Here's the delayed choice quantum eraser setup from Kim et al. (1999), from Wikipedia. Don't worry, we're going to throw out almost all of it.
An argon laser generates individual 351.1 nm photons that pass through a double-slit apparatus (vertical black line in the upper left corner of the diagram). We color code the two paths, but of course all photons and paths are the same color (wavelength). After the slits, spontaneous parametric down-conversion (SPDC) creates an entangled two-photon state.
The 702.2 nm photon that moves toward the upper right is referred to as the "signal" photon, the other the "idler" photon. I'm modifying the setup so that the idler photon goes through an equivalent setup as the signal photon:
My question is: do we see an interference pattern at both detectors $D_0$ and $D_*$? Or no interference pattern?
If there's no interference pattern: where is the which-way information? Either branch is basically a traditional two-split experiment, so I don't understand why there would be which-way information here, when there's none in the traditional. Where did it come from? Where is it living? How does it escape the setup into the macroscopic world?
If there is an interference pattern at both: then extend the length of the bottom path. This shouldn't change interference/no interference pattern. Now, after we get a detection at $D_0$, which we agree should follow the interference pattern in aggregate, swap out $D_*$ for a different setup that detects the two beams independently, thus measuring the which-way information.