I suppose we should clarify failure. For me your question sounded like a measure of complete or at least significant failure of the seal.
However there's also failures that are potentially more contentious
- Reductive aromas (personally I've yet to experience this, but for some people it's an issue). This is pretty contentious and I'd rather we didn't get into a debate about it
- Other winemaking faults: Screwcapped wines can still be tainted by TCA (when present in barrels or winery), Brett, plus various other lesser problems.
The reason I mention this is I think it's fair to compare not just failures, but magnitude of failure. Some TCA in cork sealed wines could feasibly have come from the barrel, not the cork. A very small %, but let's have a clean fair fight
. Also some failures such as reduction are very minor, and indeed lower levels of TCA will only slightly change the flavour profile of the wine, which will remain drinkable.
What I'd love to see, is a single study (or more realistically a collection of studies) where the results are organised something like
Fault Undrinkable tolerable barely perceptable
TCA 8 4 7
Brett 5 9 2
Oxidised 5 1 0
Based across various wines/seals
At this stage, such a study is both pie in the sky and also not particularly scientific as it still relies heavily on human interpretation. It also could only be a snapshot as methods are changing every year.
Maybe one day we'll get something balanced and scientific, but I think we're most likely to end up relying on limited personal experience and personal perceptions. My perceptions currently favour Screwcap over cork with cork variants (e.g. Diam) coming up on the rails - I'm yet to feel comfortable with this latter category yet.
Having said all that, I am happy to see more and more results published, even if that means accepting human variation that is implicit in them.