[FRIAM] by any means necessary
Steve Smith
sasmyth at swcp.com
Tue Feb 15 15:26:15 EST 2022
Thanks for having this conversation in front of us, I'm pretty invested
in these kinds of issues and they are rarely discussed openly IMO.
Perhaps you can unpack for me a little (or say it another way so I can
gain my own parallax):
/In our capitalist society, is it reasonable for Neuralink to be
less susceptible to the flattening you describe by aggregating (not
summing over) all subjects' projections from a high-dimensional
construct?
/
/
/
On 2/15/22 12:56 PM, glen wrote:
> Excellent! Thanks. However, it's also important to note that the
> lawsuit is against UC Davis, not Neuralink. So, to whatever extent
> that Neuralink funding, mixed with tax payer funding, drives
> university research (and possibly other things like overhead or paying
> a percentage of salary for some with teaching loads, etc.), those
> backseating costs can deeply impact whatever it is we call a research
> university.
>
> I'm about halfway into my "evaluation" of
> https://consilienceproject.org/. What I've seen so far has a healthy
> plating (I was going to say veneer, but that's too thin) of pretty
> words. But those pretty words sound a tiny bit like Neuralink's
> corporatized strawman/response to these accusations. I bring up
> Consilience because it's placed in between a for-profit company and a
> research university. On Consilience's About page, you see 2 ethical
> commitments:
>
> • collective attribution of authorship, and
> • transparency in methodology
>
> These may seem a bit contradictory to some observers. My guess is
> that, given some time and effort (maybe even semi-automated NLP
> computation), I could ferret out who wrote which featured article.
> What I'd like to be transparent is who contributes what to each
> article. (This is a professional task I have to some extent with my
> clients ... so it's not mere hobby.)
>
> Going back to the lawsuit against UC Davis and the 3 example spectrum
> (and perhaps even the political tangent SteveS raised), where does
> Neuralink end and UC Davis begin? In our capitalist society, is it
> reasonable for Neuralink to be less susceptible to the flattening you
> describe by aggregating (not summing over) all subjects' projections
> from a high-dimensional construct?
>
> We see a similar thread in the "academic free speech" rhetoric the
> alt-right is pushing these days (though there are lefty exceptions)
> ... aka when is an academic not talking as an academic? And in the
> Barret and Gorsuch exhortations that they're not partisan hacks ...
> even when talking at a partisan event.
>
> [sigh] I know these fluffy issues aren't interesting to most people.
> It's way easier to shut up and calculate. But not only are they
> interesting to me, I think they're necessary, then, now, and later.
>
> On 2/15/22 11:30, Marcus Daniels wrote:
>> For some activity there will be a mesh of consequences, that perhaps
>> with enough transparency, debate, and observation the facts of the
>> matter could be quantified as a large graph. Across this graph, one
>> could apply a subject's function of the utility of each one of those
>> consequences. If some of the consequences are both illegal and
>> observable and a node represented a risk to the subject doing the
>> assessment of the graph, then that node would probably result in a
>> negative utility for most subjects and perhaps it will overwhelm
>> other positive evaluations across other nodes. One could perform the
>> same procedure across all possible subjects. The sum would be a
>> social evaluation of the mesh of consequences. I think it would not
>> be very useful, and not even address externalized costs.
>> Throughout this procedure the subjects' utility functions would all
>> be subject to advertising, propaganda, religion, blood sugar and
>> hormones. Measure twice you could get different answer.
>>
>> If there are externalized costs that need to be recognized for the
>> survival of humans, then humans will have to create laws with large
>> risks for those that don't comply with them. (Case-by-case
>> harassment, vigilantism, or terrorism wouldn't scale as well.) My
>> guess in this Neuralink case, is that if there were any deviations
>> from best practices, they will be aware of this risk in the future.
>> In the cynical view of it being propaganda, well, yes, they'll be
>> motivated to make the best kind they can and to set things up to
>> compartmentalize the most sensitive or emotionally charged information.
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20220215/50acfb3f/attachment.html>
More information about the Friam
mailing list