Comments on NIST Interagency Report 7977, Draft 2

To: crypto-review@nist.gov

Cc: crypto@nist.gov

Re: NIST Interagency Report NISTIR 7977 “NIST Cryptographic Standards and Guidelines Development Process” January 2015 Draft.

Date: March 24, 2015

The January 2015 Draft is a big improvement. In addition, the Summary of Public Comments on NISTIR 7977 (Feb 2014 (First) Draft) was quite helpful to compare how the current draft addresses comments on the first draft. The authors of both new documents are to be commended for their work.

General review comments:

A. Review comments on the first draft of NISTIR 7977 reflect a distrust of the NSA, and by extension of NIST processes. Suppose China sends NIST an email that says, “Hey, we’ve been studying elliptic curve E over field Z/p with base point Q. It has some neat properties, and China recommends that NIST adopt E for its elliptic curve encryption.” Should NIST consider China’s suggestion? A good test of the processes in NISTIR 7977, would be to answer “yes”, even though we suspect that China knows a backdoor to encryption that uses E. Now, receiving such a suggestion from the NSA should be little different. Even though NIST is required to consult with the NSA, most independent security consultants, post-Snowden, would not trust a curve suggested by the NSA any more than one suggested by China. I certainly would not. Therefore we look for processes in NISTIR 7977 that politely look with great suspicion on suggestions for tools and algorithms from the NSA. NIST should get expertise and knowledge from the NSA, but should not blindly accept its tools and algorithms. NIST processes for analysis and acceptance of any suggestion, be it from the NSA, from China, or from any other source should be equally stringent. In particular, cryptographic technology should not be standardized without significant consensus from a wide swath of independent cryptographers. The current draft of NISTIR 7977 does not have an emphasis on public analysis and consensus for acceptance.

B. Of course, NIST could standardize, say an encryption method, even from a private and independent source, and the NSA, China’s PLA Unit 1398, the UK’s GCHQ, or other nation-state cryptography group could know how to crack this encryption method, remaining silent during and after the standardization process. One would hope that NIST, through its own efforts and its financial support of academic and other cryptography research would facilitate the discovery of the weakness and would quickly retire the standard. Such post-standardization life cycle efforts by NIST also need to be part of NISTIR 7977 [line 727].

C. Now if NIST were to publish a proof that a tool or algorithm received from China, the NSA, or another source in fact had no back doors and was totally secure then a consensus on standardization might well be easily achieved. After believing such a proof, I might recommend the proven technology to my clients, but I probably would wait until a huge number of non-government cryptographers also believed the proof and also were recommending this technology. NISTIR 7977 says roughly that NIST will “pursue” to find and use proofs [line 55]. I’d be happy if NIST worked with the NSA and other agencies on such proofs and would recommend such efforts be part of NISTIR 7977.

D. NIST publishes security papers at a prodigious rate. So fast that reviews are deemed inadequate. In light of post-Snowden caution around NIST processes, people naturally ask if these poorly reviewed papers can be trusted. It isn’t going to help if NIST says, “It’s ok, the NSA has reviewed it…” Look, not only does the current draft of NISTIR 7977 fail to convince that future NIST papers will receive good independent reviews, there was no indication that past NIST papers will retroactively receive good reviews. This is a very sad state of affairs, but it is fixable.

Some more specific review comments:

  1. Clarity of the NIST Mission: To develop strong cryptographic standards and guidelines for meeting U.S. federal agency non-national security and commerce needs. This mission should be parsed: To develop strong cryptographic standards and guidelines for meeting 1. U.S. federal agency non-national security needs and 2. commerce needs. My point is that the needs of commerce should treated by NIST as equal to the needs of any federal agency. [line 202 Balance]. For example, federal agencies may well be happy with NSA algorithms, but general commerce may not be.
  2. I do not agree with the NIST Response (page 7 of Summary) to Technical Merit comments that NIST should give priority to non-national security federal information systems. NIST should always make commerce needs equally important. Such a priority statement doesn’t seem to be in NISTIR 7977 explicitly, but there are several statements about NIST being legally required to give such a preference when so ordered by a government entity.
  3. NIST’s statement that it will “never knowingly misrepresent or conceal security properties” [Summary page 3; line 215 Integrity] reminds me of Barry Bond’s statement that he “never knowingly took growth steroids” when his hat/head size at the end of his career was three sizes larger than when he was a rookie. I would prefer a more proactive statement such as “NIST will make every reasonable effort to ensure that military, intelligence and law enforcement agencies by their suggestions, review comments, or contributions do not compromise any security tool or algorithm recommended by NIST.” For NIST standards to regain the confidence of the security and general commerce communities, NIST processes should convincingly ensure by NIST public actions that its tools and algorithms do not compromise the privacy or the integrity of any commercial or private message being protected by NIST standards.
  4. The FISMA requirement that NIST consult with certain federal agencies including the NSA to avoid duplication of effort and to maintain synergy of federal information protection efforts, but NIST can never in the future blindly accept a recommendation from any public or private agency. What is important is that NIST actively regain and maintain its process integrity via the new NISTIR 7977. The current draft falls short.
  5. NIST should consider resolving the conundrum of needing NIST output frequently and needing adequate public reviews of such output by the creation of additional outside paid review boards and conformance testing bodies. Such a review board should be established to review annually the entire Cryptographic Technology Group. [lines 316 and 326]
  6. Minutes of the monthly NIST/NSA meetings should be published. [line 377]
  7. Independent review boards should have the power to reject a proposed standard, say if NIST could not convince the board that the NSA or another agency has not compromised the standard. [page 4; lines 47, 403, and 464]
  8. The NISTIR 7977 Development Process itself should undergo regular review and updates at, say, an annual frequency.

Requested Comments [line 125]:

NIST Question Comment

Do the expanded and revised principles state appropriate drivers and conditions for NIST’s efforts related to cryptographic standards and guidelines?

Yes, but if the word “appropriate” were replaced by “adequate” then No. Neither the integrity of NIST processes in face of NSA influence, nor the issue of adequate review are satisfactorily answered. [A, C, D]

Do the revised processes for engaging the cryptographic community provide the necessary inclusivity, transparency and balance to develop strong, trustworthy standards? Are they worded clearly and appropriately? Are there other processes that NIST should consider?

No. “Trustworthy” standards need public confidence that the NSA or another agency have not added or know of backdoors or other weaknesses to their contributions. Wording isn’t an issue. Different new processes are necessary to separate NIST from NSA and other related agency influence. After-standardization research efforts should be funded as part of all life cycles [B].

Do these processes include appropriate mechanisms to ensure that proposed standards and guidelines are reviewed thoroughly and that the views of interested parties are provided to and considered by NIST? Are there other mechanisms NIST should consider?

No. Cf. A, C, 2, 3, 4, and 7 above. Regarding 7, if NIST won’t vest veto power to independent reviewers, such experts will tend to not participate. Lack of review resources also seems to be a problem. Cf. 5 above.

Are there other channels or mechanisms that NIST should consider in order to communicate most effectively with its stakeholders?

Yes. More paid outside reviewers including an annual review of the Cryptographic Technology Group. Cf. D and 5 above.

Respectfully submitted,

Gayn B. Winters, Ph.D.

Technology Consultant

Advertisements

Tags: , , , , ,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s


%d bloggers like this: