Suggest: "Focus on the User (meaning reader)."
I realize this is closed now. But this comment here is perhaps the clearest illustration of the problem:
"British culture used middle names from the mid 1800’s which are based primarily on the surname of ancestors, this practise has given genealogy a huge ability to research families back further generations with great success."
There are so many issues. It is respectfully not true the "primary" purpose of middle names is preserving ancestral surnames. Even within "British culture" whatever that is; but especially beyond that obvious limitation or bias. This is simply not a British website. Nor American. Nor for and by only these users or ancestors. It's WikiTree! This is self-evident in that even "British culture" is massively diverse and admixed with dozens or hundreds of non-British names and origins. Always was. Always will be.
So:
Many families use middle names exclusively for extra given (not sur) names. Others to preserve ancestral given (not sur) names; or immediate de facto patronyms, matronyms, or skip-generation "collateral" matrrnal-line patronyms all with given names (not sur). Yet others but especially Catholics concatenate Saint names, in ways that have nothing to do with genealogy pedigree.
Ultimately the "objective" answer to the technical question probably inhabits a more "subjective " arena, not code but rather rules: protocol & hygiene policy, for we logged-in users?
My own personal preference is: WikiTree probably wants more and more-structured database fields and data-entry name field forms in the editing interface, capable of perpetually storing different types of narrowly defined (ie discretized, unlike) middle names. Which vary and multiply by the # of cultural tradition. There will be more not fewer as the site attracts global users from more cultures, and they do their work on the universal tree. But this comes at the cost of forcing every user of the site to suffer thru scrolling thru more awkwardly complicated UX/UI, and being forced to repeatedly make decisions where to type what; complexity increases errors. The more fields or rules we have the MORE wrong-ness we harvest, not less. So what kind of problem do we prefer? Quality control cleanup labor hours, or technical simplifications which risk or actually do destroy meaning and useful nuance?
There is no one perfect solution. No one database storage & data entry regime for "Middle" names will elegantly accommodate both Welsh or say Nordic patronyms, Spanish matronyms, post-medieval manorial or noble placenames and/or French dit-names; while also effortlessly handling Catholic, German or pre-American compound given names with or without hyphens. It is just irreducibly complicated. Trying to over-simplify means creating new errors or failure modes.
Lastly the seemingly best choice is of course all a "function" of which centuries, cultures, and countries one ie we are working in. Not just personal preference, but research habits and context. For my own work here, due to my ancestry, I routinely have to teleport between centuries, cultures, languages and naming conventions. So I see lots of problems in site usage or data quality, but am also not suffering the exact same problems constantly; which I assume some folks here are, due to the nature of their personal research context. Overall I think we are better off with more not less: more discretization of name fields rather than merging information previously, intentionally separated.
But that comes at cost of "activist" use-policy and therefore ultimately the major project teams end up needing to make these decisions for themselves, or arguing to resolution between themselves. The database as "substrate" for the major project teams has to weigh irreconcilable preferences of multiple very large groups of users. It may be impossible to optimize a decision like this (middle names) without stating a larger goal and having accurate data on progress towards that goal compared to site usage and growth stars. For example, do we want to prioritize recruiting more new users ; retaining existing superusers ; adding more living family groups ; or correctly data quality problems at the "end of the line" from malplacements and GED upload fictions ingested a decade ago. Those prioritizations slice thru and confound the seemingly simple question about middle names and a large number of similar questions.
End of the day, we should do what's best for our readers. Not our editors. It's OK for the work of using the site to be slightly harder or even sub optimal, if the results are improved for posterity.
"Focus on the user."
Respectfully,