That Hadith are unreliable—that any given matn cannot be taken at face value as an accurate datum from the 1st Islamic Century, and that any given ʾisnād cannot be taken at face value as an accurate record of a matn’s provenance—cannot be seriously contested, for multiple reasons.
Firstly, there is an overwhelming prior probability based upon the ubiquity of fabrication and pseudepigraphy in Late Antique and Mediaeval religio-historical (pagan, Jewish, and Christian) ascriptions.
Secondly, there is the high frequency of contradictions within the Hadith corpus, which necessitates the occurrence of a huge amount of fabrication, interpolation, and/or mutation and, therefore, skepticism towards any given hadith.
Thirdly, there is the ubiquity of fabrication and interpolation—both reported and demonstrable — in the Hadith corpus, which again casts doubt upon the rest of the corpus.
Fourthly, there is the rapid, extreme mutation and growth of reports that evidently took place over the course of a century or more of oral transmission, which means that any given matn—regardless of the ʾisnād—is likely at best heavily distorted and at worst obliterated beyond its original form.
Fifthly, there is the belated emergence of Hadith as a genre and corpus, largely during the 8th and 9th Centuries CE, which straightforwardly precludes the authenticity of most ascriptions to the 7th Century CE.
Dr. Joshua Little, PhD Thesis
It is available from his own site, or from here: