This is true in Manager, but not necessarily in other systems.
To give a specific example, GnuCash’s statement import has surprisingly good matching of imported entries with manual entries. So if I’ve been entering information manually, but missed 5 or 6 transactions, the best way to correct the situation is to import a bank statement; GnuCash will happily match all the transactions that already exist (and not double-import them), and then the only things that get imported are the 5 or 6 transactions I missed. It’s the best of both worlds!
If I try this in Manager, then seemingly every transaction on the statement is imported and I have to go through and manually delete them one by one. So while we can pick at the details of the proposed implementation, the basic idea of “Hey, recognize when a transaction has already been entered, and don’t import a duplicate” is a very reasonable suggestion and one that deserves consideration. (The Guide claims that Manager recognizes duplicate transactions but I did an import tonight and I don’t think it caught a single duplicate out of 50 or so, probably because the matching conditions are too stringent. So perhaps what I’m really saying is “duplicate detection should be better?”)
Anyway I’m on this thread largely because I am really enjoying Manager but feel that the reconciliation flow is noticeably more clunky than many other parts of the system.