In this post, I address a question that pops up from time to time as a possible objection to the transactional picture. The scenario involves a very distant star that engages in a transaction with a person’s eye, so that they see the star as it existed billions of years ago. But suppose the star has long since ceased to exist, and that it sent out that photon long before this observer was born? How did the star “know” that the observer would be in the right place at the right time to engage in this transaction?
Actually, the star didn’t need to know, because it didn’t send out the photon “long before the person was born” in any absolute sense. There are two main issues overlooked in the construction of this little paradox:
(1) In view of relativity, distances and time lapses are only relative, as is the time order of spacelike-separated events.
(2) No transaction can be set up without the availability of an absorber, in the present, so that any photon transfer establishes the emission event in the past.
First, consider point (1), with reference to the diagram below. Assume the star and the person (call him Bob) are in the same inertial frame. The star’s timeline is on the left; event E is the emission, and event F is the star’s demise. Bob’s timeline is at location x. The story in which the star has ceased to exist before Bob comes along only holds relative to certain inertial frames, among them Bob’s rest frame. In fact, there is no invariant distance between the star and Bob, nor is there any invariant time of travel for the photon to get from the star to Bob. There is also no invariant time order of the events involving Bob’s birth (denoted by B) and the star’s emission event E, since these events are spacelike-separated. Consider a rocket ship traveling very fast from the star towards Bob. Its spatial axis (compressed to one dimension) is the slanted line intersecting point C. This means that point C is simultaneous with the star’s emission according to the rocket’s perspective. From the rocket’s perspective, Bob was born before the star emitted, even though from Bob’s perspective, he was born after the star emitted.
According to the rocket, the spatial distance from the star to Bob is x’, and the time it takes for the photon to reach Bob is t’. These are much smaller than the values x and t assigned by Bob. Thus, from the rocket’s perspective Bob is much closer to the star, and the time of the photon’s travel is correspondingly reduced. In addition, from the standpoint of the rocket, the star dies well after Bob becomes available as an absorber (well after C, as can be seen by drawing a line parallel to the x’ axis from F to Bob’s timeline).
The lesson here, courtesy of relativity, is that the ability of a source to engage with absorbers is not restricted by sequences of events or spatiotemporal displacements relative to any particular inertial frame, since those are not absolute conditions. In this case, we see that according to the rocket ship, there is nothing “out of order” about the star engaging in a transaction with Bob.
Regarding point (2): The advent of incipient transactions is governed by absorbers in the present, and the actualized transaction acts to extrude the new spacetime interval from the present into the past, as a new element of the “spacetime fabric” (think of a knitting process). In this sense, all transactions have a form of built-in retrocausation, but it’s limited to the establishment of new spacetime events (not any changes to already-actualized events). Emitters and absorbers negotiate in the present (which we can identify with “Quantumland,” the quantum substratum which is a precursor to spacetime) via offer waves (OW) and confirmations waves (CW). It’s only at the final stage of an actualized transaction that a “past event” is established— the actualized emission event. So generation of OW and CW, which act beyond spacetime in the quantum substratum, must be carefully distinguished from the actualized real photon that is a spacetime entity—the connection between actualized emission and absorption events—and is represented by a projection operator. Emitters and absorbers remain in the quantum substratum. The actualized events that make up spacetime are activities of emitters and absorbers; the latter never become part of the spacetime manifold. In this sense, they are “eternally present.”
Transactional Interpretation 
This is reminiscent of the homework problems given in undergrad electrodynamics (such as the bomb problem). It also reframes the contingent absorber experiment in more accessible language. We could even appropriate the system for a delayed choice experiment. In this form, the star and stargazer are causally separated; however, one could imagine the rocket outrunning the light (for instance, the light spending a long time orbiting a massive body just above its Schwarzschild radius). The rocket then could “choose” to intercept the light.
Similarly, one could substitute mirrors or a lossless dielectric for the curved spacetime. This paper (https://arxiv.org/abs/1601.03796) regarding the wavefunction in such media notes the problem of superposition within single photon interferometry. If we were similarly careful with this example (explaining the entire system in terms of unitary quantum mechanics within the emission and absorption timeframe), we might have difficulty explaining how the light from the star and the entire rocket are not in a superposition.
In your 2012 response to Maudlin, you mention that the choice could be made by a quantum random number generator rather than a macroscopic chooser. This doesn’t necessitate questions regarding causality unless the emitter sends a signal to the chooser before the observed photon arrives. Of course, this would then raise the superposition question. Shouldn’t a fundamental physical theory be interrogated on that theory’s own terms? I would like to see versions of the delayed choice and contingent absorber experiments that engage in such a treatment.
Thanks, Adam.
First, I should note that Maudlin’s contingent absorber experiment can’t get off the ground at all when the relativistic form of TI is taken into account (RTI). This is because, at the relativistic level of interacting fields in the direct-action picture, we find that there really are no “slow-moving offer waves” that are required to instantiate the experiment. OW/CW apply only to massless bosons (photons). Quantum systems with non-vanishing rest mass (fermionic systems) do not engage directly in OW/CW transactions (they don’t have their own ‘CW’), but instead are detected indirectly through photon transactions. This is discussed in my 2019 treatment of the Maudlin issue here:https://arxiv.org/abs/1610.04609
You propose slowing down the photon OW, but in fact, we can’t really do this. Even according to standard quantum theory, photons are not localizable and can’t be outrun by any material object. While we can make a timelike ‘path’ for the photon to increase the time required for it to reach us relative to our inertial frame, the OW/CW are always on null cones and are Fock states that directly link emitter and absorber, so they can’t be interceded by other objects or signals.
RTI also explains why there is no overall entanglement or superposition of the photon OW and the rocket. OW are generated based on direct participation of absorbers, and are generated directly to the absorber in a non-unitary process. In any case, the emitter could never send a signal that could outrun an already-emitted photon, for the reason above.
Some of this is newer research that will be forthcoming in the 2nd edition of my 2012 CUP book. Let me know if you need further details and I can arrange for you to have a preview of relevant sections.
Thanks again for your great questions!