From: Olie Lamb (email@example.com)
Date: Thu Aug 31 2006 - 03:08:05 MDT
On attempting to reply to this, I hit "Report Spam".
May I cite the fact that I passed primary school as evidence of my
non-illiteracy and be forgiven my momentary mistake in thinking that
all words starting with "re" mean "reply"?
* rapidly gains crackpot points *
On 8/31/06, Eliezer S. Yudkowsky <firstname.lastname@example.org> wrote:
> At the end of August, i.e., at the end of tomorrow, this topic is
> officially dead and killthreaded, so if you've got something you
> absolutely need to get off your chest, get it off your chest before then.
> My guess was that Loosemore was a crackpot, or, even if not a crackpot,
> someone who was rather unlikely to contribute usefully to SL4. I acted
Eliezer, I've always worried about your apparent eagerness to apply
dichotomies. The world is not as Boolean as some people seem to
Crazy people come up with good ideas. Often.
Although reasonably-smart people tend to have fewer personality flaws
etc than the average, more extreme mental abilities _often_ involve
increasing divergences from normal, sanish behaviour. I'm not going
to try to make a statistical case going this way, I'll work from the
I've been familiar with a lot of inventor-types. They seem to have a
characteristic set of behaviours...
Great with tinkering. Love to take things apart and put them back
together again. Fabulous with a lot of mechanical issues. Invariably
Generally, these people /aren't/ great with academia, and don't get
into engineering degrees, thus excluding them from utilising their
skills appropriately. Their research abilities often suck.
Each of them will, at some stage, have a stab at a number of classic
crackpot problems - like trying to build a perpetual motion machine.
If they're lucky, they will have enough engineering knowledge to
realise what they're doing and stop. If not, they will keep trying
and trying - obsessing.
Often after pouring huge amounts of effort into a project, they will
lie to protect it.
Now, having worked in Automotive technology for a bit, I am familiar
with the ideas that get sent to them... there are a huge number of
ideas for (1) fuel-saving and (2) power-increasing gadgets/engine
Many of these are based on alternative engine designs (Scotch Yoke,
radial, rotary) or different fuel mixes.
It is typical for a radically different engine design to be
demonstrably better than the best known endeavours into the field by
Qualified Engineers, although because Qualified Engineers gave up on
that approach 50+ years ago, their prototype is still nowhere near as
good as a conventional engine.
Now, another thing about inventors is that they are worried about IP
protection, and they can't afford patents. So they will ALWAYS try to
describe their system as "The problem with the other approach is that
Now, with an inventor, they have the option of creating a working
prototype. This distinguishes the talented crackpots from the
incompetent ones. But, due to the problems mentioned above, this is
generally only a "proof of concept", and generally doesn't deliver
tangible benefits over the conventional models.
At this stage, if they're lucky, an inventor might get someone else to
listen to them, and they might be able to build a /real/ prototype.
And here comes a tragedy of success: After a long period of failing,
they finally have some real success. There's proof in others'
beliefs: the invention works! The orthodoxy was mistaken! And so
begins their ego growth (UNDER NO CIRCUMSTANCES LET AN INVENTOR AT
THIS STAGE BE IN CHARGE OF ANY MONEY!!!) And so the inventor brings
out their other ideas - their old perpetual motion machine, the other
inventions, and tries to foist them onto their new benefactor.
If the benefactor has any sense, they'll say: Shut Up, get back to the
thing I was interested in. If not, the operation collapses.
If this goes well, then comes the real testing etc. In Automotive,
this is very expensive.
...But the conventional engine has been very extensively refined, and
a lot of things are fine-tuned to work with each other. Even a good
new concept will only make a very small improvement... So the inventor
may claim that the testing is unfair to the product, and may have
legitimate reasons for why the device doesn't show up as well as it
Furthermore, a lot of the testing relies on statistical
interpretations, and as a rule, inventors do not understand statistics
at all... so the inventor may recklessly misinterpret the data
Furthermore, the established manufacturing companies care a hell of a
lot more about reliability and durability than performance under
To any of these, the inventor may try to go their own way, get
paranoid - and given their inflated-by-success ego, it can get very
ugly. The path from idea to incorporation in mass production is long,
expensive, and arduous. There has to be a shortcut!
So even with a genuinely, demonstrably beneficial product, there is a
strong motivation for the inventor to "protect" their investment by
lying or misleading, which can damage even the best ideas.
I can think of several decent inventions that got to the stage of
aftermarket release, only to have the crackpot inventor collapse and
bankrupt the company - often through misleading with the test data.
However - and this is the key bit - just because the company collapses
doesn't mean the idea was a dud. With any luck, the idea can be
picked up and developed by Real Engineers, not crackpot Inventors.
Even if someone /is/ a crackpot, their crackpot theories might still
have something to them.
Yes, I know that some of us don't have the time to sort through the
shit for the gold-ore. (Note the "ore").
With Automotive inventors, we can say "build a prototype", which sorts
the talented crackpots from the untalented ones. Unfortunately, with
AI, there's no easy mechanism. Readability at least makes some
crackpots more bearable than others (I quickly stopped bothering with
Marc's posts not because of content, but because they are too damn
awkward to read)
But, seriously, didn't anyone else think about Loosemore: "This guy's
a nutbag" when he started talking about PARAPSYCHOLOGY?
Yeah, he might have legit reasons for thinking that the
school-of-thought is unjustly denigrated. Give a fuck. He supported
something that the establishment lumps together with the toothfairy.
Why not ban him then? Because we're so incredibly smart and wise that
we know that smart people get stupid ideas. But when Loosemore *shock
horror* does other crackpot things, he gets banninated.
Quite frankly, I don't care about any of the other things. If he, or
any other crackpot, can put together reasonable, coherent and
_readable_ theories, that can generally meet the standard criteria for
scientificness*, then all their crackpottiness does is put my bullshit
detector on high alert.
*Falsifiability is a good starting point, but not always necessary (eg: Maths)
I damn well expect Loosemore to meet the challenge to give literature
references (even if he has to write them) to this:
>This impression would be a mistake. To take just the issue of
>friendliness, for example: there are approaches to this problem
>that are powerful and viable, but because the list core does not
>agree with them, you might think that they are not feasible, or
>outright dangerous and irresponsible. This impression is a result
>of skewed opinions here, not necessarily a reflection of the actual
>status of those approaches.
I also expect him to elaborate on WHAT tools are required for his approach.
When (if) he does so, I will forgive him any beliefs he might have in
the toothfairy or perpetual motion.
If others interested in AI can't do the same, I think that's a pity.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:57 MDT