Mathematicians created various notations to make complex ideas easier to remember, keep track of, and communicate. But sometimes we forget this. Instead we start to imagine that notation (in and of itself) somehow makes our work more rigorous, more objective, or more accurate. The first step towards mathtimidation.

Eventually, this transforms our shorthand into a way of drawing group boundaries, asserting authority, and silencing questions. It becomes a performance that loses it's initial purpose of simple communication.

For a concrete example, I see this all the time at presentations by modelers. They will throw a slide of equations up, maybe some other mathematician will ask something about it. None of the domain experts -- be they biologists, clinicians, social scientists -- will feel that they are qualified to comment on that slide. But that slide is the model, and the part that should be transparent to all and communicated to all. Which can often be done easily (and just as 'rigorously') with something more accessible like an interaction network, but without alienating people that don't read systems of ODEs for a living.

In this talk, +Cathy O'Neil follows this line of reasoning from the small weapons of mathtimidation to the weapons of math destruction made possible by the algorithms intervening in our lives. She stresses the importance of transparency. Of course, this transparency doesn't just mean throwing some code up on GitHub; sure, that is more public than some of the opaqueness that Cathy highlights, but it is still not a great way to inform or communicate to the people affected by the algorithm.

In an open democracy, we have to understand (or at least it should be reasonable to expect us to understand) the rules of the systems that affect us. That way we can be properly informed in our decisions, and can more easily see and critique the weaknesses of those rules. We also need to realize that in this realm, objective is -- more often than not -- a weasel word that we use as a short-hand notation for culturally engrained opinion. Something that can easily perpetuate social injustice.

/cc +Arturo Araujo, +John Baez, +David Basanta, +Suresh Venkatasubramanian 

Pranav Warman, +David Basanta, and I know how to spend our Friday nights: building models of prostate cancer in the bone. Yesterday on #TheEGG , I posted a quick write up of the model that we hacked together.

After some reflection this weekend and discussion today, I don't think that this captures exactly the intuitions we want, but it is still a fun model to play with. +Arturo Araujo might enjoy it, and can probably point out the places where we fail to capture the proper intuitions, since those intuitions mostly come from his much more extensive work on the topic.

Last night on #TheEGG, I published a technical post on the stability of fixed points in tag-based models under replicator dynamics. But I disguised it with a long into and footnotes on moral and political philosophy. This was mostly inspired by +David Pizarro and +Tamler Sommers of the Very Bad Wizards podcast. There is no central thesis to this post, since it is a sketch of on-going thoughts and work, but I hope it amuses some.

/cc +Nick Byrd, +Abel Molina, +Alexander Yartsev

Is the Internet like the lead lined aqueducts of antiquity?

In this article, Rose Eveleth shares Finn Brunton's analogy between the Internet and lead pipes. With two thousand years of hindsight, will we look back on today as indulging in a status granting activity that we kind of knew was bad for us?

I do like looking at the symptoms mentioned for lead poisoning and comparing them to typical complaints of heavy net users:

"According to the EPA lead poisoning can cause behavior and learning problems, lower IQ, hyperactivity, slowed growth, hearing problems, decreased kidney function, reproductive problems, and more."

At least the Internet doesn't decrease kidney function, eh? Although more seriously, I suspect the severity of the ill effects -- except maybe hyperactivity and attention deficits -- is much lower for the Internet. And some features, like learning and IQ effects, might even reverse if you use the Internet for the right things. But good luck with that. The severity being lower doesn't alleviate the problem, though, if a much larger population is effected. This is the largest disanalogy for me: since when is the Internet a status symbol for the aristocrats? Unless we consider all of western society as the global aristocrats, but even that is being changed in terms of web access.

I'll close with a quote and let you guess if it is about the Romans or us:

"this thing that was simultaneously a fascinating part of how their culture worked, and the invention of a new kind of urban living but also as something that was slowly but surely making the ruling class into people who were desperately ill with terrible impulse control without ever realizing it or understanding why."

