RIP and UDP Port 520

Here's a somewhat basic and probably dumb question - but here goes!

I know that Ports are used to separate different "conversations" so packets are directed to the correct application or process. As far as I understand, RIP is a Layer 3 protocol. It's also described as using UDP Port 520 (hxxp://www.iana.org/assignments/port-numbers). I'm not aware of any other Layer 3 protocols (ARP, RARP, ICMP) having a port associated with them, so why does RIP?

As RIP is a Layer 3 protocol, surely the information contained in the PDU travels down the OSI model, having headers added as it goes and then headers are stripped off at the other end as the information travels back up through the model? If it works at Layer 3, I can't understand how it will ever get to the Transport Layer where the port numbers would have relevance.

There's obviously something wrong with my logic and I'm missing something - but don't know what!

Thanks for your time and patience.

Should answer your question www.tcpipguide.com/free/t_RIPGeneralOper...ssagingandTimers.htm

Thank you for the rapid response. It's interesting that there's the comment on the site:

"Thus, even though RIP is considered part of layer three like other routing protocols, it behaves more like an application in terms of how it sends messages."

This concurs with my understanding. I suppose a supplementary question is why the decision was made originally that routing protocols should be Layer 3 pretending to be Layer 7 rather than actually Layer 7. I realise that the answer to this might not be easily available but such cock-eyed logic doesn't make it easy for newcomers like me! I dislike simply learning things A --> B --> C but prefer to know why A leads to B leads to C.

Interesting point. And this is probably not a very interesting answer :

I guess, since RIP is a routing protocol, placing it on top of layer 3 will raise the argument (among designers/engineers) that RIP needs another routing protocol for it to be routed it self. Although it obviously uses broadcasts. On the other hand they used a UDP port. And they could have also used a flag or separate field just to indicate to routers that it is RIP.

I accept it as fact (even though I hate to do so!) but still don't see the logic in a Layer 3 protocol having to go up a layer to use a UDP port. It flies against everything that I've learned about how the OSI model works. It's rather like me going from Texas to New York to get to Mexico.