Previously: 2011; 2012-2013; 2013-2014; 2014-2015

"When I was one-and-twenty / I heard a wise man say, / 'Give crowns and pounds and guineas / But not your heart away; / Give pearls away and rubies / But keep your fancy free.' / But I was one-and-twenty, / No use to talk to me."

My past year of completed writings, sorted by topic:

Genetics:

• Embryo selection for intelligence cost-benefit analysis
  • meta-analysis of intelligence GCTAs, limits set by measurement error, current polygenic scores, possible gains with current IVF procedures, the benefits of selection on multiple complex traits, the possible annual value in the USA of selection & value of larger GWASes, societal consequences of various embryo selection scenarios, embryo count versus polygenic scores as limiting factors, comparison with iterated embryo selection, limits to total gains from iterated embryo selection etc.
• Wikipedia article on Genome-wide complex trait analysis (GCTA)

AI:

• Computational Complexity vs the Singularity
• Adding metadata to an RNN for mimicking individual author style
• Armstrong’s AI control problem: Reinforce.js demo

Biology:

• vitamin D & metformin for life-extension: cost-benefit analysis

Statistics:

• Candy Japan new packaging decision analysis
• “The Power of Twins: Revisiting Student’s Scottish Milk Experiment Example”
• Genius Revisited: Critiquing the Value of High IQ Elementary Schools
• Inferring mean ethnic IQs from very high IQ samples like TIP/SMPY

Cryptography:

• shell scripts for easy cryptographic timestamping

Misc:

• Newton’s comet apocalypse system

gwern.net itself has remained largely stable (some CSS fixes and image size changes); I continue to use Patreon and send out my newsletters.

We've started a habit of creating periodic Bitcoin threads to confine discussion thereof to those threads and prevent excessive proliferation of Bitcoin topics in the discussion section.  Here is a link to the last one, which links the other discussions.  Lot's to talk about, and another bounce in Bitcoin's value (up to 33 then down to 24), so share your links and thoughts!

Annie Lowrey discusses Bitcoin in Slate.  No clear thesis, but important that it gets attention there.  She gives a general overview, with emphasis on its benefits to fringe elements on society, and gives quick attack at the end.  The attack seems misinformed, but it links to something more interesting, specifically...

A technical critique by Victor Grishchenko, PhD, who was mentioned here in the context of causal trees.  He describes a few problems he sees with Bitcoin:

1) Asymmetry favors attackers, in that it takes a lot more effort to check for double spending than to attempt a double-spend, eventually requiring "supernodes" that have disproportionate influence over the network.

2) It needs to continuously spend spend cycles to stay free from attackers.  He then describes an attack I don't quite understand that involves holding on to a discovered block and then broadcasting it at just the right time

3) It doesn't compare well against existing systems in terms of privacy, speed, or transaction cost.  (I found this questionable because the system he's comparing it to is still subject to warrants, and Bitcoin takes significantly less time -- 1 hour or so -- to ensure a transaction than the wiring transfers Grishchenko discribes.)

Finally, he credits Bitcoin in being advantageous similarly to Bittorrent: the latter was clumsy and complicated compared to regular downloading, but could perform well enough in a niche niche to force change in the broader markets.

BitCoin is a recently introduced currency, based on public-key cryptography combined with a peer-to-peer network for verifying transactions. I've been thinking a lot about BitCoin recently, and particularly about BitCoin's main weakness: if your computer is compromised, an attacker could copy your BitCoin wallet and use it to steal coins. That's bad. But I've come up with a possible improvement that would greatly mitigate this risk, and was hoping for some help confirming its viability and filling in the details.

The basic idea is to make it so that rather than having a single computer which can steal your coins if it's compromised, you have two computers (or a computer and a phone), such that your coins can only be spent if both devices cooperate. It is much harder to break into two computers belonging to the same person than just one, so this makes coins harder to steal. You could also have one of the computers involved be a third party that you trust to keep its files secure, and while that third party would be able to freeze your funds, it wouldn't be able to steal them. Using a third party this way, you could also add withdrawal rate limits and time delays, further improving security.

I believe that this can be done in a fully backwards-compatible way, without any changes to the BitCoin protocol, using homomorphic encryption. BitCoin is based on elliptic curve cryptography; a receiving address is a public key, and a wallet file is a collection of private keys. The goal is to create a protocol where two cooperating computers produce a split key, such that they can use it cooperatively to sign transactions later, but neither one can sign transactions or determine the whole key on its own. My understanding is that homomorphic encryption can be used to implement a simulated computer that does arbitrary trusted computation, so this should be possible. However, I'm a bit fuzzy on the details, and I don't have the time or comparative advantage to implement this myself.

(To deal with the risk of one one computer being lost or damaged, there could also be an override key; both computers would have the public half of the override key, and the private half would be kept offline in a bank deposit box or something similar. Then both computers use the override key to encrypt their halves of the split key, and send the encrypted keys to a cloud backup provider.)

Short version: Why can't cryptanalysis methods be carried over to science, which looks like a trivial problem by comparison, since nature doesn't intelligently remove patterns from our observations?  Or are these methods already carried over?

Long version: Okay, I was going to spell this all out with a lot of text, but it started ballooning, so I'm just going to put it in chart form.

Here is what I see as the mapping from cryptography to science (or epistemology in general).  I want to know what goes in the "???" spot, and why it hasn't been used for any natural phenomenon less complex than the most complex broken cipher.  (Sorry, couldn't figure out how to center it.)

EDIT: Removed "(cipher known)" requirement on 2nd- and 3rd-to-last rows because the scientific analog can be searching for either natural laws or constants.

Though not yet an "official" project, Google has released a Bitcoin client. As you may remember, there were concerns here about what the government/legal reaction to Bitcoin [1] will be, and the significance of certain groups lending their support to it.  EFF and SIAI accept Bitcoin donations, which helps, and this action by Google is another big step.

Previous articles: SIAI accepting Bitcoin donations, Discussion on making money with Bitcoin (Clippy warning on the latter)

[1] In short, it's an anonymous P2P crypto-currency with no transaction fees, in which new units are generated by spending computer cycles computing hashes until you find one with specific properties.