/cc +Forrest Barnum, +David Basanta, +Abel Molina, +Dirk Puehl

A nice article via +Russ Abbott on the abuse of autonomous cars as a way to motivate trolley problems. I think it can be summed up with an old adage from Donald Knuth: "premature optimization is the root of all evil".

I've wanted to write an article like this for a while, so I am happy that Brett Rose did it for me. However, there are two threads that I would have liked to see added:

[1] The life or death ethical design decisions involved in this are not new to engineers. People imagine that everything else is a simple optimization problem with a clear peak for them and no ethical content. But that isn't the case. For example, consider designing an airbag. If you make it deploy too hard, it might help save a heavier person but kill a lighter one. If you make it deploy too light, it might not save the heavier person. How to choose?  And this isn't even a crazy hypothetical. I am sure you could come up with a fat-fetched trolley problem equivalent for airbags, and that would be at least more fun to read since it would require more creativity than the autonomous vehicle case and probably be equally far-fetched.

Maybe engineers like +Edward Morbius or +Yonatan Zunger can weigh in.

[2] This does not mean that morality and ethical algorithms aren't important to software engineers. But the real difficult and important problems aren't about trolleys and autonomous cars, but how we manage data and how we use it to affect life outcomes for people. Does a credit card score  calculation algorithm inadvertently disempower people? Is your predictive ad sexist? There are the real questions and they are much better addressed by +Cathy O'Neil and +Suresh Venkatasubramanian. The ethics of autonomous vehicles craze just distracts us about these impossible but empathy-yanking scenarios (oh no, 10 kids on the road!) to blind us to the less 'sexy' but much more important effects that weapons of math destruction are having on us (read more: https://egtheory.wordpress.com/2014/09/05/ethics-of-big-data/).

Synthetic ecology is being introduced as an experimental technique at the Max Planck Institute for Chemical Ecology. This is basically lab-based evolutionary game theory. The experimentalists engineer strategies in bacteria that might not exist in nature and then watch them compete in well mixed media or biofilms. In this case, the researchers showed that cooperation between two cross-feeding bacteria could be sustained against defectors through spatial segregation in a biofilm but not in well mixed solution. Their work combines heuristic models, simulations, and well-controlled experiments in a way that +David Basanta, +Jacob Scott, +Robert Vander Velde, and I would love to do.

For the whole paper, see:

Pande, S., Kaftan, F., Lang, S., Svatoš, A., Germerodt, S., Kost, C. (2015). Privatization of cooperative benefits stabilizes mutualistic cross-feeding interactions in spatially structured environments. The ISME Journal. DOI:10.1038/ismej.2015.212


The problem with conspiracy theories is that they aren't always false. In this case, it seems like ExxonMobile and the Koch family are the central players of a small network that has a big punch.

A few months ago, +Helga Vierich left a detailed and insightful comment on #TheEGG. She discusses the importance of reputation and image scoring to cooperation among hunter-gatherers, and provides some insights into their social structure.

It took me nearly three months to approve it because I knew that I should write an equally detailed reply. I am very happy to have such great engagement from the blog's community, and I apologize to Helga for taking so long to respond.

This morning on #TheEGG  is a technical post remembering some of the progress that +Robert Vander Velde, +David Basanta, +Jacob Scott and I have been making on our double public goods generalization and synthesis of Archetti (2013,2014). I know that +Philipp Altrock and +Philip Gerlee do not like the approach we are taking here, so maybe they will have some constructive criticism to add in the blog's comments.

Last night I posted a new analytic listdex to #TheEGG. Or at least that is the fancy name I want to give to my annotated collections of links. Especially when there is more annotations than links.

The post opens with a tale of +Jacob Scott and has shout ours to +Joerg Fliege and +Jordan Peacock. It might also be of interest to +Sergio Graziosi and +Abel Molina. If you've been following my stream then these links, and my comments, are nothing new for you. They are just in one place and a bit more tied together.