Kozen pointed out that, at least at Cornell, recruiting is discussed a lot at faculty meetings. For example, right now there is huge load on the systems faculty, and therefore candidates in systems are desirable. In other words, area counts along with references, publications, and so on. The committee along with input from the entire faculty will converge on a list of names and decide on people to invite for interviews. The committee offers a report every week to the faculty on what the current status of the search is. He also pointed out that reference letters in general tend to be a bit overblown, and thus are usually taken as such.
An interactive discussion about various aspects of the process followed Kozen's introduction. Zabih pointed out that while the committee will decide whom to invite (based on input from committee and other faculty members) the faculty as a whole will decide whom to hire. Cornell is not necessarily representative in this respect. Other departments are known to have more powerful committees or other sorts of decision-making arrangements. Zabih suggested that the single most important thing in the applicants folder is the area of specialization. The faculty will need to decide how much emphasis to place on different things. Kozen suggested that sometimes faculty will hold a position open while waiting for just the right person to fill it.
Tardos offered some insights into the interviewing process. She pointed out that interviewing in industry is very different from interviewing for an academic position. In academia, you will talk to people who know nothing about your area and will be more interested in general sorts of things. They will want to know what your research plan is, what you expect to do yourself, what ideas you might have for your students to work on and where you're planning to get funding. They'll likely also be interested in what you expect to teach. Tardos' advice is to be prepared to teach just about everything at the undergraduate level and also graduate courses in your own area. Kozen mentioned that a common complaint among interviewers was candidates who only know about their own specific research area. Tardos agreed and added that it's impressive when a candidate is aware of local research projects, particularly those even tangentially related to their own. Zabih gave examples of other sorts of general questions that might be asked, e.g., What is the next hot thing in your area? What's going to make you famous? Tardos added: What will be happening in your area in the next 10 years? What major results came out of the last 10 years? Zabih pointed out that during the interview process (which is extensive and exhausting) the faculty are on the ball. They're trying to figure out if they can live with you for the next few years/decades. During interview talks people are paying attention, which makes giving the talk both an opportunity and a challenge. Tardos added that applying to all of the top schools, even if you know some are not where you'd like to be is a good thing. Being invited to give a talk not only helps your job prospects, but also gives you some exposure and the chance to meet more people in the field.
A question was raised about what sort of flexibility there was in scheduling the interviews. Kozen answered and pointed out that there is a huge host of constraints, and especially near the end of a term things can become very tight. Along the lines of scheduling and constraints, Tardos offered some insight into how to handle being half of an instance of the "two-body problem." Essentially her advice is to keep quiet. This is a very big problem with no easy solutions. When the two bodies are in completely different fields, it can be even more trying, as different fields have different interviewing schedules. One partner may have to make a decision before the other has even been given any offers. Even when both are in CS, it's best to refrain from commenting on the problem and wait until offers show up to make a decision. During the interview process, you may be asked whether such a problem exists. (For the record, Tardos pointed out that it is illegal to ask personal questions, but in general people are well-intentioned, so reacting badly is not in your best interest.) When deciding how to answer, keep in mind that every such illegal question has a legal alternative. Thus, the best thing to do is to answer the legal question. For example: My interviewing and decision is not related to that of any other person. Obviously, don't point out that the question is illegal and be as friendly and polite as possible.
Zabih pointed out that the entire interview process is grueling. Essentially it will consist of 10 or more half hour discussions with faculty members who are not in your area. The faculty will be pushing you both to find out if you have a vision for your research and if you will be a reasonable colleague. Kozen advises that maximizing the number of people to connect with at a given place is a good strategy. Tardos pointed out that the web has become a great resource. Before arriving you can get a lot of information about the people you'll be meeting. Kozen strongly urged that if you're going to be meeting someone even close to your area, to make sure you've read that person's papers closely enough to have more than a passing familiarity. Zabih pointed out that from the time of your arrival to the time of your departure, everything is about business, including breakfast, lunch and dinner. In other words, don't make any other plans for your time there.
Several questions were asked regarding post-docs, teaching positions and publications. Tardos noted that post-docs can strengthen your position as an applicant. The drawback is that in a short (one-year or less) post-doc, there's barely time to do any research at all before the interview cycle starts. People will be looking for post-Ph.D. research and therefore a two-year post-doc might be preferable. It was pointed out that in general you will get one shot at a faculty position. It is extremely rare if not unheard of for someone to be interviewed twice at the same place.
Regarding publications, the entire panel seemed to agree that quality matters more than quantity. Serious publications are a necessity. Zabih pointed out that being known (through conferences, publications, etc.) is a very good thing. A question was raised about whether teaching at a small college (as opposed to taking a post-doc) was an effective strategy immediately after the Ph.D. The general consensus seemed to be that a post-doc would allow more time for research and that it was hard to move "up" from a college to a university. The institution you're immediately coming from matters a lot. Recall that Cornell automatically collects letters from the top few places. Kozen pointed out that some might prefer a position at a "teaching college" and Tardos agreed, noting that taking a position at a place like Cornell is a complete gamble, whereas a tenure-track position at a lesser-ranked university or a small college is probably not as risky.
After a question on gender ratios, Kozen pointed out that it is illegal to discriminate in hiring practices. There is a lot of interest in balance and diversity. Tardos mentioned that there may be some bias in interviewing. That is, there is a lot of pressure to interview women and other minorities, but that this pressure does not at all reflect in hiring decisions.