Skip to the Main Content

Note:These pages make extensive use of the latest XHTML and CSS Standards. They ought to look great in any standards-compliant modern browser. Unfortunately, they will probably look horrible in older browsers, like Netscape 4.x and IE 4.x. Moreover, many posts use MathML, which is, currently only supported in Mozilla. My best suggestion (and you will thank me when surfing an ever-increasing number of sites on the web which have been crafted to use the new standards) is to upgrade to the latest version of your browser. If that's not possible, consider moving to the Standards-compliant and open-source Mozilla browser.

April 11, 2005

Susskind on wormholes

Posted by Aaron

Lenny Susskind was recently here and discussed his recent paper on wormholes. As you probably know, he’s mentioned some of the counterarguments in his most recent paper. I think that there’s another possible perspective on the time difference. It’s sort of an intriguing combination of classical stuff and the uncertainty principle(s). I wrote it up formally, but it’s sketchy enough that I thought I’d just put it up here. Feel free to poke holes!

(Or I’ll just do it myself. I made a silly mistake in the last part. I’ve made some changes, and I’ll leave it up, but it’s late at night, and I can’t come up with a quick fix tonight.)


It has long been known that general relativity seems to allow interesting solutions such as wormholes [1] and warp drives [2]. This is perhaps not terribly surprising in that we can take any metric on a spacetime, apply Einstein’s field equation and declare, by fiat, that G μν=8πT μν. Thus, in order to restrict the possible solutions, we must supplement Einstein’s field equation with a restriction on the types of stress-energy allowed. These are termed energy conditions. It can be shown that traversable wormholes [1,3,4], warp drives and, indeed, any method of faster-than-light travel [5,6], must violate an energy condition, usually the averaged null energy condition.

Nonetheless, we are not out of the woods. There are many quantum systems which are known to violate various energy conditions. The Casimir effect serves as the most common example. Other examples include the theory of inflation which depends on the violation of the strong energy condition. There are also known to be exotic fluctuations around the horizon of an evaporating black hole that allow the violation of Hawking’s area increase theorem. There has been research in constraining the types of energy condition violation allowed by quantum mechanics [7] but the situation remains uncertain. To my knowledge, while very much doubted, neither the existence of wormholes nor of warp drives has been ruled out.

Susskind [8] has recently proposed two gedankenexperiments that purport to show how wormholes, if they exist, lead to some unexpected weirdness irrespective of the energy conditions. In the first scenario, he uses the fact that a charge traversing a wormhole adds a unit of flux along the wormhole. Thus, if we take two systems in some correlated charge superposition state and pass one through the wormhole, it becomes correlated with the flux quantum number of the wormhole. If we then attempt to measure the correlation between the two original systems, taking one the long way around, we will not measure any correlation because of the entanglement with the flux.

The subject of this paper is the second scenario described by Susskind. There, he uses the fact that the wormhole seems to exactly measure energy to show that the uncertainty in time is infinite, as follows from the energy/time uncertainty relation. While there are many subtleties associated with this uncertainty relation, they will be irrelevant for the purposes of this paper. In section two, we will review Susskind’s argument in more detail. In section three, we will resolve the apparent paradox in 3+1 dimensions and show how the energy/time uncertainty relation is always satisfied. We demonstrate how this (generally quite small) time uncertainty can be seen to follow from the usual uncertainty relation for position and momentum. Interestingly, this argument does not seem to apply to the 2+1 dimensional case, leaving an intriguing problem. As is usual in such arguments involving the uncertainty relations, we will assume that they are saturated. We will also neglect leading constants and only keep track of the dependence on the various parameters of the problem.

Susskind’s Wormhole

To model Susskind’s wormhole, we can begin with flat 2,1. In the two dimensional space, select two circles (cylinders in spacetime) extremely far apart. If we identify these circles, we have created a wormhole. While in higher dimensions, the wormhole mouths will generally warp the spacetime far away, in 2+1 dimensions they only create an angle deficit. Thus, we can measure their masses. Call the two mouths A and B.

Now, let us place two identical synchronized clocks near mouth A. These clocks must have some mass, so as to be able to measure time with any accuracy. Now take one of the clocks to mouth B, going the long way around, ie, not through the wormhole. Pass the other clock through the wormhole. The question is whether the clocks are still correlated. For example, if we were to use our knowledge of general relativity to compute the time change for the clock that took the long way, would that be the difference in times shown on the clocks? Or, if we were to repeat the experiment with any number of clocks, would the differences in time have a highly peaked distribution or be relatively flat?

An interesting fact about wormholes is that they change in mass as an object passes through them. To see this, imagine a wormhole connecting two distinct asymptotically flat spacetimes. In each spacetime, the ADM mass is conserved. Thus, if we pass an object of mass m through the wormhole, from A to B, the ADM mass on either side cannot change. This means that the mass of the mouth A increases by m and the mass of mouth B decreases by m. In other words, the wormhole has measured the mass of the object. Note that this argument applies in any dimension in which the ADM mass makes sense.

