Proof of Work
In a sentence, "If it's good enough to be peer reviewed and put on a retail wiki as fact and not speculation", it's more than good enough. You've done a good job.
Ideally, you would have a complete packet capture from retail of whatever it is you're implementing, fixing, or updating. This is very important, as sometimes there are conflicting messages that appear to have the same text but aren't identical IDs. For all we know, the text is different for the Japanese client and could be wrong. 
is not "maybe" to the Japanese client.
That's fine. We know it's impossible to capture everything all the time. What is important is documenting what you don't know. TODO: is not a curse, it helps us know what we need to capture the next time around. We encourage TODO: within reason, there are many things that can't be known easily with gargantuan levels of effort required to verify. If something looks like a 100% proc rate but you can't spend weeks gathering the data to figure it out? Put a TODO: on it, but we do expect a reasonable effort. 5 samples of something isn't enough.
No problem. Mark it as TODO: with an applicable message indicating what needs testing. It's not your fault something is what it is, the feature might have changed since it was developed or wasn't tested well enough, possibly even more research revealed that we always had a poor understanding of it. Be sure to reference the most up to date and reliable wiki updated by players who still play the game.
You'll need to do your own research or make a discussion about getting research. Note that partial implementations are OK, but you can't just make stuff up. See #1442's foil implementation. Foil's mechanics at the time were unknown, but the spell is still useful as part of the game for enmity purposes.
Complete captures for the mission/quest you are implementing are required for correct cutscene events, and parameters. It may not be possible to capture all options on a single character on retail and that's OK. Remember how many options there are to get access to ToAU? You'll need to test that on LSB, not on retail. Do note that wikis are often wrong and they sometimes tell you to skip things you don't need to. We expect you to check all the optional paths within reason. Captures also tell us about NPC spawn positions, models, etc. These are critical for a correct implementation of missions, quests, and battlefields.
Captures are required for messages, especially for action packets. Until recently, COR rolls were crashing clients due to the original implementation using only 2 message types, even though there were several more. It's very important to get things right, this way nobody has to revisit a broken implementation, only to change an implementation if/when it is changed in retail. For job mechanics that appear completely unknown and you want to test them, make a discussion! We have plenty of people who've contributed in the past that have tested obscure mechanics and their findings are now the de-facto knowledge of retail until proven otherwise.
How am I supposed to know the NM's level? Good question! It turns out the widescan packet has all monsters level embedded in it... including Notorious Monsters. This way there's no reversing or guessing and it comes straight from the servers.
Sometimes, videos are cited as evidence. Some videos are from retail and some are not. Often times on Youtube newer videos of "era" content is done on private servers, and it may be impossible to tell at a glance that it's not retail. Check the channels other videos, comments, video descriptions. Be particularly wary of words like "Classic", as there is no FFXI classic, only private servers that attempt to replicate that.
We want your contributions and if you need help, please make a discussion. Remember, a TODO: doesn't indicate failure to implement, it indicates that we need more information, a much deeper look or backend changes in core.