more prevalent in adults with the mentioned hub networks?


Am I the only one who looked at the diagram and imagined it as a great party planning tool?

Couldn't something like this be quickly built with a school-wide hot-or-not website? People publish far worse stuff on Facebook.

Hmm... Orkut has a network like this, in the form of crush-lists. In fact, I know two couples who were "aided" by this feature.


If they're describing the Moody paper I'm thinking of, this is _Chains of Affection_.

Here's a link (PDF!): http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.131...


(2004)


That, and the picture says everything a lot more succinctly than words: http://researchnews.osu.edu/archive/chainspix.htm


Where is the geek independent set? G = (V, empty)


How do they know that the data they collected in their interviews is reliable?


Teenagers have sex. Who knew.

The interesting part to me is that if you look at individual students in the big group, most have only 1-3 relationships. They just so happen to link up into a massive web.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: