a7fdd45bf2
Change-Id: If161fd8b6c96d66aa63cfb22f8a8bb26d71c5caf Reviewed-on: https://cl.tvl.fyi/c/depot/+/6885 Reviewed-by: tazjin <tazjin@tvl.su> Autosubmit: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
142 lines
4.5 KiB
Markdown
142 lines
4.5 KiB
Markdown
<!--
|
|
|
|
This file contains a bunch of random thoughts I don't want to lose,
|
|
often resulting from conversation with other people, but that are
|
|
too far removed from what most people can relate to for me to just
|
|
publish them. Sometimes it's convenient to be able to share them,
|
|
though.
|
|
|
|
For that reason, if you stumble upon this file without me having
|
|
linked it to you intentionally, feel free to read it but keep the
|
|
sharing to a minimum (though do feel free to share the thoughts
|
|
themselves, of course).
|
|
|
|
-->
|
|
WARNING: This is not intended for a large audience. If you stumble
|
|
upon this page by chance, please keep the sharing to a minimum.
|
|
|
|
TIP: It's always work-in-progress. Things come and go. Or change. Who
|
|
knows?
|
|
|
|
---------
|
|
|
|
### Three things
|
|
|
|
*[mid/late 2020]*
|
|
|
|
All things in the universe take the shape of one of approximately
|
|
three things. If you had Hoogle for the entire universe, you'd
|
|
probably find that one of them is `fmap`.
|
|
|
|
There might be a few more, or a few less (or some may have been
|
|
deprecated), but you get the idea. I guess [five][] would be a good
|
|
number.
|
|
|
|
[five]: https://principiadiscordia.com/book/23.php
|
|
|
|
----------------------
|
|
|
|
### Free energy principle
|
|
|
|
*[mid/late 2020]*
|
|
|
|
Karl Friston wrote:
|
|
|
|
> The free-energy principle says that any self-organizing system that
|
|
> is at equilibrium with its environment must minimize its free
|
|
> energy.
|
|
|
|
Or, somewhat paraphrased:
|
|
|
|
> Any Markov blanket capable of modeling its environment aims to
|
|
> reduce its level of surprise by either adapting its model, or
|
|
> through other action.
|
|
|
|
Seems reasonable to me.
|
|
|
|
### More bizarre universe
|
|
|
|
*[many years ago]*
|
|
|
|
Douglas Adams wrote:
|
|
|
|
> There is a theory which states that if ever anyone discovers exactly
|
|
> what the Universe is for and why it is here, it will instantly
|
|
> disappear and be replaced by something even more bizarre and
|
|
> inexplicable. There is another theory which states that this has
|
|
> already happened.
|
|
|
|
### Alpha decay
|
|
|
|
*[late 2022]*
|
|
|
|
Finance people say:
|
|
|
|
> Alpha Decay is commonly referred to as the loss of prediction power
|
|
> of a trading strategy over time. As a consequence, the profitability
|
|
> of a strategy tends to gradually decrease. Given enough time, the
|
|
> strategy converges to having no superior predictive power and
|
|
> returns when compared to a suitable benchmark.
|
|
|
|
A market is a big optimiser. Any successful trading strategy adds
|
|
friction in a place that the optimiser wants to remove.
|
|
|
|
Alpha decay is unavoidable without changing and adapting the strategy.
|
|
|
|
### Optimising universe
|
|
|
|
*[late 2022]*
|
|
|
|
*(thanks edef for helping me think through this one!)*
|
|
|
|
Assume that the universe acts as a giant optimiser, and consider that
|
|
the three things above are related and specialisations of more generic
|
|
ideas:
|
|
|
|
1. Every delineable entity in the universe (i.e. every *Markov
|
|
blanket*) attempts to reduce its level of surprise (the free energy
|
|
principle).
|
|
|
|
2. The universe needs replacement (a more bizarre universe) if global
|
|
surprise drops to a minimum[^heat].
|
|
|
|
3. Without improvement that outpaces the optimiser of the universe,
|
|
any strategy leading to (2) will get eroded by alpha decay long
|
|
before.
|
|
|
|
4. We don't know if it is possible to outpace the optimiser from
|
|
within.
|
|
|
|
On a personal note, it seems to me that achieving (2) is likely
|
|
undesirable. It probably takes god[^god] a lot of resources to create
|
|
an ever more complex universe and this process might be much less
|
|
enjoyable than "running" (for lack of a better word) a universe. Under
|
|
this assumption, a universe that achieves (2) faster than others might
|
|
be a failure, and on a higher level conditions leading to its creation
|
|
might be subject to another optimiser.
|
|
|
|
Or it could be the other way around, but this seems more likely to me
|
|
personally.
|
|
|
|
### Superintelligence
|
|
|
|
*[late 2022]*
|
|
|
|
Under the previous assumption, achieving superintelligence is likely a
|
|
bad idea for anyone feeling some kind of attachment to *this*
|
|
universe.
|
|
|
|
Or it might be the exact opposite, but I don't think so.
|
|
|
|
-------------------------------
|
|
|
|
[^heat]: Note that this is consistent with the heat death of the
|
|
universe.
|
|
|
|
[^god]: I'm using the word "god" as the best English approximation of
|
|
a concept that different religions and philosophies all attempt to
|
|
approach. I think that for many cognitive purposes, an
|
|
anthropomorphised idea (as in the abrahamic religions) is useful,
|
|
but ideas from some Eastern religions or modern philosophers like
|
|
Bach or Watts are likely more aligned with the "nature of things"
|
|
as such.
|