Monthly Archives: March 2021

The Lack of Courage and Clarity in the Rush to Judgement against Andrea Bertozzi.

I was disappointed, but not at all surprised, by the news that the Association for Women in Mathematics (AWM) had recently decided to condemn Andrea Bertozzi, through facts that turn out (after careful, nuanced investigation) to not be facts at all. The evident lack of care is not surprising, but the incidents point to much deeper problems.

I am referring to the articles you can find here.

Now, while I deeply respect the right of the article and letter writers to write what they wrote, the whole spirit of this ongoing saga, with its deeply religious/ideological perspective — one that crosses the line repeatedly to the “the end justifies the means” principle — is something I find very disturbing.

I suppose that if was in favor of policing as it is, thought that racism was not a problem, or believed that the left/progressive side of things has nothing to offer, I would shrug this off as another example of the “idiotic leftest academics” not getting their facts straight.

But I don’t believe those things.

In fact, I have had very deep problems with the way policing is done, with the cultures that are tolerated, with the extreme lack of wisdom with the new fad of machine learning for everything, and the fact that the data we use is extremely problematic.

So why am I not a fan of this edition of the AWM newsletter?

To begin with, the facts, with respect to Andrea Bertozzi, are wrong. She did not come up with the core algorithm developed by PredPol, the paper by Lum and Isaac did not even use the actual PredPol algorithm in daily use by the company (that, again, Andrea did not develop), and, as can be seen by a careful look here, while Andrea’s name was on the list of organizers, she did not even attend the workshop on predictive policing at ICERM, so she could not have actually led the event. As mentioned above, the PredPol algorithm was not developed by Andrea, as evidenced by the fact she is not on the patent underlying the company and the one paper that has her name on it, was a paper to which she contributed only a very small amount, but again, long after the invention of the method (which is based on a model of earthquake aftershocks). Additionally, the Lum and Isaac claim that PrePol attempts to “identify future offenders” cannot be true, since Prepol only uses the distribution in time and space of crimes (including their types) in their prediction calculations, without any personal data or identification of previous criminals being used. And as others have noted, Lum and Isaac apply this to the one type of data the PrePol algorithm is not intended to be used with — drug offender data. The fact that the perpetrator and victim are usually the same person in drug crimes, is very significant. Finally, since the article was clearly intended for the general public, the speculation and hypothetical nature of various assumptions should have been couched in much more careful language. The “scientism” of the general public often leads them (and, actually, too many scientists) to treat hypothesis and conjecture as fact. (It was only clear that they used the algorithm published openly by those associated with PrePol, instead of the proprietary version used in reality, in their 2018 Medium article. Those acquainted with how what is used in practice differs from what is released to the public, in companies like PrePol, understand the significance of this difference. While Lum and Isaac could not be expected to use the proprietary version, the fact they were not, should have been made very, very clear, especially given the certainty of the controversy the paper would generate.)

There is another fact that seems lost on the writers in the AWM newsletter — the study and description of crime (just one of Andrea’s many threads of research) is not the same thing as its use and does not dictate whether that use is positive or negative. That is up to the (non-academic) culture that the activists are correctly wanting to change.

A deeper, more nuanced analysis, looking for the true source of the negative uses of what academics discover, finds it in academic culture itself, in the culture of separation, disconnection and lack of grounding in the barehanded reality of the world we live in. (As a result, everyone in this saga of judgement and protest against judgement is implicated (to some extent) in doing too little to connect and be relevant — and that includes me!) I am convinced that if those academics that study and discover (all sorts of) things were deeply grounded and connected to the nuanced reality we live in, their empathy would kick in and help them craft how they innovate and how the connect so that the uses of what they discover would be positive.

But, with the facts wrong, and nuance jettisoned, there is no way that the freedom I support in terms of what was written in the newsletter will also get my agreement or applause.

What makes this all the more distasteful is the fact that I agree that data science and machine learning are very frequently, deeply unwise and damaging, and that trauma, the true elephant in the room, is screaming at us to be healed, yet is in actual practice, ignored. We would rather give that trauma a pill to silence it, immerse ourselves in struggles against symptoms, with a deep sense of self-righteousness and an invigorating belief that our adoption of “the ends justifies the means” is OK because what we fight for is so clearly, so good.

For what is being aimed at is truly good. And this is clear to anyone with any kindness or love.

Because, those not outraged by the way many young black men are treated by the police are truly calloused and lacking in basic human decency. Anybody not deeply disturbed by the outrage of what is called the justice system in the US clearly needs help, empathetically speaking, and the fact that so many groups of people have been mistreated in systematic ways is depressing when the comprehension begins to grasp the enormity of it all.

I suspect that the difference, for me, originates in the experience of being raised very religiously, leading to a deep acquaintance with a very wide variety of very enthusiastic coreligionists. While eventually I was led to a personal spiritual walk that might be described as primitive Christianity, not attached to any particular denomination, that did not happen before I developed a deep allergy and acute sensitivity to self-righteousness, to good people believing that “the end justifies the means” if the end you are trying to reach is good enough. I suspect that many of the (typically) privileged souls that are driving the kind of thing we see in the AWM newsletter, do not have similar experiences that would have, at least to some extent, inoculated them against self-righteousness and misguided principles.

But there is something even more disturbing — and that is the silence of almost all the leading mathematicians in this matter. On the other hand, this is not surprising. Prestige and advancement in academia (and in society in general) is systemically biased against boldness, against nuance and wisdom and towards cleverness that masquerades as wisdom, and towards attention seeking that masquerades as boldness. The result is “leaders” and “experts” that lack real courage, have little to no wisdom, and no willingness to risk their social capital for the principles that are the actual foundation of the freedom and rich flourishing that make living a joy.

