Tuesday: Turns out that skewness isn't as helpful as I'd hoped

My thought was that based on the assumption that an underlying distribution is symmetric (so, Gaussian distributed; or, Poissonian distributed with N large, such that it's effectively Gaussian distributed), you can use the measure of skewness to reweight the samples into a symmetric form, and by doing so, downweight outliers.

It still might work, but it's not clear that it's going to do what I want.  Part of the problem is that extreme outliers push the skewness above 1.0, which means the reweighting starts to fall apart.  I suspect that this means I need to measure the skew, determine if it suggests multimodality, remove outliers, redo the skew, and then reweight.  Unfortunately, this starts to get to the point where I start to think that it's mathematically unjustifiable.

Thanks for the vote of confidence, Peebles.

Wait...squirrels don't eat hot dogs.  No, this is all wrong.

• Bunny!
• All of these links are interesting, but I think this one about the dismantling of Utrecht and this one about the problems with higher learning are the most interesting.
• '“We had a lot of young people here as delegates. And they had input and they were part of the process," said Sen. John Hoeven, the platform committee co-chair. "And I think if you talked to them, they went away feeling that they had a big say and a big part in the document that was crafted.”'  Seriously.  I can't wait for the current batch of Republicans to just die already.  Pandering to crazy old racists is pretty much guaranteeing that they won't last long.
• Flying squirrel!