[FRIAM] The epiphenomenality relation
uǝlƃ ☤>$
gepropella at gmail.com
Mon Nov 29 15:10:29 EST 2021
I fixed the larding for you. 8^D
Yes, I would accept "played little role in". That's what I've been arguing for the entire time ... a *scoping* or scalar measure. When you say "epiphenomenon", you imply a clear cut, obvious category. When you say "played little role in", you're making it obvious that there is a *measure* in there somewhere. What measure are you using that shows the "little role"? Can we change the measure so that what seems like "little role" by the first measure shows "lots of role" with the next measure? Etc. Such measure-variation is necessary to the parallax method you mentioned using the words "flashlight" and "across points of view".
Ranked choice voting is just a human-in-the-loop example of dimension reduction. And reduction usually involves abstraction, the ignoring of detail. So, if, during your exploration of the object using multiple points of view (parallax), you *discover* that all the possible points of view show effect E to be of small primacy, then sure ... reduce/abstract/elide it away. But if E shows small primacy under one flashlight and large primacy under another, then you want to keep it for awhile at least.
Ideally, some subset of all effects will show small primacy under *all* feasible flashllights. Then you get dimension reduction that limits the amount of information you throw away.
But the above depends, even further back in our assumptions/workflow, on a stable set of distinguishable effects. You can't really talk about the primacy of effect E1 versus E2 without slicing the object into 2 disjoint effects (E1 vs E2). What if the distribution of effects is kindasorta smooth? And there is no clear line between E1 and E2? Then, we have an even more primitive analysis to execute where we parallax our binning methods ... before we even engage in the process of reducing the number of bins. Binning method B1 might produce 10 effects, where binning method B2 produces 2. And b3 produces 72. What does that mean? Etc.
On 11/29/21 11:43 AM, thompnickson2 at gmail.com wrote:
> On 11/29/21 10:51 AM, uǝlƃ ☤>$ wrote:
>> My only beef in this conversation is the implicit intelligence, intention, and purpose in Nick's incessant use of "epiphenomenon". The 1st and 2nd paragraphs of Nick's trolling text were well composed, nothing with which to disagree. It was the 3rd that triggered me (making me wonder if I understood the 1st 2). Therein, he said "We will argue for a definition of an epiphenomenon as a consequence of a structure’s (or behavior’s) design which has played no
> */[NST===>Ach! I always get in trouble for overstating my cases: whistling in the dark, it is. Obviously, every cause feeds back on its effects to SOME degree. So, would you accept “played little role in”? If so, my apologies. <===nst] /*
>> part the development of that structure (or behavior)." Every consequence plays a part in the development of the structure. That's, to some extent, what we *mean* by "consequence".
> */[NST===>I like the idea that we limite the idea of consequence in this way, but I don’t think it will fly out there in the world. <===nst] /*
>> I'd be fine if, for example, our chosen measure was that some outcomes (and their generating machinery) dissipate through the iterations. Some dissipate faster than others. We could talk about the half-life of a given consequential pattern. And if the half-life was short, we could call that a less primary effect than those with large half-lives, perhaps even so drastically less as to call it "ancillary" or some other such word.
> */[NST===>I totally agree with all of this, and that makes me think that many of my disagreements with Glen arise from my rhetoric, my asserting greater confidence than I actually feel. As I say, Whistling in the Dark. <===nst] /*
> */[NST===>Do we need ranked choice voting to determine which effects are primary? <===nst] /*
--
"Better to be slapped with the truth than kissed with a lie."
☤>$ uǝlƃ
More information about the Friam
mailing list