[hounddogjr]

hounddogjr asked the Naked Scientists:
   
If I was to down-load 500 Gigabytes of information onto a computer, would that computer weigh more than if the computers hard drive was blank?

Does information weigh anything?

Keep up with a great programme.

Oney Pantoja
Phoenix Arizona

What do you think?

[peppercorn]

No.
The data is encoded by changing tiny magnetic regions.
Nothing material is added and mass unaffected by magnetism.

If anything, a wiped blank disk with all '0's (or '1's, it doesn't matter) would have a very slightly more polarised magnetic field overall (I think we can assume that enough data as a whole would have equal 1's to 0's).
Placing the blanked platter on top of a ferrous metal should (I think) pull it infinitesimally downwards compared with a 'full' disk - strange huh!?

[JP]

I agree with you for the most part, peppercorn.

Of course, if he's trying to get at weighing more through E=mc², you could argue that it would weigh an minute bit more, since you've put energy into it to store that data.  However, the energy is small and c is huge, so the increase in mass is minuscule, so you'd never really notice it.

[imatfaal]

PCorn is of course right but also agree with JP - increased information, increased energy, increased mass.  Actual amount completely negligible.

[Geezer]

Ah, but there was already information there. All that happened is different information was substituted, so there is no net change in information and therefore no change in mass.

[JP]

Someone always has to be a wiseguy, eh Geezer?

Good point.  It also depends on how the information gets stored--whether the storage mechanism takes energy or not.  If it's a magnetic storage, then I think there is generally some energy gain on storing information as opposed to nothing (you have to line up magnetic domains, which takes energy).  You're right, of course, that writing new information on top of old information (which is generally what you do) won't change much of the energy content, since things were probably already lined up.

But even writing new energy over old takes an energy expenditure which gets dissipated as heat.  So if you didn't let your laptop cool down, storing data would make it weigh more. 

But all of this is just me nitpicking, since this energy gain is too tiny to matter anyway. 

[Geezer]

Well, it depends  []

Even if a disk never had any user data on it, it's still formatted with "information". Highly repetitive information admittedly, but information nonetheless. Depending on the encoding scheme, random data might actually take less energy to encode than all ones or all zeros.

BTW, my middle initials really are W and G.

[imatfaal]

I will dig out the reference to this but Leonard Susskind is quite insistent on two occasions that I have seen that to take a computer hard drive full of information and return it to the initial "unwritten" form will, through change in entropy (ie apart from magnetic, electric, etc  considerations), require a release of heat (incredibly small amounts) from the computer.  clearly if this is the case (and I realise I have not convinced) then the opposite must apply - that to put information on requires an increase in information and an increase in energy thus mass.     Basically the complication and non-randomness on the hard disc when it is written must be maintained, as information is always conserved - that information can only be preserved by a change in the environment. 

[Geezer]

I will dig out the reference to this but Leonard Susskind is quite insistent on two occasions that I have seen that to take a computer hard drive full of information and return it to the initial "unwritten" form will, through change in entropy (ie apart from magnetic, electric, etc  considerations), require a release of heat (incredibly small amounts) from the computer.  clearly if this is the case (and I realise I have not convinced) then the opposite must apply - that to put information on requires an increase in information and an increase in energy thus mass.     Basically the complication and non-randomness on the hard disc when it is written must be maintained, as information is always conserved - that information can only be preserved by a change in the environment. 

I suspect the catch here is the definition of "information". When the disk was first formatted, I could say information was encoded on it. It may be a continuous string of zeros, but it's still information. If that's true, when we write different information on to the disk, the net change in entropy is zero.

It would be interesting to see what Susskind had to say about it. I'd be highly amused if I was able to poke a hole in his argument! 

[RD]

... It may be a continuous string of zeros, but it's still information. If that's true, when we write different information on to the disk, the net change in entropy is zero.

This could get confusing as there is a computing version of the word "entropy" ...

Entropy is a measure of disorder, or more precisely unpredictability. For example, a series of coin tosses with a fair coin has maximum entropy, since there is no way to predict what will come next. A string of coin tosses with a two-headed coin has zero entropy, since the coin will always come up heads. Most collections of data in the real world lie somewhere in between.

http://en.wikipedia.org/wiki/Entropy_%28information_theory%29

A "string of zeros" would be like the two headed coin, a string of random "1" or "0" would be like a regular fair coin.
Changing the data from the former to the latter would increase entropy, in the informational sense of the word.

[Bored chemist]

"there is a computing version of the word "entropy""
It's the same thing.
If you were to store the information in, for example, the orientation of carbon monoxide molecules in frozen CO then the thermodynamic entropy would be the same as the computer science entropy.
This means that having ordered data stored would increase the weight of the computer and the extant of the weight change would depend on the temperature.
However, it's my understanding that  the people who designed the coding for the patterns of ones and zeros on disk drives did so in such a way that there are never long strings of 1 or 0 so the entropy is kept high.
I suspect that the change of entropy when you write data to the disk would be rather small

[Geezer]

From the little I can remember, I think the erased state of a disk is more ordered than the (typical) written data state. (Real data tends to look a lot more random.)

Does that mean the entropy of the erased state is actually lower than the entropy of the typical data state, and if so, would the weight actually decrease as data is written?

[Bored chemist]

"Does that mean the entropy of the erased state is actually lower than the entropy of the typical data state, and if so, would the weight actually decrease as data is written?"
Yes, or at least I think so.

[mattyh]

1's and 0's (Binary) is just open and closed or on and off switches.

These switches (billions of transistors in a PC)turn on and off and this encodes information.

I'm guessing that the harder a PC is working, the more electricity is being used and since electrons have a mass (i think), a PC that is idle or turned off would actually weigh less than one that was running at full capacity.

If the PC was turned off it wouldn't matter whether the hard drive was full or empty, the mass would remain the same.

Please correct me if I'm wrong.  Regards Matt

[Atomic-S]

Please do not confuse energy and entropy. in "E-mC²" E is energy, not entropy. During the time a hard drive is being written to, it experiences a temporary increase in energy. This however dissipates as it returns to its previous temperature (except insofar as there may remain different magnetic interactions between the various parts of the disk than existed previously.) Considering only the magnetic intereactions, the energy of the disk after it has been written to and returned to a static state, may be more than, the same as, or less than it was previously, depending on exactly the coding scheme and the information.

[imatfaal]

Atomic there are two clear parts to the argument and I dont think anyone was confusing S and E.  Conservation of information would say that for a system to move from an ordered to a disordered state there will be a heat change (absolutely mimiscule) - that is why people were discussing entropy

[qazibasit]

lets assume if 500 GB weigh 1 gm, still its a good deal for us, because carrying books equal to 500 GB would be like carrying a harvard university. Apart from the joke, hard disks are electromagnatic plates, they already have the charges on their plates, its just we alter them everytime we read or write data, so no net change in any mass or energy, u can say that it produces heat due to the head rotation.