Monday, January 7, 2019

Thoughts

If an embryo of two cells can be separated into two embryos, from which twins can develop, can this be continued more? So we get triplets, fourplets, fiveplets, etc.
And if that is so, would they have shorter lifespans due to cell division shortening the telomeres?

  • Race against the machine
  • Race with the machine
  • Race into the machine
  • Race of the machine
What would happen if sex organs are plug-and-play devices?
 - Well for starters it'd be pretty horrific if someone loses their vagina during pregnancy.
 - Maybe not so much. Hospitals would have emergency vaginas for that. Or just do cesarian.

  1. All models are wrong, but some are useful. ---- George E. P. Box
  2. Ethics are models of the world. ---- One possible stance of ethics 
  3. All ethics are wrong, but maybe some are useful. ---- Conclusion

Fermi estimate for intelligence:

Reading this, it makes me think. The chimps and humans differ in DNA only in 125 million base pairs out of 3000 million, that's a difference of 4%. 125 million base pairs is 63 megabytes. That gives an upper bound to complexity of intelligence: intelligence can be programmed in 64 MB... can it?
Some problems:

  1. Maybe humans just have a few more tricks compared to chimps, and most of intelligence relies on abilities that are shared with the chimps. For example, think back to "elephants don't play chess"
  2. Maybe the environment is very rich and a lot of complexity is gained by interacting with the environment. This is not a practical problem as long as we develop the AI on earth and make it interact with it, but if we ever launch AI into space, it may have to package the richness of the environment with it.

How optimized is the code? Well let's see... The genome of C. elegans has 97 million base pairs, giving 50 MB data, and OpenWorm project has 40 MB of code... so a comparable human project would simulate a human with about 1.2 GB of code???

Nah, this is nonsense.

Bayesian good luck

If someone keeps having good luck, maybe it's rational to assign higher probability to the hypothesis "they have good luck and will keep getting good events", if the hypothesis is simpler than the alternative of "they happen to get the lucky events and won't get them anymore later".

Integration of commodity derivative markets: Has it gone too far?
Well, if you need to use the Fundamental Theorem of Calculus in the name of the market, then I suppose it is too far!

Steampunk Venus Rover

This is inspiring. Automaton Rover for Extreme Environments (AREE). Finally, a mechanical robot with practical use! Strandbeest on Venus!

It will use passive communication: it use a simple optical reflector to transmit its data to orbiting satellites by reflecting radar in morse code. AREE is projected to transmit 1,000 bits per day, compared to Curiosity’s 1 million bits.

The danger of Humanism

Being human is addictive, people would do anything just to stay human.

The key to plausible deniability is to be a false negative by pretending to be a false positive.

No comments:

Post a Comment

Let's Read: Neuropath (Bakker, 2009)

Neuropath  (Bakker 2009) is a dramatic demonstration of the eliminative materialism worldview of the author R. Scott Bakker. It's very b...