The 2027 anchor calendar, checked 2026-07-08 (SIAM SODA27/ALENEX27/SOSA27 pages
via search renderings; siam.org blocked direct fetches): papers due July 9,
2026 (AoE) on soda27.hotcrp.com; reviews to authors by September 1; responses
due September 4; decisions October 2026; conference January 24-27, 2027,
Philadelphia, co-located with ALENEX (papers due July 20, 2026) and SOSA
(papers due August 6, 2026). SODA appoints a new PC every cycle under joint
SIAM/ACM SIGACT sponsorship — every date here is a 2027-cycle snapshot, void for
any other year until re-checked.
Theory groups plan around three rotating deadlines, and SODA's is the summer one:
| Beat | Deadline season | Conference season | 2026-27 anchors |
|---|---|---|---|
| SODA | early-mid July | January | July 9, 2026 → January 24-27, 2027 |
| STOC | early November | June | pattern; 2027 dates 待核实 |
| FOCS | early April | October-November | pattern; 2027 dates 待核实 |
Sitting in SODA's seat, the planning consequences are:
| Checkpoint | State of the project |
|---|---|
| T-12 weeks (mid-April) | Result proved to the group's satisfaction; claims ledger opened (soda-reproducibility) |
| T-8 weeks (mid-May) | Full proofs written as prose — writing is the verification |
| T-6 weeks (late May) | Blind proof-check pass: each proof verified by a non-author of that proof |
| T-4 weeks (June) | Front matter built: title-page abstract, overview, prior-bound table (soda-writing-style); final arXiv sweep for concurrent work |
| T-2 weeks | Format and anonymity audit (soda-submission); HotCRP record created; conflicts entered |
| T-1 week | Full-document read by the least-involved coauthor; freeze |
| July 9, AoE | Upload confirmed by re-downloading the stored PDF |
The classic failure is proving through June: theorem-finishing scheduled inside the writing window converts both into deadline-night work.
This pack's anchor deadline (July 9, 2026) sits in early July — so a reader in the first week of July 2026 is not planning, but triaging. Honest triage:
| State of the paper today | Verdict |
|---|---|
| Proofs complete, front matter rough | Submit: polish the abstract and overview first, in that order — they carry triage review |
| One proof "almost works" | Do not submit the claim. Either weaken the theorem to what is proved, or hold the paper for STOC |
| Complete but nobody re-checked the proofs | Submit only with a same-week blind check of the main theorem; a withdrawn SODA paper costs less than a refuted one |
| Result strong, writing chaotic | Submit if the first ten pages can be made honest by Thursday; the back matter is verified, not enjoyed |
| Waiting on an experiment | Cut it; experiments are optional evidence at SODA (soda-experiments) and never worth a missed AoE cutoff |
The one thing not to do in deadline week: broaden claims to justify the rush. Reviewers arrive in August with time you did not have.
soda-author-response).soda-camera-ready); registration logistics for Philadelphia.Every deadline on this page needs exactly one named owner, decided in June:
soda-reproducibility
through submission and into rebuttal week.Decision tree for a group with theory + implementation results in June 2026:
1. Theory result stands alone? -> SODA by July 9
2. Implementation contribution disjoint? -> ALENEX by July 20 (AE follows)
3. Simplification of a known result? -> SOSA by August 6
4. Same claims in two of the above? -> pick one; overlap risks both
Each paper needs its own contribution sentence that mentions no sibling.
One calendar per paper, one owner per deadline. Shared-author papers at SODA and a satellite also share the January trip — one registration decision, one talk rehearsal block, and no scheduling collision (satellites run January 24-26, inside the SODA window).
[Cycle verdict] Submit July 9 / hold for STOC / re-route
[Backward plan] <checkpoint dates with owners, from today>
[Quiet-phase tasks] <ledger, rebuttal prep, arXiv, satellite decisions>
[Contingency] <October rejection branch: repair-for-STOC or wait>
[Satellite calendar] <ALENEX/SOSA/ITCS deadlines in play, if any>