it seems that awareness of the need to introduce economic incentives for services provided by full nodes are rising among devs community, see this thread on the btc dev mailing list:
http://lists.linuxfoundation.org/pipermail/bitcoin-dev/2015-August/009879.html
let's see how core devs will react to this idea.
How would this work if you issue (say) one txn per day on average? I mean, wouldn't it take a month or even a year before the SPV client summed up enough use to make it worth committing the txn? Are there risks to having a payment channel or LN connection open for these durations?
I agree that implementation of such payment system has to be carefully crafted. that said I'm not knowledgeable enough to evaluate which solution are better. (1)
I just want to say that even at current level of adoption there's enough SPV traffic to support a market based on supply and demand.
e.g. 1 txn per day per SVP client, supposedly half a million of active SPV clients would generate:
500000÷(3600×24) ~ 5.7tx/s
clearly we are not there since 5 tx/s per second is above the actual network capacity, but it is not so far fetched to estimate at least 500K SPV clients once we gain a wider users base.
(1) fwiw I'm following LN mailing list and it seems to me that keeping a channel open even for long period of time is not impossible.