At present, the methods described above can be applied only to bright and local AGN (with a few exceptions, like the hard X-ray diagnostics, and the ISOPHOT colors). Indeed, almost all of the examples shown here are about observations of nearby, powerful IR galaxies. These sources are not only interesting in themselves but represent analogs of the population dominating the submillimeter background, which probably comprises high redshift galaxies. Understanding the energy source of these populations would ultimately give us the ability to estimate the total fraction of the radiated energy of the universe due to AGN. It is therefore very important to understand the perspective of extending these methods to fainter and more distant sources.
There are two main limitations of the methods described above, one due to physical limits, and one due to observational capabilities. Observations of better quality can be done with forthcoming instruments and can extend these diagnostics to a much wider range of sources. However, all the methods described above are based on the detection of either direct AGN emission or scattered, reflected, or reprocessed (in the case of emission lines) emission. If the gas column density is higher than ~ 1025 cm-2, the dust extinction is higher than ~ 8 - 10 magnitudes (even in the IR), and the obscuration of the source is complete along all lines-of-sight, then none of the above methods can be used. Even the mid-IR emission will be self-absorbed and reprocessed at longer wavelengths, so no big differences are expected in the far-IR spectra of AGN and starburst dominated sources. In this extreme case, all the observed emission of the active nucleus will be thermally reprocessed by the circumnuclear obscurer. There is then only one method to search for AGN in these sources: spatial resolution. In principle, high spatial resolution is rather simple: if high far-IR emission is detected in too small of a region, so that even the most compact starburst can be ruled out, then only an AGN can be the energy source. In order to be effective, this requires resolutions of at most a few tens of parsecs for ULIRGs, which is far from being reached for the nearest ULIRGS, even with the new Spitzer Space Telescope. (One milliarcsec at z = 0.05 corresponds to ~ 1 pc with H0 = 70 km s-1 Mpc-1). However, this remains a possibility (maybe the only one) for a future solution to this problem.
In summary, will we ever be able to understand the origin of the emission in the powerful far-IR and submillimeter sources and to estimate the total contribution of accretion to the energy output of the universe? The answer depends both on the improvement of instrumentation (and here no limit can be put on the development of science and technology) and on nature - even the most powerful and compact emission in the universe can be completely hidden by a thick enough screen.