As is known, the majority of hadith literature available today was compiled well after the death of the prophet, due in part to the nigh-unmanageable number of questionable or downright fabricated ahadith which were being narrated. The hadith sciences were developed, giving scholars criteria with which to distinguish between authentic ahadith and weak ahadith.
Take, for example, Sahih Bukhari. Bukhari was notoriously discriminating, and went through significant effort to collect and compile only the most authentic ahadith that he could; for the sake of argument let's assume that he was successful and that this collection was 100% authentic.
However, this merely determines that these collected ahadith were authentic at that particular point in time. Given that these books were written centuries before any form of "perfect" scriptural reproduction was available, any copies made thereof would have to have been transcribed manually. And without any original manuscripts to refer to, any current references would be based upon these copies, or copies of copies.
Due to simple human error, every manually-copied edition increases the chances of mistakes being introduced into the text, be it intentional or accidental. To me, this parallels very strongly with the earlier hadith sciences, wherein determining the authenticity of any particular hadith requires a thorough understanding of every narrator in the isnad. A hadith is only considered sahih if each and every narrator in the isnad is considered reliable.
The only difference in the case of transcription is that this "narration" is transmitted textually, rather than orally.
No hadith text I have seen lists a "scribal isnad;" the chain of narration, such as it is, tends to end with the fact that it was compiled by Bukhari. Given that any scribe between the original compilation and the invention of perfect digital copies may have (inadvertently) changed the text, what we have today as "Sahih Bukhari" may or may not be the original text compiled by Bukhari.
I realize that just because I've not seen such a "scribal isnad" does not mean they do not exist (limited to English as I am, many ahadith do not even bother to translate the normal isnad), and I also realize that a number of sahih ahadith would've been recorded during that time in disparate enough works that they could (by the same parallel) be considered mutawatir, but I know little (if anything) about the sciences of hadith which specifically deal with ahadith after they're compiled. Do the extant hadith sciences account for such (potential) scribal errors, or must we resort to textual criticism to determine authenticity?
To put it another way, if (for example) Imam Bukhari were alive today, would he call any hadith, even those compiled in his own Sahih, "sahih"?
(note that Sahih Bukhari was used as an example, but the question is applicable to any hadith which is called "sahih," regardless of where it was compiled or who authenticated it)