I'm not sure if there is any historical evidence backing that up (IE, Tim Berners-Lee used Finger protocol as an inspiration. A lot of the UNIX protocols of the time were like that (NNTP in particular), simple call/response with textual commands and arguments.
I thought Finger itself was a copy of the Whois protocol, which runs on port 43. But that's backwards! According to https://datatracker.ietf.org/doc/html/rfc812, sri-nic.arpa supported Whois in 01982, saying, "The NICNAME[/WHOIS] protocol is similar to the NAME/FINGER protocol (RFC 742)," and https://datatracker.ietf.org/doc/html/rfc742 is from December 01977, explaining, "The FINGER program at SAIL, written by Les Earnest, was the inspiration for the NAME program on ITS."
Things like NNTP, SMTP, IRC, and FTP were pretty different from this family of protocols. They're textual, yes, but they're highly stateful protocols with lots of back-and-forth to get anything done. DNS, NFS, and SNMP (01988: https://datatracker.ietf.org/doc/html/rfc1067) were stateless, like Finger, Whois, and HTTP, but used optimized binary structures over UDP. In my view, though, the protocol semantics are a lot more important than its syntax, so I think HTTP is a lot more similar to DNS than to NNTP.
Later, FTP-like numerical status codes and long-lived connections got added back in to HTTP, but they weren't there in HTTP/0.9. Designed at the same time as HTTP, Gopher (port 70) was also a finger-style protocol, and I don't think it has status codes either.
According to https://www.iana.org/assignments/service-names-port-numbers/... the lowest unused ports are currently 4, 6, 8, and 10. I think these were originally the port numbers for the opposite direction of data transmission for services on ports 5, 7, 9, and 11 (RJE, echo, discard, and systat) but that function became obsolete with the switch from NCP to TCP in 01983. I may be misunderstanding this a bit because I don't really understand NCP.
