Mystery Constants in SHA256 Encryption

The real NSA backdoor, which is probably what you’re remembering, is Dual_EC_DRBG. It’s a random-number generator standardized by NIST in 2006.

  • Its constants (two elliptic-curve points, P and Q) came with no explanation of where they came from.
  • Researchers pointed out in 2007 that whoever knew the secret relationship between P and Q could predict its output.
  • Snowden documents in 2013 indicated NSA engineered it, and RSA was reportedly paid $10M to make it the default in its BSAFE library.
  • NIST withdrew it in 2014.

That’s exactly the “fixed number” pattern you describe, but in a different algorithm.

Other things people cite:

  • NIST P-256 and other elliptic curves were generated from “seeds” (hash inputs) that were never explained. That’s suspicious but unproven, and it’s why many people prefer Curve25519, whose constants are explained.
  • SHA-0 → SHA-1 (1995): NSA made an unexplained one-line change. It later turned out to fix a weakness that academics only found in 1998.
  • SHA-256 itself is NSA-designed (2001) and has had more than 20 years of heavy public analysis. The best published attacks only work on reduced-round versions (about 31 of 64 rounds) and break nothing practical. Bitcoin puts trillions of dollars of incentive on breaking it.

Realistic risk of NSA sabotage lies in how algorithms are chosen and implemented (random-number generators, curve parameters, standards defaults), not in SHA-256’s constants.

Hashes with no NSA involvement, SHA-3/Keccak (designed in Belgium) and BLAKE3.

NIST a “Captured Agency” ?

Why would they release as “standard” something having known weakness?

NIST was involved in both cases, but the constants in each came from NSA.

SHA-256 (SHA-2)

  • Designed by: NSA
  • Standardized by: NIST (FIPS 180-2, 2001)
  • Constants: square and cube roots of primes; anyone can verify them

Dual_EC_DRBG

  • Designed by: NSA
  • Standardized by: NIST (SP 800-90A, 2006)
  • Constants: unexplained points P and Q; the backdoor

P-256 and the other P-curves

  • Designed by: NSA (Jerry Solinas)
  • Standardized by: NIST (FIPS 186-2, 2000)
  • Constants: unexplained seeds; suspected, not proven

SHA-3

  • Designed by: Keccak team (Belgium)
  • Standardized by: NIST, chosen in an open public competition
  • Constants: derived openly

NIST’s failure with Dual_EC was process: it accepted NSA-supplied constants without asking where they came from, and kept the standard after researchers raised the issue in 2007. After the 2013 Snowden documents, NIST withdrew it (2014). It also reviewed every standard developed with NSA input and changed its process so constants must now be explained or come from public competitions.

The constants in SHA-256 can’t hide anything, whoever chose them, because they come from public math anyone can recompute. In NIST standards, suspicion should go to anything with unexplained parameters, which today mostly means the P-curve seeds.

Leave a Comment