I flipped through Mann's Tinseltown yesterday. The book is terribly padded--95% context, 5% mystery. As for Mann's new solution to poor William Desmond Taylor's murder, I suggest skipping the rest of this post if you want to avoid spoilers.
Ready? Spoilers ahoy!
Mann's solution ties the murder to Margaret Gibson's deathbed confession of killing Taylor. In his solution, Gibson didn't pull the trigger--she was partnered with a group of blackmailers, one of whom, Blackie Madsen, was directly responsible for the killing. It's a plausible solution, since Madsen and his accomplice seem to fit the eyewitness descriptions of the men who asked for directions to Taylor's bungalow and of the man seen prowling around there and leaving the residence shortly after the Taylor's murder.
The problem is there's no actual proof tying Madsen to Taylor. Nor was it clear how Gibson and company would have amassed the neccesary evidence for blackmail (I'll have to re-read some parts to see if they had contact with Taylor's ex-servant, who apparently was also a blackmailer). So while it's a plausible theory, it remains a theory. I'm also a skeptical of Gibson's confession, which could either be a old rogue's last swindle or the delusion of an otherwise normal person (just as otherwise sane people sincerely believe they've seen UFOs or ghosts).
And to be honest, Mann's solution is also a rather dull one. That's not a knock against its possible verity, since real crimes often have dull and/or anti-climactic solutions, but it did leave me nostalgic for the Charlotte Shelby theory, which has an emotional power and resonance that the stories of the other culprits (Madsen, drug hitmen, burglars) lack. That reaction is undoubtedly a tribute to A Cast of Killers, which wove a captivating mystery tale and shimmers with nostalgic glamor (King Vidor and Colleen Moore as detectives! Interviews with Alan Dwan and Gloria Swanson!). And since every theory has holes (if they didn't the murder would be considered solved), I'll stick with the Cast of Killers version (while admitting Mann's might be the right one). As Self-Styled Siren wrote, "whether or not Vidor had the right perp, it plays."
In any case, it would be terrific if the research King Vidor compiled could be digitized and placed online--you can bet the Taylorologists would have a field day. Those of us who care about King Vidor would have fun too. The Taylor research is presumably in the set of Vidor papers at the University of Southern California, along with a bunch of enticing unproduced scripts.