This is a deep shame.

Because the fundamental aims of the activist — stop the killing, remove prejudice, create a human culture in which everyone can truly flourish — are incredibly important and timely. It seems that there are reasons to think that the current chaos could, with wisdom and true empathy, be transformed into real progress.

But the misguided principles in evidence, the lack of courage and wisdom that can be seen from spending the time to think and see and hear and feel all suggest that this opportunity will be missed.

But, let us say for the sake of argument, that Andrea was actually much more involved in predictive policing (which she is not), was actually making money from the technology (which she is not and did not), and was actually somewhat insensitive to the plight of the criminals police are trying to obstruct (which again, she is not). Even under these (false!) hypotheses, a fair and balanced look at her, to determine whether or not she should be uninvited to give talks (as she has been) or that honors should be rescinded (as they have been) would have to look at the rest of what she does.

Because, even though this is an unpopular idea in the current climate of instant, knee-jerk shaming of people not in your tribe, the whole person must be understood to correctly assess any action of that whole person.

Such a look would be deeply revealing.

Those looking would rapidly find a woman who has bent over backwards to help junior mathematicians, both men and women, who has worked hard to create opportunities for students and postdocs. Digging deeper, they would find some of odds stacked against her that she overcame, that helped encourage her to become the encouraging champion of younger people. They would find, at her core, a very warm heart and someone who is fundamentally kind, even though her energy and enthusiasm for progress can make that hard to see at times when she moves fast and makes things happen.

If in fact, those that have a problem with her took the time to sit down with her, they would see this side of her as well, and the more reasonable ones might see the wisdom in starting a debate with her, instead of a war. Because, they would understand whatever mistakes she makes (she is human) are precisely the kind of mistakes they have made themselves!

In a nutshell: such a strategy would not only have a higher chance of effecting change, it would also reveal that the negative things they observe are not a result of any kind of racism, but in fact are much, much more nuanced and complex — in the same way that they themselves are complex and a mixture of light and dark.

Of course, it is rare to find people who are honest and disciplined to this extent, because it moves them out of their comfort zone (which is another of Bryan Stevenson’s principles of change — be willing to move out of your comfort zone!)

In the end though, Andrea is not actively involved in predictive policing, she is not involved in the company PrePol, she did not actually make money from the company, she did not attend, let alone lead, the workshop at ICERM (that her name was on the organizer list is actually a clear indication of her generosity in lending her stature to those that ask for help), and her real interest is in the science of crime — something that is not the same as predictive policing, any more than biochemistry is the same as Purdue Pharma’s deep abuse of the products of biochemistry).

Actually though, this whole saga is no surprise if you believe, as I do, that the trauma almost all humans experience, is the true source of misery in the world. Trauma is experienced by the vast majority of people, but is almost never effectively dealt with. This unhealed trauma leads to a seemingly infinite variety of dysfunctions and disconnections that in turn opens the door for mistreatment and atrophied empathy. I am speaking from experience here: both my brother and I ended up with severe PTSD after watching both parents die from cancer over an 8 year period, starting when we were both preteens. In addition to that, I experienced severe psychological abuse from an (extended) family member that took years to recognize and deal with.

All this has convinced me that the methods themselves dictate the results, that the means has in itself the seeds for the end you actually reach, and that love and healing are at the roots of all means moving us to any truly good end.

So, while I am a general foe of the idea that machine learning can add anything of value to the world, while I believe that the culture of policing must change, and even more radically, that love is the only thing that cannot be defeated in the long run, I cannot align myself with much of what currently claims to be for progress.

In the same way that so much Christianity has too often aligned itself with evil in spite of its truly phenomenal foundation and (I believe) divine origin, social activism that uses impure principles to get ahead is doomed to fail.

Being intent on helping create an increase in human flourishing, means, for me, that I have to use my freedom to protest the methods of those who claim to be for progress, but instead are adopting methods and strategies that auger against their success.

There is one last point that I believe has been lost because of the lack of imagination that seems to have overtaken our current culture. Instead of simply opposing all predictive policing and “mathematics of crime” methods as fundamentally racist, why not attempt to inject corrective ideas and thoughts into what gets produced? This would have the advantage of simultaneously correcting what is done and building defensible trust between the police and the public. It should be clear to careful observer on both sides, that policing would be far, far more effective and humane with an informed, engaged and supportive general public. The data that could be used to help solve the trouble at the roots of the current crisis is actually only going to be available in a society that is characterized by trust. The evidence for this is very, very strong. It is also clear that in a society based on trust, tools like data analysis can be positive. While I am for thick data and grounded, action based on insights like those in Bryan Stevenson’s Just Mercy, and I also believe the knee-jerk “big data” and “machine learning for everything” instincts are dangerously foolish, tools that mathematics gives us can, with wisdom and constraint, be very useful.

With enough imagination and creativity, in an environment rich with thick data, filled with people having widely different experiences and perspectives, in an environment with a dedication to taking the time to think and see and hear and feel, and a real focus on wisdom and groundedness, the tools of mathematics and the hearts of healed humans can help craft a future that anyone would want to be a part of and in which everyone is welcome.

I suspect, that if I were able to sit down with the letter and article writers, in the patient proximity that Bryan Stevenson so effectively argues for, and we were to actually see and hear each other, the power of that proximity and basic human empathy would enable us to see a path forward based on nuanced awareness of the barehanded facts of each other’s lives. I can easily imagine that while we might not agree on everything, we would agree deeply on the supremacy of the goal of human flourishing for everyone.