Information domains

This post is intended as a half-baked idea which we will discuss at the local Less Wrong meetup and either discard or improve. In particular, at the last Singularity discussion we seemed to feel it would be useful to come up with a non-confused notion of a meme, and this might start off some kind of a sequence with that goal.

It’s hard for information to cross between different domains:

  • genetic information
  • knowledge in human brains
  • human written knowledge
  • human-readable computer information
  • machine-readable computer information

This is due to differences in storage medium, bandwidth between systems and differences in how information is encoded/interpreted. Something which we might regard as a “paradigm shift” occurs when either a new information domain is created (presumably with some advantage over previous ones) or when a partition is removed and information can flow between previously separate domains.

This idea is supposed to be obvious, so expect nothing profound here. I hope that by stringing together obvious ideas, we can end up with a non-obvious one.

Memory systems

Whenever a system can be in more than one state, we might regard it as containing information. But we’re interested in something slightly less general than that – something I’ll call memory systems. Memory systems are made up of a largish number of memory units, each of which can be in a few different states. The large-scale structure of a memory system is therefore fairly uniform, but the memory system has 2^largish possible states, because each memory unit can be twiddled independently.

Examples of memory systems:

  • A book
  • The DNA inside the nucleus of a single cell
  • An SDRAM chip on a computer
  • A hard disk
  • A human brain

The last example – brains – doesn’t fit the pattern very well. The state of a “memory unit” must include any memory that individual neurons have (which I know nothing about, so if there’s anything going on here then someone please fill me in). But it must also include connections between neurons. So the system seems to be at least partly analog and it’s not very obvious what constitutes a “memory unit” in this case. So I don’t know how much special treatment we’ll need to give to brains here.

The reason these memory systems emerge is because they’re really useful. This implies that there’s another system – a memory user – which makes use of the memory. And it must make good use of it because memory systems are expensive. Thermodynamically, they’re somewhat low entropy so you need to use up some of your precious negentropy if you want to create or maintain one. The memory user must be getting enough benefit from the memory system to make it worth that cost. The information has to pay its rent.

The other obvious point here is that the memory user provides an interpretation of the information. Stick your genetic code in a book, or a Shakespeare play into the DNA in one of your cells, and nothing terribly interesting is going to happen. The memory users interpret the information completely differently. And there’s no shallow conversion process either – in these examples the information means entirely different kinds of thing.

Sharing information

Memory systems differ in their ability and willingness to accept information from other memory systems.

Sometimes they’re completely read-only, and in this case it can be difficult drawing boundaries around memory systems (is your copy of Gödel Escher Bach the same “system” as mine, or a separate one? The question doesn’t seem entirely meaningful). So maybe we need some new notions in order to reason sensibly about read-only memory. (In the read/write case it’s pretty obvious. My brain is a separate system from your brain because they exchange information far more slowly than information can be copied about internally).

The case of computers suggests it’s a bit more complicated than that though. Each storage device on my computer should probably be regarded as a separate memory system. But the entire computer, and all computers on the Internet should maybe be regarded as larger systems. The horde of information interpreters on the Internet will each provide a similar interpretation of the same information. And data can be copied between memory systems relatively quickly. So the Internet is a candidate for an information domain.

Should we extend that a bit, to include all computers including those not on networks or on separate networks? I think… maybe. If we’re interested in huge economic paradigm shifts, we’re interested in the space of things which could happen at any point in time, not the things which are actually happening. So two memory systems which are capable of exchanging information and which have compatible interpretations should probably be regarded as belonging to the same domain, even if there are physical barriers.

The notion of an information domain isn’t 100% robust here. Clearly information does flow between different domains – information is flowing from your computer to your brain as you read this. But it seems robust enough that we build our paradigms of how the world works around the fact that information flows within one of these domains far more easily than it jumps between them.