Now we come to the crux of Susskind’s argument. Since the wormhole has measured the mass of the clock, we have ΔE=0. By the usual energy/time uncertainty relation,


we have Δt=. In other words, the times on the clocks are completely uncorrelated. If you were to enter the wormhole, you would have no idea when you would come out. Performing the experiment repeatedly, we would observe a distribution with infinite standard deviation for the time differences of the clocks. If not a stark paradox, this makes wormholes significantly weirder than they might appear at first glance.

The resolution

Energy and time uncertainties

The problem in the above is the idea that we can measure the mass of the mouth of the wormhole exactly. The see the problem, let us recall the definition of the ADM mass. We will hence forth work in 3+1 dimensions1. I will be schematic; for more rigor, see [9] for example. In an asymptotically flat spacetime, we know that the metric approaches a flat metric as some coordinate, r, approaches infinity. The ADM mass is then defined by the integral

M116πGlim r μ,ν=1 3(g μν,μg μμ,ν)N νdS.

Here, we integrate over a surface at infinity where N ν is the unit normal to this surface.

One immediately sees a problem. In our wormhole spacetime, there is only one asymptotic infinity and, thus, only one ADM mass. While we might consider this as invalidating the scenario, this does not accord with our intuition that, when the wormhole mouths are sufficiently separate, we should not have to worry about the other mouth. Thus, we ought to be able to measure some sort of mass for each mouth in this limit. To see how we might do this, note that, sufficiently far away from the mouth, we can recast the metric in the following form

g μν=η μν+(η μν+2V μV ν)2GMr+O(r 2).

The M that appears here will serve as a definition of the mass of the wormhole mouth. The Vs are just velocities that we can set to zero.

Now we can see the difficulty with Susskind’s scenario. In fact, we cannot precisely put the metric in the form (1). As we can superimpose the perturbations, the true metric will be of the form

g μν=η μν(1+2G(M Ar A+M Br B))=η μν+η μν2Gr B(M B+M Ar Br A).

Assuming we are close to mouth B, we can approximate r Ad. Comparing with (1), this gives the measured `mass’ of mouth B as

MM B+M 𝔸r Bd

Thus, we have


Note that, in accordance with our intuition, as we separate the wormhole mouths, we decrease the uncertainty. Now, if we want to transit an object of mass m from mouth A to mouth B, the value of M B must be at least m so as to not give the mouth a negative mass when the object emerges2. This implies that the best accuracy we can expect is GM Am/d.

Now, we plug this into the energy/time uncertainty relation to obtain:

Δt/ΔE=dGM Am.

This is generally a quite small number. However, we can certainly imagine an enormous universe where we can set d as large as we wish, giving arbitrarily large time uncertainties. Thus, we have apparently not eliminated the weirdness from our world.

Momentum and position uncertainties

In fact, there is more uncertainty in the problem. At the beginning, each box has to reside in the region of the wormhole’s mouth. Let’s say it begins at a radius q. Thus, the uncertainty in its position is certainly less than q. This means that the usual uncertainty relation, ΔpΔx, implies an uncertainty in the momentum


From the standard fact, d=vt=pt/m, we have

Δt=mdΔpp 2=dΔpKEdqKE

where KE is the kinetic energy of our object, and we have plugged in (3).

Now, in order to escape the gravitational well of wormhole A, assuming no other energies in the problem, we must have an initial KE of at least GM Am/q. Plugging this into (4), we obtain

ΔtdGM Am

which is exactly the relation from equation (2) that we derived from the energy/time uncertainty relation. One might ask what would happen if we were to decrease M A. In fact, because we passed one clock through the wormhole, M A has a minimum value of m and cannot be decreased to zero.

There is still a problem here, however. We have shown that we can account for the uncertainty in one scenario, but it seems as if we can increase KE to lower the uncertainty in the time of transit. Must think more….


I would like to thank Jacques Distler, Lenny Susskind and Uday Varadarajan for useful conversations.

1As noted above, in 2+1 dimensions, the ADM mass corresponds to a deficit angle. As no matter implies flatness in 2+1 D, we can measure the mass of an object unambiguously without going to infinity.

2 To my knowledge, this statement is unproven. Normally one would expect that the ADM mass must be positive from the positive mass theorem. However, this theorem depends on the dominant energy condition which will be violated in a wormhole spacetime. Nonetheless, the prospects of an unbounded from below mass are disturbing enough that this seems like a reasonable assumption.


[1] M. S. Morris and K. S. Thorne, “Wormholes In Space-Time And Their Use For Interstellar Travel: A Tool For Teaching General Relativity,” Am. J. Phys. 56, 395 (1988).

