Where does photon interference happen

[zonde]

If we look at http://en.wikipedia.org/wiki/Mach%E2%80%93Zehnder_interferometer" I think it's fair to say that any interpretation of QM says that there are appear no photons (in idealized case) at detector 2 as there are no detections. Basically that means that photon interference takes place at beam splitter.

As I see this approach conflicts with results of this experiment:
http://physics.nist.gov/Divisions/Div844/publications/migdall/psm96_twophoton_interference.pdf"
where interference is observed after "postponed compensation" that takes place after beam splitter.

To make reasoning more straightforward let's say we use photon pair from donwconversion process where signaling photon goes through interferometer but idler is detected just for tracking purposes. So the setup is like this:

Will we see 100% coincident detections between Di and D2 with ideal detectors and ideal source (all photons are paired up)?

 

[Greg Bernhardt]

@zonde did you find any more insight on this topic?

 

[zonde]

Here are valid links in place of those broken ones:
https://en.wikipedia.org/wiki/Mach-Zehnder_interferometer
https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.77.1917
And another similar paper https://arxiv.org/abs/quant-ph/0304030

These two-photon interference experiments demonstrate the passive role of first beamsplitter in SA type experiment but they extend these results to HOM experiment and claim that in case of HOM experiments beamsplitter has passive role as well.
In my question I was relying on correctness of this conclusion but my insight on the topic is that this conclusion is just wrong. HOM and SA experiments are very different despite some similarities in setup so that conclusions from modified SA experiment can not be extended to HOM experiment.