Paradigm shifts

Creating a new type of memory system (e.g. solid-state drives) doesn’t appear to be enough to really count as a paradigm shift. It’ll just make existing stuff happen more efficiently. On the interpretation side, coming up with a shallowly different interpretation (e.g. EBCDIC vs. ASCII) doesn’t appear to be enough either. If you could come up with an entirely new kind of thing that information could mean then it might count – but (almost by definition) it’s kind of hard to imagine what such an event would look like. In the past, I think such events have been associated with the creation of new information domains also, but I’d be interested in possible counterexamples. The problem is that my idea of what “interpretation” means is somewhat vague at this point (and probably deserving of a separate writeup).

So currently I’m thinking that (information-related) paradigm shifts are associated with the creation of new information domains, or the merging of existing ones.

The origin of life is associated with the creation of the first information domain (although I think it’s still an open question as to whether there were evolution-capable systems that didn’t contain anything like a genetic code, and if so how long they lasted for. E.g. bags of miscellaneous chemicals  catalysing each other’s reactions, with many stable equilibrium points).

Brains were a new information domain too. Human written knowledge might have been (printing almost certainly isn’t). Computers might have been, although I’m not sure to what extent they can be regarded as an extension of human written knowledge. We haven’t really seen any domains merge yet (or have computers achieved this somewhat?), but current and near-future advances in technology may cause it to happen. Bioinformatics might start to merge the genetic and computer domains. Neuroinformatics might start to merge the brain and computer domains. Narrow(ish) AI might start to merge the human-readable and machine-readable domains.

I think the possible consequences of such an event happening are more controversial than whether it will happen.

Summary

  • Properties a memory system:
    • Digital or analog? Ordered or messy? (Brains stand out as the exception here)
    • Read-only or read/write (Definition of a “memory system” is vague in read-only case)
    • What sort of memory interpreter is associated with the memory system
    • Read/write bandwidth
    • Connectedness to other memory devices
  • Aspects of interpretation:
    • My concept of “interpretation” is still somewhat vague. (Worth considering both procedural and declarative interpretations)
    • Each memory systems will have one or more interpreters. One interpreter may handle multiple memory systems.
    • Memory devices are expensive, so information must pay its rent. We expect to encounter information which is “meaningful” according to the relevant interpretation
    • Connected memory devices are assumed to permit compatible interpretations, otherwise exchanging information would be pointless
    • (I also need to describe conversion between different interpretations)
  • Classes of memory system are partitioned into different information domains because of:
    • Radically different interpretations of information
    • Lack of ability to copy information between domains (e.g. even if I had a sensible genetic code to hand, I couldn’t easily reprogram my own DNA). (Does this relate to read-onlyness?)
    • (Concept is not completely robust because information still flows between domains, just more slowly)
  • Not obvious whether barriers which are conceptually easy to overcome should count as partitions between information domains
    • Low bandwidth or lack of connections
    • Deliberate security policy
    • e.g. computer disconnected from Internet
    • e.g. two separate human brains (actually I’m not sure this one is conceptually easy to overcome)
  • Information domains include:
    • genetic information
    • knowledge in human brains (should each brain be regarded as a separate domain?).
    • human written knowledge
    • machine-readable computer information
    • human-readable computer information (maybe this should be lumped together with either “written knowledge” or “machine-readable” or all three lumped together).
  • Paradigm shifts:
    • The creation of each of these domains was associated with a paradigm shift
    • Merging two of these domains would be a paradigm shift too. (I’d be interested in whether anyone thinks this has happened in the past)
    • It’s at least somewhat plausible that one or more of these merge events is happening now and will complete on a <century timescale
    • If we expect a paradigm shift, we shouldn’t necessarily trust models that worked well in the previous paradigm. They are probably over-fitted.
    • What are the most important paradigm shifts that aren’t covered here? (Robin Hanson thinks agriculture is important)

Wikipedia pages which I probably should have read:

Advertisements

0 Responses to “Information domains”



  1. Leave a Comment

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s





%d bloggers like this: