Without going too much into practical implementation I would refer to the long time goals of Interledger
If Interledger is intended mainly to be used by banks, institutions, companies, then running almost anything shouldn’t be a big issue.
If also most of the end-users including lay mens - the “average Joe” should be able to comfortably use Interledger in casual life situations like paying for parking or at the Supermarket, then, usability-wise, it makes sense to simplify everything as much as possible (of course trying to keep it as secure as possible in the same time).
Going this branch, what is the level of control the users should have over their assets?
When I think about Interledger, sometimes I make analogies and references to classic banking. My money are in banks. The banks keep them - they are “the Ledger”. But they also operate them, they send and receive payments on my behalf. At least in practice, I fully trust the bank, which can in theory also freeze and/or confiscate my money on government or police orders, in case of a financial crisis, etc. The advantage is that I don’t care about operating anything but at most an app on the phone, which I start and stop anytime I want. The app is secure and easy to use by an “average Joe”.
As such, if the intent is that average people and casual situations should become a significant chunk of Interledger’s user pool, then, in my opinion, simplicity, user experience, comfort, ergonomy would be a relevant concern, besides security
People like simplicity and mobility, in my opinion many would be reluctant to starting computers, installing and configuring Moneyd, SPSP or equivalent, keeping them online, etc. Although we can imagine a phone app connecting to a system running at home, use cases would still be more limited in my opinion.
Perhaps pre-packaging everything and offering an (almost) automated deployment in some secure authenticated (cloud) environment with minimal setup would be a way to go? In my opinion, the casual user doesn’t need to know almost anything about the tech behind, unless they want to; My feeling is that it would be preferable for them just to subscribe, enter some initial minimal config and be ready to go on their phone or some minimal thing as a PC browser. Like subscribing to email or to a bank. Such a cloud environment could be ran/managed/supervised by Ripple, possibly? The same way Google manages the servers hosting the user email accounts; a company can have an email server, but casual users just have a light phone app or use a PC browser. These are money not emails, so possibly extra guarantees might need to be added, up to backing up and ensuring, globally or individually the cases where users would place extra trust in a third party (cloud hosting for example).
If everything would run somewhere else, a user could only run a light app with minimal config on his phone/PC (browser) from where he could authenticate, check and submit transactions in different forms - NFC, QR, manual, etc. I think this kind of simplicity and lightness in everyday usage would truly make ILP useful in casual life situations like paying a bill online or at gas station, restaurant, supermarket; and consequently, penetrate all market segments/regions and become global.
Indeed Interledger is about Money so this is not so easy. These are just some random thoughts maybe out of place, maybe said or thought before,. I just felt it might worth sharing, though