[2] M. Alcubierre, “The warp drive: hyper-fast travel within general relativity,” Class. Quant. Grav. 11, L73 (1994).

[3] M. Visser, Lorentzian Wormholes: From Einstein to Hawking.

[4] J. L. Friedman, K. Schleich and D. M. Witt, “Topological censorship,” Phys. Rev. Lett. 71, 1486 (1993) [Erratum-ibid. 75, 1872 (1995)].

[5] R. J. Low, “Speed limits in general relativity,” Class. Quant. Grav. 16, 543 (1999).

[6] K. D. Olum, “Superluminal travel requires negative energies,” Phys. Rev. Lett. 81, 3567 (1998).

[7] L. H. Ford and T. A. Roman, “Quantum Field Theory Constrains Traversable Wormhole Geometries,” Phys. Rev. D 53, 5496 (1996), and many more by the same authors and by others.

[8] L. Susskind, “Wormholes and time travel? Not likely”.

[9] R. M. Wald, General Relativity.

Posted at April 11, 2005 11:47 PM UTC

TrackBack URL for this Entry:

5 Comments & 3 Trackbacks

Read the post Wormholes
Weblog: Musings
Excerpt: Lenny spends a week in Austin.
Tracked: April 12, 2005 4:53 AM

Re: Susskind on wormholes

Is there a landscape-related reason for Susskind to be interested in wormholes?

Of course wormholes could sort of connect otherwise far apart points in a landscape multiverse, but is their existence or non-existence thought to have any consequences for the question of which vacuum we should expect to find ourselfs in?

On a related note:

Does Susskind consider any stringy effects? There have been various claims that stringy effects would do away with pathologies in classical GR, as one would expect of a working theory of quantum gravity. In particular I am thinking of

L. Dyson,
Chronology Protection in String Theory,

where it was claimed that closed timelike curves (CTCs) in classical GR solutions are elimiated by stringy effects which manifest themselves as unitarity conditions in a duala CFT.

So I’d expect that any string theorist interested in the pathology or viability of wormholes would like to take such stringy effects into account. Does this play a role anywhere here in the problem that you are disucssing?

Posted by: Urs Schreiber on April 13, 2005 7:54 PM | Permalink | Reply to this

Re: Susskind on wormholes

Is there a landscape-related reason for Susskind to be interested in wormholes?

Not as far as I know. I just don’t think he likes them very much.

In particular, for the purposes of his first paper, he really didn’t care about CTCs, chronology or anything else. Just wormholes and the uncertainty relation.

Posted by: Aaron Bergman on April 13, 2005 11:25 PM | Permalink | Reply to this

Re: Susskind on wormholes

I should mention that Roman Buniy and Stephen Hsu have a paper on wormholes and energy conditions and stuff. I don’t know much about it, but check it out if you’re interested.

Posted by: Aaron Bergman on April 15, 2005 6:45 AM | Permalink | Reply to this
Read the post Warped Ambitions?
Weblog: Asymptotia
Excerpt: Just spotted this in the Guardian: Next Thursday, the British Interplanetary Society is bringing together physicists for a conference entitled Faster than Light: Breaking the Interstellar Distance Barrier. “The main purpose is to raise awareness ...
Tracked: November 18, 2007 3:58 AM
Read the post Warped Ambitions?
Weblog: Correlations
Excerpt: I noticed that there was a conference hosted by the British Interplanetary Society about research on things like warp drive. You know, the way they zip around space in Star Trek. Is this crazy stuff? Where does the science begin and the science fiction...
Tracked: November 18, 2007 4:03 AM

Re: Susskind on wormholes

Re Wormhole mouth mass…

“imagine a wormhole connecting two distinct asymptotically flat spacetimes. In each spacetime, the ADM mass is conserved.”

I’ve never got that argument (see e.g. Matt Visser on Lorentzian Wormholes…; ditto charge)

In the case of traversable wormholes in particular, there is no horizon, so why do the two spacetimes count as separate regions? I can see the angle deficit creating the appearance of mass but have never been able to see how passing a mass through such a wormhole cause the masses of the mouths to change.

Enlightenment welcomed!

Posted by: Julian on March 31, 2009 9:24 AM | Permalink | Reply to this

Re: Susskind on wormholes

read it again.

How does one reconcile the facts that “In our wormhole spacetime, there is only one asymptotic infinity and, thus, only one ADM mass.” (which is what I had inferred myself) and that in the argument for separate wormhole mass conservation - which leads to the potential of negative mass - Visser specifically relies on the fact that there are two spacetimes with separately conserved ADM masses (even though an intra-universe wormhole is considered).

Posted by: Julian on November 24, 2009 10:41 AM | Permalink | Reply to this

Post a New Comment