Dominik Wujastyk's blog
I mark conjectures with a wavy line underneath so no mistake is made that the conjectured word or phrase is a solid reading of the text. I also question the claim that 95% of conjectures are wrong. How wrong are they? Do they simply not get the right word or do they skew the entire sense of the passage? If they help to understand a passage that is otherwise non-sense and turn out to have been approximately the same idea of the author, I would not count that as a failed conjecture at all.
You would need to read the Dawe and Boeckh references that Kenney gives, and his earlier passage about papyri. You make a good point about reconstructing the author's intention. But a text is made of sentences and words, and if the words are wrongly conjectured then it's really not what the author wrote, however plausible or useful the conjecture.
I was about to write something similar to Michael's point: where does the estimate 95% (or 99%) come from? And what does it mean? I agree with Michael that if the text says, e.g., "darśapūrṇamāsābhyāṃ svargakāmo yāti" and I restore it as "…yajati" (although the original text would have been "…yajeta"), I am still doing a good service to the text and to its author. After all, philosophical authors might be happy enough to convey a meaning (instead of the traded non-sense), even if not in the exact form they chose.This being said, I agree with the general principle that one should ponder each conjecture for a long time. In my own limited experience I have sometimes had to go back to the text I thought was faulty after having better understood it in the light of other parallel texts.