Knows about Jim Morris but didn’t get him in this particular line of questioning.
Not impressed.
It didn’t get Sonmi 451 from Cloud Atlas
It didn’t get Daneel Olivaw from Asimov’s Foundation series, Nor Giskard
It took 72 questions to get me
OK, if I were Akinator, I’d be embarrassed by this one: He couldn’t get the Tin Woodsman, even after 80 questions. But he did, somehow, guess the Tik-Tok Man and the Scarecrow. I’m baffled how he could have been thinking Oz, and knew “made of metal” and “appeared in a movie”, but not get the Tin Woodsman.
It knew Angus Young From AC/DC, Samuel Eto’o from Anzhi Makhachkalabut but not Ed Wynne from Ozric Tentacles.
Stumped it with a couple of Australian comedians. It thought Catherine Deveny might be Gina Rinehart (ha!) or Germaine Greer (ok, that’s close).
Also stumped it on John Clarke, who was in the database as “Olympic Games Organiser” 
I did the tin woodsman just yesterday and it got it in about 12 questions, on the first try.
That’s because you got the question about “Does your character like to use lube?”
Possibly there’s confusion in the database over what name to associate with. (The character name is the “Tin Woodman”, though he’s popularly referred to with the inserted ‘s’.)
I also noticed some people are trying to mess with it. I got the question “Does your character live in your country?” on the road to Samuel Pepys (45 questions).
Surprised it eventually got Junko Enoshima.
Stumped it with Jake Logan (Tachyon) and Katharina Bora Luther.
There’s also redundancy problems. I’ve had to answer some questions 4-5 times on occasion. I’m pretty sure Akinator is just a really simple tree, if you choose yes, it chooses one branch, if you choose “no” it chooses the other branch, if you choose “don’t know” it picks uniformly randomly, if you choose “probably yes” it chooses randomly weighted towards the yes answer, if you choose “no” it chooses randomly weighted towards the no answer. When it hits a node with a leaf attached, it guesses, if you say no, it continues along the path. If there’s nowhere else to go, it gives up.
It may be a little more complicated (i.e. not a binary tree, a few more nodes each time), but not much. Its seeming power comes from high userbase more than the technology, I’d wager.
Oops, got distracted and missed edit window:
Because of all this, even if you’re not deliberately screwing with it it’s relatively easy to mess it up just by the sheer intractability of maintaining a clean tree.
I stumped him with Clamantha from Fish Hooks. He did guess multiple other female characters from that show.
He got Perry the Platypus pretty quickly.
I tried a couple of the ones that people said stumped Akinator, and got them all in less than 20. I wonder what accounts for the discrepancy? I’m using the iPhone app… I wonder if it uses a different database?
Tin Woodman, 17 questions.
R. Daneel Olivaw, 15 questions.
And I just did Douglas Adams, correctly guessed in 16 questions. Weird!
Stumped with Imoen from Baldur’s Gate
Stumped Akinator with Bobbie Jo Bradley of Petticoat Junction – guesses included Alice Kramden, Thelma Lou, Helen Crump, and Mary Ann Summers.
Satchel Paige was guessed on Question #18, though.
Just now stumped Akinator with 5 95472 after guesses of other Peanuts characters (Charlie Brown, Linus, Pig-Pen) and one name I didn’t recognize (should have written it down) from a Japanese cartoon. The database did list my character by his full name, 555 95472, but victory was nevertheless mine!
I stumped him with Petey Otterloop. Then tried Alice and he got that.
He had no idea who Amanda Huggenkiss was.
It got the Fruit Fucker from Penny Arcade in about 40 Qs. Surprised it didn’t censor the name.
Yep, that’s what I’m thinking, too. After it got Noam Chomsky on the second guess, my daughter came in to ask what I was doing, so I showed her, using Mercy Watson (a pig from one of her favorite series of children’s books) as the example. Akinator was stumped, but in the end it asked me for clarification, and I showed it two names in its database with the same meaning; my guess is that now, if someone tries Mercy Watson, it’ll compare your answers to mine and be able to get it right this time.