3,102 karma · joined August 26, 2016
However, it's still better to recognise a problem, so you can at least look into ways of improving the situation.
(Replying to the comment that you made before completely rewriting it.)
You're completely wrong about the sugar pill (or at least whether or not it matters) - it's one of the issues with psychedelic medicines: there's no viable placebo at higher doses, and it's the higher doses that seem to have the deepest and longest lasting impact on depression.
I know, from repeated personal experience, that psychedelics (or mushrooms at least) can be highly effective against even severe depression.
They have been the single most effective anti-depressant that I have ever taken.
Now, I haven't had the depression curing wonder of an experience that some people claim;
It never pulled me out right out of it - but it pulled me back from suicide (on multiple occasions) and a single (rather large) dose is effective for 3 or 4 months before I find myself spiraling back down.
And if we're gonna have a conversation about anti-depressants that are being used despite scant evidence in their favour, perhaps we should start with SSRIs - which not only made my depression worse, but also left me almost asexual for ~20 years.
It's to illustrate the hypocrisy.
Ranked choice allows for a middle ground to form independently of the parties at the outer fringes - it doesn't necessarily mean the centre wins, but it does exist and have a chance.
Burnout's more insidious and pernicious than that - and it takes far more than a few days of rest to get over it.
And you're right, it would eventually go away - but only if and when late adopters find themselves being punished by the market.
Often by the end of the week, people in complex positions are well past the point of diminishing returns - pushing them harder may work in the short-term, but it's a recipe for burnout.
It's better to have someone operating closer to their peek performance for longer - even if you sacrifice a day overall.
Those guys had some ridiculous job titles.
But the most fundamental boon that it offered was in terms of planning and organisation. Before language we'd point and grunt and go there and do the thing that we were gesturing that we were going to do.
But that's a very crude form of planning - you're pretty much just going all in on Leeroy Jenkins.
But actually (and horrifically) I think it's the gift of Kane that speaking and planning permitted; well organised humans (the smartest things on the planet) have been figuring out increasingly better ways to both kill each other and not to die themselves in a brutal feedback loop for a very long time now.
It's brutal as fuck, but it's Darwinian gold.
Perhaps, but the relative success of trained LLMs acting with apparent generalised understanding may indicate that it is simply the interface that is really an LLM post training;
That the deeper into the network you go (the further from the linguistic context), the less things become about words and linguist structure specifically and the more it becomes about things and relations in general.
(This also means that multiple interfaces can be integrated, sometimes making translation possible, e.g.: image <=> tree<string>)
Beyond the point that custom fields are pretty much always advanced, there has to be a point at which a term is so fundamentally non-specific that it cannot be subject to trademark.
I just think it's bloody madness sometimes.
Intel was crossing many lines in terms of the anti-competitive practices they were willing to engage in, of which hobbling their own internal development was probably the least damaging to the market overall.
The internal communications of the mind have no need for such constraints (and evolved hundreds of millions of years beforehand).
Anyway, I don't know what you were actually trying to argue here: you just built a simulated brain out of people, and the massively multi-agent distributed nature of the language of that machine is (emergently) incomparable with vocalised language.
Language is designed to be expressible with low fidelity vibrating strings - it is very clear that the available bandwidth is in the order of bytes per second.
Verses a fucking neural network with ~100 billion neurons.
Come on man, seriously - the two communication modalities are completely incomparable.
Language is a very poor substitute for freely flowing electrical information - it is evolved to compensate for the bottlenecks to external communication - bottlenecks that are lacking an internal analogue.
It's also a highly advanced feature - something as heavily optiimised as evolved life would not allow something as vital as cognition to be hampered by a lack of means for high fidelity external expression.
I don't think that there's spooky action at a distance (it's fundamentally equivalent to retrocausality, and the consequences of the distant foreign event cannot outpace its light cone anyway).
I think its a superposition of states of a closed time-like curve thing being fleshed out as its contradictions are resolved and interactions are permitted between its colocated non-contradictory aspects.
But I'm not a physicist, so that's probably all just bullshit anyway.
Laziness isn't necessarily a bad thing, depending on how it's implemented.
Don't consider this from the perspective of a reasonable person - ask yourself: what would psychopaths do?
It was a deeply cynical way of doing an end-run around the GPL, and I've held them in utter contempt ever since.
It converts a regular rent payment into a one time fee, and scares the crap out of anyone thinking to do something similar.
I'm guessing you'd need to solder on your debugging tools (oscilloscopes?) in parallel, since anything else would completely invalidate anything that you were trying to do.
It honestly sounds like a recipe for Heisenbugs.
Unlike the -e solution, it wouldn't make the command significantly more verbose than the default option.
You could also print a list of commands for the specific options (or allow for index based selection) in cases where there were collisions.
Defaulting to the only result in your library is perfectly reasonable, but it the case of a collision, what does it do?
My guess is that it runs through in a loop (I haven't looked at the code) and simply goes with the first result that it finds, but that doesn't cover all possible use cases.
(I often consider it shorthand for "I'm not an American".)
Much like he seemed to have missed the point of the post, you seem to have missed my point about him seeming to miss the point of the post.