Follow-up to Journal article about politics and mindkilling.  That post showed that people can be convinced that a view is correct by being told that their political party endorses it, even if their party actually opposes it.  A similar, but stranger, effect, is that people can be convinced that a view is correct because their favorite software implements it - even if they have stated that the view is wrong just minutes ago.

I think I’ve always had certain stereotypes in my mind about research. I imagine a cutting-edge workplace, maybe not using the newest gadgets because these things cost money, but at least using the newest ideas. I imagine staff of research institutions applying the scientific method to boost their own productivity, instead of taking for granted the way that things have always been done. Maybe those were the naive ideas of someone who had never actually worked in a research field. 

At the medical research institute where I work one day a week, I recently spent an entire seven-hour day going down a list of patient names, searching them on the hospital database, deciding whether they met the criteria for a study, and typing them into a colour-coded spreadsheet. The process had maybe six discrete steps, and all of them were purely mechanical. In seven hours, I screened about two hundred and fifty patients. I was paid $12.50 an hour to do this. It cost my employer 35 cents for each patient that I screened, and these patients haven't been visited, consented or included in any study. They're still only names on a spreadsheet. I’ve been told that I learn and work quickly, but I know I do this task inefficiently, because I’m not a simple computer program. I get bored. I make mistakes. Heaven forbid, I get distracted and start reading the nurses’ notes for fun because I find them interesting.

In 7 hours, I imagine that someone slightly above my skill level could write a simple program to do the same task. They wouldn’t screen any patients in those 7 hours, but once the program was finished, they could use it forever, or at least until the task changed and the program had to be modified. I don’t know how much it would cost the organization to employ a programmer; maybe it would cost more than just having me do it. I don’t know whether allowing that program to access the confidential database would be an issue. But it seems inefficient to pay human brains to do work that they’re bad at, that computers would be better at, even if those human brains belong to undergrad students who need the money badly enough not to complain.

One of the criteria I looked at when screening patients was whether they did their dialysis at a clinic in my hometown. They have to be driving distance, because my supervisor has to drive around the city and pick up blood samples to bring to our lab. I crossed out 30 names without even looking them up because I could see at a glance that they were a nearby city an hour’s drive away. How hard would it be to coordinate with the hospital in that city? Have the bloodwork analyzed there and the results emailed over? Maybe it would be non-trivially hard; I don’t know. I didn’t ask my supervisor because it isn’t my job to make management decisions. But medical research benefits everyone. A study with more patients produces data that’s statistically more valid, even if those patients live an hour’s drive away.

The office where I work is filled with paper. Floor-to-ceiling shelves hold endless binders full of source documents. Every email has to be printed and filed in a binder. Even the nurses’ notes and patient charts are printed off the database. It’s a legal requirement. The result is that we have two copies of everything, one online and one on paper, consuming trees. Running a computer consumes fossil fuels, of course. I don’t know for sure which is more efficient, paper or digital, but I do know that both is inefficient. I did ask my supervisor about this, and apparently it’s because digital records could be lost or deleted. How much would it take to make them durable enough?

Summary: Rationalists should win; however, it could take a really long time before a technological singularity or uploading provide powerful technology to aid rationalists in achieving their goals. It's possible today to create assistant computer software to help direct human effort and provide "hints" for clearer thinking. We should catalog such software when it exists and create it when it doesn't.

The Problem
We may be waiting awhile for a Friendly AI or similar “world changing” technology to appear. While technology continues to improve, the process of creating a Friendly AI seems extremely tricky, and there’s no solid ETA on the program. Uploading is still years to decades away. In the meantime, aspiring rationalists still have to get on with our lives.

Rationality is hard. Merely knowing about a bias is often not enough to overcome it. Even in cases where the steps to act rationally are known, the algorithm required may be more than can be done manually, or may require information which itself is not immediately at hand. However, a lot of things that are difficult become easier when you have the right tools. Could there be tools that supplement the effort involved in making a good decision? I suspect that this is the case, and will give several examples of programs that the community could work to create -- computer software to help you win. Because a lot of software is specifically created to address problems as they come up, it would also be worthwhile to maintain an index of already available software with special usefulness and applicability to Less Wrong readers.

Stigmergy is a notion that an agent's behavior is sometimes best understood as coordinated by the agent's environment. In particular, social insects build nests, which have a recognizable standard pattern (different patterns for different species). Does the wasp or termite have an idea of what the standard pattern is? Probably not. Instead, the computation inside the insect is a stateless stimulus/response rule set. The partially-constructed nest catalyzes the next construction step.

An unintelligent "insect" clambering energetically around a convoluted "nest", with the insect's local perceptions driving its local modifications is recognizably something like a Turing machine. The system as a whole can be more intelligent than either the (stateless) insect or the (passive) nest. The important computation is the interactions between the agent and the environment.

This is an account of downloading and using SMILE, a free-as-in-beer-but-not-open-source bayes net library. SMILE powers GENIE, a graphical bayes net tool. SMILE can do a lot of things, but I only used the simplest features - building a network and, given evidence, inferring probability distributions on the unobserved features.

In the post Outlawing Anthropics, there was a brief and intriguing scrap of reasoning, which used the principle of reflective inconsistency, which so far as I know is unique to this community:

If your current system cares about yourself and your future, but doesn't care about very similar xerox-siblings, then you will tend to self-modify to have future copies of yourself care about each other, as this maximizes your expectation of pleasant experience over future selves.

This post expands upon and attempts to formalize that reasoning, in hopes of developing a logical framework for reasoning about reflective inconsistency.

In reply to: Community Epistemic Practice

There are software tools, possibly helpful for community truth-seeking. For example, truthmapping.com is described very well here. Also, debategraph.org, and I'm sure there are others.