This was a review paper, so that's one of the reasons for the high number of citations. I prepared it carefully. First, there was about a year between receiving the request to write it and the actual writing. During that period I carefully collected all relevant publications, and tracked down many of their citations. Using this and other procedures it was possible to write a fairly comprehensive review, without much important material falling through the cracks.

My strategy for the writing was first to create a comprehensive outline. It was of course modified as I went along but it was always designed to incorporate everything I wanted to discuss—it was pretty detailed. Then each day I worked on only one tiny bite, so I could do each bite justice without getting overwhelmed. Finally I circulated the draft version rather widely to collect comments.

The paper was almost entirely observation and interpretation rather than theory. This made it simpler and more tractable, and more limited to what I know best.

The topic, "Unified models for active galactic nuclei and quasars," touches the appearance of all active objects at some level, so it was relevant to a very wide body of research.

  What circumstances led you to your work?

I'd been lucky enough to discover, with my former thesis adviser Joe Miller, one of the tenets of these orientation-based unified models in the early 1980s: it was possible, through the polarization property of scattered light, to look inside some galactic nuclei for which the direct view is obscured. This showed that two phenomenological spectroscopic categories of active nucleus differed only in orientation with respect to the line of sight. In particular, the active galaxies without "quasar-like" bright optical/UV sources directly visible from Earth do have these power houses detectable via scattered light.

While the generality of the result is still being debated today, it's clear that it's at least fairly generic. Another key to the popularity of the result is the robustness: I think a 99.9%-sure result is much, much more precious than a 90%-sure result. Perhaps the reason is that scientific arguments typically use a multiplicity of putative results, and if each is only 90% certain, the arguments can quickly go astray.

I think a major reason that I was asked to write the article, and that it came out as well as it did, is that after finding the result unifying some classes of object based on optical spectra, I went into a fruitful line of research with Jim Ulvestad on the radio aspects of orientation. Thus I was one of the relatively few people who had first-hand knowledge of both the optical and the radio results.

  Can you describe the significance of this work for your field?

The unified model explains that what had been a bewildering variety of galactic nuclear activity properties and classes is really a result of something as simple as highly anisotropic objects seen along different lines of sight. This made the whole subject seem more tractable, and perhaps brought the physics issues into sharper focus. What the unified model doesn't do is shed any light on the workings of the central energy source, which takes place on very tiny scales compared with those at work shaping the orientation- dependence of the observable properties.

  Where has this research gone since the publication of your paper? Where do you see it going 10 years from now?

I don't think the observational situation on unified models has changed a great deal since that review was written. Later studies provided much more insight into the systematic properties of the total population of active nuclei. But much of the physical basis of the scenario was developed by Julian Krolik and collaborators on the optical side, and Roger Blandford and collaborators on the radio side, in plenty of time for inclusion in the review. So the most basic aspects were seemingly in hand.

There is one area in which great progress is being made right now. The spectral/polarimetric discovery of hidden nuclei of one type inside another type of nucleus applies to most or all radio quiet nuclei, and more or all of the most powerful radio loud nuclei. But since the review, evidence has been emerging that many of the weaker radio galaxies do not have hidden nuclei (work by D. Chiaberge et al., D. Whysong and myself, among others). This turns out to be crucial information regarding the accretion modes and power sources for active nuclei. In 10 years we'll know exactly which kinds of galaxies have hidden nuclei, an important clue for theorists.

  What lessons would you draw from your work to share with the next generation of researchers?

Go for results that are going to be interpretable and general. It is sometimes better to have a few such results and many failures than to have many minor successes. When writing a review paper, start collecting literature as early as possible and try to be comprehensive. When writing a long paper or doing any other large difficult task, break it into tiny pieces and only do one tiny piece per day.