A belated analysis: hypothesis confirmed to my satisfaction - as expected, age is strongly related to Usenet-familiarity. (There's even a hint of a quick shift in the histograms.)

# <https://www.dropbox.com/s/d5kddi1ckl3qbxk/2015-02-26-lw-usenet.csv>
usenet <- read.csv("2015-02-26-lw-usenet.csv", header=FALSE)
usenet2 <- data.frame(Usenet=usenet[1:67,]$V3, Age=usenet[68:134,]$V3)
## the default LW poll encoding is yes=0, no=1; this is very confusing, so let's reverse it
usenet2$Usenet <- (usenet2$Usenet==0)
wilcox.test(Age ~ Usenet, data=usenet2)
#
# Wilcoxon rank sum test with continuity correction
#
# data: Age by Usenet
# W = 183.5, p-value = 3.893e-06
g <- glm(Usenet ~ Age, data=usenet2, family="binomial"); summary(g)
# ...Coefficients:
# Estimate Std. Error z value Pr(>|z|)
# (Intercept) -7.33329350 1.93381008 -3.79215 0.00014935
# Age 0.24311715 0.06695693 3.63095 0.00028238
## alternative plot:
# with(usenet2, plot(Age,Usenet,xlab="Age",ylab="Probability of Usenet familiarity"))
# curve(predict(g,data.frame(Age=x),type="resp"),add=TRUE)
# points(usenet2$Age,fitted(g),pch=20)
library(popbio)
with(usenet2, logi.hist.plot(Age,Usenet,boxp=FALSE,type="hist",col="gray"))
## <https://i.imgur.com/W6DT9Tu.png>

## specific example: 51yo vs 20yo probabilities based on the model: predict(g, data.frame(Age=51), type="response")
# 1
# 0.9937299491
predict(g, data.frame(Age=20), type="response")
# 1
# 0.07791991187

(If anyone is curious, my original motive was wondering about Satoshi Nakamoto & Nick Szabo - both are familiar with and have used Usenet. We already know Szabo is old and very similar to LWers, so being Usenet-familiar turns out to be entirely ordinary as I guessed, but if Satoshi Nakamoto were a young college student as some people thought, then being Usenet-familiar is pretty surprising.)