Eliezer

15 ms·

121.
by Eliezer 7y ago
Sometimes real life lacks a satisfying conclusion?
122.
by Eliezer 7y ago
Assets being priced proportionally to their currently-generated use-value isn’t how efficient markets work even in principle. If that were so, you could make predictable profits just by buying anything whose current adoption was currently
123.
by Eliezer 8y ago
I claim not that it is the whole effect size, but certainly the massive drop in testosterone must account for some of it. Would be interesting to see the state-by-state, year-by-year correlation.
124.
by Eliezer 8y ago
You are doing an obvious right thing to do and I wish you huge success.
125.
by Eliezer 8y ago
It seems disingenuous that this article fails to quote examples of GPT-2’s stunning results, or give any contrasting results from BERT to support the claim that this is all normal and expected progress. Like many, I was viscerally shocked t
126.
by Eliezer 8y ago
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3048809/ That was a shockingly low figure relative to what I’ve heard from friends in medicine. Looking up the original paper showed that the estimators used for t
127.
by Eliezer 8y ago
What if Apple buys a meth lab with it? You’ve just deployed a fully general argument against the existence of money.
128.
by Eliezer 8y ago
Failed? There are very few top theoretical physicists (though still some) who still believe in collapse, and the rest believe something morally equivalent to an Everettian theory.
129.
by Eliezer 8y ago
They say phishing is impossible and don’t immediately explain why? All kinds of alarm bells going off.
130.
by Eliezer 8y ago
Yes, still wrong if the police ran it. I think that in the end, the fate of the world is more likely to hang on how much people can trust each other, than on having even fewer assassins.
131.
by Eliezer 8y ago
You so rarely run into a news story that considers a genuinely difficult ethical question. I think I’d say that it’s morally wrong to run a scam hit darksite to steal money from people who want assassins, but it is not obvious. Especially
132.
by Eliezer 8y ago
And people say that Moore’s Law has broken down! Sure, processor speeds aren’t increasing as fast, but that’s more than made up for by the incredibly rapid increase in the number of Internet addresses per device.
133.
by Eliezer 8y ago
If interpreted that way, storing the noise and memorizing the offset, it amounts to having a privately stored one-time pad to be used as key... with a passphrase with as many bits in it as the offset. That’s probably not a lot of bits. You
134.
by Eliezer 8y ago
Don't roll your own crypto. If that was much better encryption, everyone would be using it.
135.
by Eliezer 8y ago
"What an unusually low Lizardman's Constant!" is my first reaction here.
136.
by Eliezer 8y ago
It's a root cause of bank irresponsibility. If not for free FDIC insurance, a bank making risky loans would cause the cost of insuring an account at that bank to rise. Securitization does change this, but all the CDSs on AIG's m
137.
by Eliezer 8y ago
FDIC insurance. A private insurance market for bank accounts sends a price signal about reliability. Subsidizing that insurance in a way that doesn't eliminate the price signal and the customer risk, like requiring 20% private insura
138.
by Eliezer 8y ago
"Already 7 inside a heptagon is much, much bigger than Graham’s number." Did I completely misunderstand that the polygon series is pretty much the Steinhaus-Moser polygons? Because Graham's Number is way out beyond what can
139.
by Eliezer 8y ago
Okay, but if you thought it was all your dream, why would you write up that belief in a presumed attempt to communicate the same idea to another figment of your imagination?
140.
by Eliezer 8y ago
Who did the people who wrote that up think they were talking to?
141.
by Eliezer 8y ago
I'd like to assure anyone reading this that the above commenter has no idea what Eliezer Yudkowsky thinks. I, with my considerably greater knowledge of that subject, can testify that among Yudkowsky's beliefs is "Bounded age
142.
by Eliezer 8y ago
Having read not more than a little amount of your take on my material, I suggest you look to your own reading comprehension skills.
143.
by Eliezer 8y ago
I'm not even sure I should bother replying to this. But, like many, I'd consider the scientific method properly applied to be a special case of probability theory, modulo some social rules that might technically depart from Bayes
144.
by Eliezer 8y ago
For a more detailed take advocating the particular solution "report likelihoods, not posteriors or p-values", see "Likelihoods, p-values, and the replication crisis": https://arbital.com/p/likelihoo
145.
by Eliezer 8y ago
I was on mobile so didn't have the time to post a link to more than "search on star lifting", but now that I'm off mobile, have a link. https://en.wikipedia.org/wiki/Star_lifting EDIT: more material
146.
by Eliezer 8y ago
To accelerate stars, you squeeze them until a jet of matter comes out in the opposing direction. For ideas on getting matter out of stars, see "star lifting". Furthermore, most of the useful energy in stars is not their nuclear bi
147.
by Eliezer 8y ago
The stock market disagrees with you.
148.
by Eliezer 8y ago
If I used the term, I would reply, "A safe powerful AGI is one with a less than fifty percent chance of killing more than one billion people." (Because most of the work is pushing the failure probability sigmoid from 0.999+ to bel
149.
by Eliezer 9y ago
The startup is not claiming to have current uploading technology. The startup is researching (does not claim to have) brain preservation techniques leading to long-term structure preservation in hopes of uploading sometime in the indefinit
150.
by Eliezer 9y ago
Gradient descent: Run the neural net, evaluate the fitness function, shift parameters in the direction of the derivative of the fitness function. Genetic algorithm: Shift parameters in a batch of random ways, evaluate using fitness functi

More ›