...A 25% chance of failure per encounter is challenging. However, it's not a cumulative factor ensuring a failure every four encounters.
Ensuring failure, no. It means about a 70% chance of at least one failure in four encounters, however. One "adventuring day" in three you get to not either die or run away. Sounds like a blast... but count me out.
Arguing that successive encounters weaken characters is beside the point, proper encounters are designed with that reduced condition in mind. Assuming that the successive encounters fall on the same day.
This has nothing whatsoever to do with weakening characters or reducing resources - it's just a consequence of the way stacking probabilities work.
..Fleeing an encounter is also a failure. It's extremely limiting to state that death is the only option to define failing an encounter.
So, what percentage will be "death"? One in three? One in four? That would make your 25% failure case a world where the average life expectancy of an adventurer is ~2-4 days (8-11 encounters), depending whether it's 1 in 3 or 1 in 4 failures = death. By "average life expectancy" I mean that each character has around a 50% chance to last this long.
...If adventurers are guaranteed a 99% success rate in every encounter, a failure rate of 1% or lower, you effectively create encounters in which no pc can ever lose or die. Such campaigns do not last.
Well, pretty obviously, campaigns where characters have an average life expectancy of under a week do not 'last', in the sense that it's a bit marginal to call them "campaigns" when the list of PCs is that unstable.
As for "never losing", this is incorrect, to put it politely. In a game (story arc, mission, adventure series, whatever you wish to call it) of 100 encounters, a 99% chance of success gives a ~37% chance of at least one failure. 99% is an exaggeratedly high figure to use, but even then to say it gives "no chance of failure" just flies in the face of probability theory.
...IME, it's a good idea to find out how tough players like their encounters. Which is what I am trying to do, in a more general community sense, in this poll.
Then here's some advice:
1) Make your agenda at least slightly less transparent. Your biased wording and selection of "success chance" points makes the view you want to push abundantly clear and your motives thus suspect.
2) Be clear about what you actually mean by your question. Poorly defined terms not only invalidate the answers you get (because nobody knows quite what question they are answering) but also make it look very weaselly when you start squirming about what the terms mean when challenged. It's far better to be clear about what you mean from the get-go.
3) Think in advance about what the options you are offering actually mean, and why they might be attractive. This links into (1), above, in the sense that you can only describe options neutrally if you first have some idea as to why such an option might be attractive to voters. If there's genuinely no way it could be attractive, as far as you can see, don't offer it as an option. The comments might show you that some folk do, indeed, like options outside your range - but the lack of a poll button means they might tell you why they do so (and thus educate you).