Peer review, decoded

Peer review is the quality-control step that turns a draft into a publication: before a venue accepts your paper, a few independent experts read it and judge whether the claims hold up. It is slow, imperfect, occasionally maddening, and still the best system anyone has found for keeping the scientific record honest.

The cast of characters

Role	What they do
Reviewer (or referee)	Reads your paper closely, writes a critique, recommends a verdict
Program committee (PC)	The pool of reviewers for a conference; “PC member” means reviewer
Area chair (AC) / meta-reviewer	Oversees a handful of papers, weighs the reviews, makes a recommendation
Editor	The journal equivalent: picks reviewers and makes the final call

Typically 2 to 4 reviewers read each paper. They are unpaid volunteers, usually researchers in your subfield, reviewing on evenings and weekends. This explains a lot about review quality variance.

Blind, and how blind

• Single blind: reviewers know who you are; you do not know who they are.
• Double blind: neither side knows the other. You submit an anonymized paper with author names removed.

Double blind is standard at most CS conferences and aims to reduce bias toward famous names and institutions. Check your venue’s rules before submitting; an accidental name in the PDF can get you desk-rejected.

The verdicts

Journals usually answer with one of four outcomes:

Verdict	Translation
Accept	Rare on the first try. Celebrate loudly.
Minor revision	Fix small things; you are nearly in
Major revision	Real work needed, but the door is open
Reject	No, but usually with reasons you can use

Conferences mostly compress this to accept or reject, sometimes with a “conditional accept.” Reviews often come with scores, like a 1 to 10 rating plus a confidence rating. Scores anchor the discussion but do not decide it; a champion reviewer arguing for your paper matters more than a decimal average.

The rebuttal, in one paragraph

Many conferences give authors a short window to respond to reviews before the final decision, called a rebuttal or author response. You get a few hundred words to correct misunderstandings, answer direct questions, and point to evidence already in the paper. It can flip a borderline decision, and writing a good one is a skill, covered in rebuttals and revisions.

Rejection is the default, not a disaster

Top venues accept roughly 15 to 25 percent of submissions. Flip that around: the typical outcome for a good paper at a good venue is rejection. Nearly every famous paper you have heard of was rejected somewhere first. The practical move is to read the reviews cold, a day later, extract every fixable weakness, and resubmit stronger.

What this means for how you write

Reviewers are tired volunteers reading fast. Clear structure, honest claims, and clean typesetting all buy goodwill. Knowing the standard sections helps too, see the paper anatomy tour, and a tidy LaTeX setup, like your first document, keeps formatting from ever being the reason a reviewer frowns. The whole path, from draft to verdict to camera-ready, is mapped in the first paper roadmap.