We are using “full double-blind” review, so at this point the reviewers still do not know your identity.
Please DO NOT – under any circumstances – reveal your identity in your author response. If for whatever reason you feel that you need to reveal your identity to the reviewers (e.g. in order to authoritatively correct some errors in the reviews), then contact the program chair (me) by e-mail and we can discuss it.
There is no hard word limit on the response. That said, reviewers have finite time: do not expect that they will read every word of your response if it is long. This is why I suggest that you adopt a structure (described below) which places the most important part of the review (Overview and Change List) up front.
Submitting a response is not mandatory, but it is strongly recommended. Admittedly, if your paper receives only negative reviews, then frankly the chance that it will be accepted is fairly small, and your response is unlikely to turn things around.
But I would still recommend in this case at least submitting a brief response thanking the reviewers for their reviews (and of course feel free to submit a proper response if you want to correct reviewer misconceptions). On the other hand, if you think your paper has a chance of acceptance, then it is definitely in your interest to submit a proper author response.
I have seen many cases in the past where the author response convinced hesitant reviewers to up their score and champion a paper. Moreover, if you do not respond, it will send a signal to the reviewers that you either do not expect your paper to be accepted or do not take their criticisms seriously.
The author response serves two purposes:
(a) To correct factual errors or points of confusion in the reviews.
(b) To make the case to the reviewers that you have a plan for revising your paper to address their criticisms and that your plan is feasible, i.e. that you can execute your plan in the 19 days in between notification of conditional acceptance (Mar 6) and the 2nd-round revision deadline (Mar 25).
These are both equally important, but in my experience the mistake that many authors make when crafting their responses is to focus too much on (a) and not enough on (b). This leads their responses to be overly confrontational and defensive, when in fact what you want to do is convince the reviewers to accept your paper, not annoy them by pointing out the myriad ways in which they are wrong. :-)
Towards that end, I strongly recommend that you adopt the following structure for your review:
Begin your review with a sentence or two thanking the reviewers for their reviews. They put a lot of work in over the winter to produce 1300 reviews for PLDI. It is polite to thank them.
OVERVIEW: A brief overview of the salient points of the response. This is where you should of course highlight any major errors in the reviews, but also acknowledge major criticisms (especially those expressed by multiple reviewers) and explain briefly how you will address them. Avoid the urge to be defensive here. For example, if reviewers say some part of your paper was hard to understand, don’t argue: just acknowledge the criticism and explain how you will revise the paper to improve the presentation.
CHANGE LIST: A bulleted list of the major, concrete changes that you plan to make to your paper to address the reviewers’ criticisms. Usually, these changes should reflect the points of criticism discussed in the Overview. Your goal here is to convince the reviewers that you know how to revise the paper to their satisfaction, so that they end up with the decision: “The paper is accepted under the condition that the authors implement their change list.”
DETAILED RESPONSE: A reviewer-by-reviewer list of answers to questions and comments from the reviews. Use Markdown syntax to structure your response clearly (e.g. with bold headings like “Reviewer A”, “Reviewer B”, etc.), and to quote directly from the reviews. (Since there is no word limit, you are encouraged to quote directly from the reviews with “>” symbols, so that the reviewers don’t have to flip back and forth between your response and their reviews.) You do not need to address every single comment in the reviews, but it is a good idea to address the more substantive ones.