Stellar, technically, is a bunch of things; gateways, trust graph, etc. But in terms of how people use it, it's a single thing: a mechanism for sending real-world money over the internet. And it is real-world money. It's represented by credit, but most people are already operating on credit; they keep all their money in a bank, and the bank tells them how much credit they have, and when they pay for something the vendor accepts their credit and redeems it from the bank (typically by accepting credit from their own bank, rather than physically receiving paper money).
This does suggest to me, though, that it would have been nice if Stellar had used simple technology (just a distributed ledger, like bitcoin) to bring about the simple use-case (an exchange).
Perhaps that isn't possible, and Stellar truly is the best we could do. But I'm not convinced it is.
Bitcoin is a distributed trustless ledger tracking inputs and outputs of transactions. It stores it in a blockchain, and determines transaction values using a script that takes some inputs and produces some outputs.
---
As far as I can tell,
Stellar does many things (gateways, trust/debt graph, asset mapping, quality conversion between assets, stellar currency), on top of which people are trying to bootstrap one thing (a big exchange).
Stellar is a centralized trust-based debt graph of arbitrarily-defined assets hopefully backed by trusted gateways. It uses consumer-defined trust and quality (exchange rate) values per asset to other nodes which creates a debit cap for the peer.
The roles between the technology vs ecosystem of Bitcoin vs Stellar are inverted. Clearly Stellar's approach has some benefits (fast transactions, regulation-compatible gateway infrastructure) at some costs (trusted/centralized, bootstrapping risks).
This is about as good of an oversimplification as I can come up with right now.