Hacker News Reader

Comment by Anon84 | Hacker News Reader

> a mechanical light switch should have different weight when switched on or off.

Well, in principle it should have a different weight to account for the extra electrons moving around when it’s turned on. Every time you have a change in charge you have a corresponding mas change.

My problem with this idea is the fact that not all information is electronic in nature, in which case you would be arguing that the weight of a completed puzzle is different than the total weight of the individual pieces, for example.

In general, this would imply that changes in entropy would correspond to changes in mass...

View on HN

It's more complicated than that. A single light switch stores exactly 1 bit of information. Its either on or off and both positions convey the exact same amount of information.

Same as an SSD. Regardless of whether you saved anything on it the individual bits are either set or unset. It always has information on it, it just might not be _your_ information.

This raises what I view to be a falsification by counter-example of the original premise. If information (and not _the representation of_ information) has mass, then does a piece of uncompressed information weight the same amount as the compressed information?

Furthermore, information is related to interpretation. A string of bits may be an uncompressed text file, a set of numbers, a compressed image, etc. But it's always the same amount of bits. How can my interpretation of that set of bits affect its mass?

There are measures that do not depend on interpretation. Entropy of the content for instance. This is what people usually refer to in this context and the ideas of the article have actually been known to physicists for decades.

I'm out of my depth here as I've never studied information theory. I suppose there's a larger gap between the theory and the practical aspect of the title of this article than I first imagined!

Yeah, our everyday technology is maaaany orders of magnitude away from where this would be detectable. However, it is still important topic if we want to design extreme efficiency computing hardware (and already studied among researchers that want to make nanotech).

If the mass of information is strictly proportional to how many bits of entropy it contains, it follows that neither interpretation nor (lossless) compression can alter its mass.

There's quite a bit of difference between how we usually think about information and how information theory treats it. There it's all about entropy.

Doesn't the very definition of "lossless" depend on interpretation? If lossy compression alters its mass, but lossless doesn't, doesn't it depend on the interpretation of "lost" information?

As I said to the above poster I don't know much about this, so hope this isn't an off-base question.

Lossless compression preserves entropy. Lossy doesn't.

The meaning of a certain sequence of bits depends on interpretation, of course. But nobody is talking about meaning here, only entropy. For our purpose it wouldn't matter if there were no sentient beings in the universe to attribute any meaning to anything.

Yes, you summarized what I feel is wrong with the experiment perhaps even better than me. They are trying to weigh not the information itself, but medium used to convey the information.

It that true, though? Does an electronic storage device with full storage have more electrons in it? I don't know enough about how modern memory works to say for sure.

That's not really the point. The point is, they would be measuring the weight of the electrons, in other words, the implementation-specific medium used to store the information. Not the information itself.

I can see this discussion go very far into metaphysical. :)

Well, what is information? Is it an abstract thing, or does it only exist in terms of the thing which is representing it?

I don't have a clue, I just think it's a relevant question at this point.

Electrons are used as the information storage mechanism in flash memory, but the data is whitened (50 % one bits on average independent of the data input) before being stored. Hence, "the number of electrons" would not vary between an SSD storing only zeroes, and one only storing ones.

But in any case, you would be measuring the mass of electrons and not the information they represent. This theory suggests that a storage device would have additional mass dependent on information content, so things like whitening / encryption would not matter.

However, here is a question that immediately occurs: If I have a message with "informational mass" x, and encrypt it, does the encrypted message has the same informational mass, or is the mass of the encrypted message the same as a random message of the same length?

Since flash is based on a charge pump, doesn't it just move the electrons from the substrate or control gate to the floating gate? Or am I missing something here?

In that case you do not have to account for the electron mass at all.

Electron rest mass is about 1E-31 Kg which is orders of magnitude more than the proposed estimated bit mass of 1E-38 Kg.

Surely not though - we're talking about information.

You don't need electricity flowing for the light switch to store information. In fact, the light switch could have no spring in it, allowing it to be in an infinity of different states.

So let's imagine a board with two such light switches. I set one to 0.9998128376666.. and the other to 0.26757328888... - out of the infinite number of possibilities, how do we quantify the information that I've just stored in these light switches, to determine how much their weight should have changed?

It's a silly question. It's a bit like saying "If a tree falls in the forest, do birds contain sodium?" Of course the two halves of the question are related tangentially, but putting them together like this is totally meaningless. Oooh... and nests are a thing, so, is it trees falling in the forest that causes nests? (dark matter)

Reply on news.ycombinator.com