Mental model
In M2 you saw HQC as an independent backup KEM to ML-KEM. This lesson brings two things together: a short reminder of HQC’s current status, and the current state of NIST’s separate “Additional Digital Signatures” program. These are different processes and should not be confused.
HQC: a short reminder
As you saw in M2, HQC is SELECTED2025-03-11 today (September 2026), with no draft FIPS yet. This lesson does not repeat that status; it only makes clear that HQC is a separate process that must not be confused with the next topic (round 3 additional signatures): HQC is a backup KEM to ML-KEM (an existing, final standard), while the round 3 candidates below are part of an entirely separate evaluation that adds to ML-DSA, SLH-DSA and FN-DSA on the signature side.
IR 8610: why there is a second process on the signature side
The “don’t put all your eggs in one basket” principle you learned in M2 applies on the signature side too: alongside ML-DSA, SLH-DSA and FN-DSA (all three already selected), NIST is evaluating additional signature candidates from even more varied mathematical families through a separate process called Additional Digital Signatures. IR 8610 (14 May 2026, final) reports this process moving from round 2 to round 3: 9SOURCED candidates advanced to round 3: FAEST, HAWK, MAYO, MQOM, QR-UOV, SDitH, SNOVA, SQIsign, UOV. According to IR 8610’s own Table 3, these candidates come from four different mathematical families (isogeny, lattice, MPC-in-the-Head, multivariate): a search for even broader diversity than the lattice, hash-based and code-based trio you saw in M2. (Live verification note: CSRC’s round 3 project page today lists FAEST under a separate fifth “symmetric-based” heading; NIST’s own taxonomy has been updated since IR 8610 was published, an example of how the “check the live source” discipline makes a difference even here.)
HAWK’s withdrawal: from 9 to 8
If you check round 3’s live project page today (September 2026), one of the nine candidates, HAWK, is marked “withdrawn”; the number of active candidates has dropped to 8SOURCED. The reason is not a procedural problem but a real cryptanalysis finding; you will see the full case (who found it, how, how quickly) in M14 (the-hawk-case). The important point here: this withdrawal is not a failure of the process but the opposite, the process working as designed. A candidate being found weak and eliminated during broad public review, before becoming a final standard, is exactly the purpose of the round structure; in M13 (cost model) and M14 you will see the economic and operational consequences of this “find early, fix cheaply” principle in more depth.
Round 3 itself is an evaluation, not yet a selection. IR 8610 itself gives no duration; it only says the 7th PQC Standardization Conference is planned for the first half of 2027 (the common secondary-source summary that “round 3 will take about two years” is an inference from that, ESTIMATED, not stated directly in IR 8610’s text). Presenting any of these candidates to an architect as “ready to use” would be the same kind of overclaim as presenting HQC as final while it is still pre-draft.