Are requirement analysis on-topic here?

Question

As we know, cryptanalysis questions are off-topic here - there's little value of a single decrypted plaintext to the larger community. (This excludes general cryptanalysis technique questions of course). But what about the opposite?

From my recollection, these types of question typically consist of a background of the application, some proposed scheme, and a request of analysis on aspects that concerns them, such as logical loopholes, alternatives' comparison, room for optimization, etc.

Often but not always, askers are building some E2EE chat or cloud storage application, which I believe we should point them to this as a friendly reminder of how fierce the competition in this market segment is.

• Specific instance of requirement analysis is probably off-topic in the same vein as cryptanalysis questions. However, I think

• General techniques for establishing requirement and refining them through thought experiment and test implementations can be okay for this site. Also,

• Some requirement analysis may receive definitive answers where a specific class of cryptographic algorithm can fulfil their need, although I think these may be so easy to answer that there're too many dups for it.

• Some requirements may already have existing solutions that the asker may not be aware of. These can be turned into reference-request maybe?

What do you think?

Answer 1

As we know, cryptanalysis questions are off-topic here - there's little value of a single decrypted plaintext to the larger community

Cryptoanalysis is generally thought of as the analysis of cryptographic schemes such as algorithms, not code. We've got another close reason for that.

Specific instance of requirement analysis is probably off-topic in the same vein as cryptanalysis questions.

Yes, mainly because they lead to opinionated answers, extensive discussion in the comments, re-edits and they won't result in objectively valid let alone complete answers.

General techniques for establishing requirement and refining them through thought experiment and test implementations can be okay for this site.

Sure, if the question is concise enough and can be answered objectively.

Some requirement analysis may receive definitive answers where a specific class of cryptographic algorithm can fulfil their need, although I think these may be so easy to answer that there're too many dups for it.

You'll see that many simple schemes will generally be treated relatively gently on the site, especially if an objective answer is at least possible. Answering them in their context and pointing to places where the issues are explained more deeply seems like a good strategy. Remember though that these questions are often not very good for building a DB of questions that are useful to others.

Some requirements may already have existing solutions that the asker may not be aware of. These can be turned into reference-request maybe?

I think the tags should mainly be put onto the question in such a way that it supports the question rather than the possibility of an answer. If there is an other Q/A that completely answers the question then it should be marked as a dupe. If it only covers part of it then it should be answered and the other answer can be used as reference. But I don't think it does transform the question into a reference request just because part of it can be solved using a reference.