Reasons why it probably wasn't Hal Finney.
1. Satoshi emailed me at the same time Hal was taking part in a marathon. Yes, technically emails can be delayed, but ... why? There was no urgency in those discussions.
2. Hal preferred to write C and was still writing C up until the moment he sadly passed away. Satoshi wrote in modern(ish) C++. Note that the same problem applies for Len Sassaman and basically everyone who has been fingered as a possible Satoshi. The combination of C++, Win32 and cryptography is not very common in the cypherpunk space.
3. Hal was a professional cryptographer who worked on an encrypted messaging product (PGP). Satoshi had some fairly basic misunderstandings about cryptography, like believing that you can't use elliptic curve keys to encrypt messages, and referenced having to do research to understand the cryptography he needed. Why would Hal pretend to be a newbie at cryptography, up to and including not having features Satoshi said he wanted?
4. Hal did try to create digital cash already, but tried to build it on trusted computing hardware. He had a long term interest in that approach, as did I, and his last project before his death was one I suggested related to this tech. In fact there are emails between me and Hal in list archives pre-Bitcoin where we swap notes on Intel LaGrande. If Hal had been Satoshi it seems likely that Satoshi would have at least commented on the ways TC could be used at some point, but Satoshi never showed any awareness the tech exists.
6. If you're trying to be anonymous, why would you immediately and pointlessly reveal your identity by messaging yourself?
7. He denied that he was Satoshi.