Connectionists: Reflections on Models

[Brian Mingus]

Daniel, thank you for the pointer. A quick look at it suggests that it is
full of "dangerous ideas." Who among us is really qualified to read it?
What if those ideas go viral. Is it safe?

We can hardly abandon the search for truth. Yet, we must recognize it has
global security implications. Instead of jumping to conclusions that impose
top-down constraints on what it means to be human, we should be more
bottom-up, looking for those truisms about being human that are essentially
self-evident. As it is safe to do so, we can peek at science and slowly
evolve our understanding together. Also, we must be careful that we don't
use the search-for-truth itself as an excuse for getting caught in cycles
of deviance that directly cause social havoc. If models are an inspiration
for you, relatively uneducated people are excellent at PAC-learning. They
know good-enough when they see it.

Brian



On Fri, Feb 14, 2020 at 2:53 PM Levine, Daniel S <levine at uta.edu> wrote:

> When you say “All models are wrong,” this reminds me of the statement in
> Kurt Vonnegut’s *Cat’s Cradle* by one of the characters who had given up
> the search for truth in favor of the search for useful lies.
>
>
>
> *From:* Connectionists <connectionists-bounces at mailman.srv.cs.cmu.edu> *On
> Behalf Of *Brian Mingus
> *Sent:* Friday, February 14, 2020 2:34 PM
> *To:* Connectionists <connectionists at cs.cmu.edu>
> *Subject:* Connectionists: Reflections on Models
>
>
>
> The most important thing to know about models is that they affect your
> perception of reality. If you or your group becomes infected by a
> sufficiently unrealistic model, the consequences are devastating.
> Maintaining an unrealistic model as your worldview leads immediately to a
> kind of dissociation and creates and over time continues to magnify your
> resulting blind spots. These blind spots, if not rectified, promise, very
> sincerely, to end the species.
>
> Consider the following models:
>
> - All models are wrong. This model hitches a ride on the truism that
> “nothing is necessary.” It is a trivial truth. Outside the synthetic
> aperture of mathematics the real world is a messy place, and you ought to,
> as a child, be made comfortable with that. Whereas “all models are wrong”
> puts you in the position of staring down a never-ending existential crisis
> - a permanent instability that fails to establish itself as a safe platform
> for the future of the species.
>
> - All models are wrong, but some are useful. This model is elitist,
> allowing the person rocking it as their worldview to have a feeling of
> superiority over other people, such as found in Nassim Taleb. The reality
> is that all models are useful. In the worst cases, we can use a model as
> part of a Rube Goldberg machine that solves a problem, such as throwing bad
> models into the landfill of time.
>
> - The No Free Lunch Theorem. This model says that for every model, there
> is a counter-model. It concludes that it is therefore impossible to have
> knowledge. In reality, the No Free Lunch Theorems simplify to the negation
> operation - for anything that is said, simply add that it might also be the
> case that it is not true. The NFL theorems are nothing more than a
> mathematical attack on your ego, strengthened by your inability to
> penetrate the extremely advanced mathematics used to “prove” them to you.
>
> Indeed, the entire idea of a model could be considered to be an attack on
> your individual person. As a flawed conception of a human being in the
> world, it leads to an endless cycle of navel-gazing-induced existential
> crises.
>
> Models, in their never-ending ambiguity, promise to erode all that is
> important to people. Our ability to love, have children who are optimistic
> stewards of the future of humanity, enjoy our lives and to live with
> confidence fall flat in the face of one simple operation - negation.
> Instead of negation, try optimistic, positive psychology and having respect
> and warm feelings and intentions for your fellow humans - all of them -
> 100%.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20200214/5a8eae53/attachment.html>