This isn’t some obscure niche case either, it happens every day on HN.
I _knew_ somebody was going to nitpick that phrase and rejected the urge to write a few extra paragraphs in the format of a HN comment expressing the idea more completely...
If we want to look at it a bit less charitably though, even things that are true "as far as anyone has been able to prove" are usually not perceived to be true for a significant fraction of the human race, so in interactions with randos (like this current thread) it's still reasonable to assume that we're talking about opinions rather than facts. Even when talking about "facts" (as you partially defined them), the core idea still holds -- your _opinion_ (maybe a very well validated opinion) is that those facts are true, and hedging language is discourteous to the reader.
Seemingly random claims, that no one has been able to provide proof for so far, but the person has some special reasons to strongly believe it is true.
E.g. ‘Back channel’ communication in diplomacy.
We find [strongest result]
Our findings strongly indicate [likely result]
Our findings suggest/indicate [result that needs more research]
The way I and many other people read papers is that we read method and appendix last, good writing can give you hints at those things without needlessly bloating the paper. The nuance also makes me trust the authors more, as it indicates humbleness and diligence
On some level, that makes even more sense in scientific writing. You already have method sections, experimental results, .... If you haven't been overconfident in those sections then the reader really ought to have all the tools they need to appropriately examine the paper.