So, what I'm hearing is:

 * lightning works great, but is still pretty small
 * zeroconf works great for txs that opt-out of RBF
 * opt-in RBF is a pain for two reasons:
    - people don't like that it's not treated as zeroconf
    - the risk of fiat/BTC exchange rate changes between
      now and when the tx actually confirms is worrying
      even if it hasn't caused real problems yet

Maybe it would be productive to explore this opt-in RBF part a bit
more? ie, see if "we" can come up with better answers to some question
along the lines of:

 "how can we make on-chain payments for goods priced in fiat work well
  for payees that opt-in to RBF?"

That seems like the sort of thing that's better solved by a collaboration
between wallet devs and merchant devs (and protocol devs?), rather than
just one or the other?

Is that something that we could talk about here? Or maybe it's better
done via an optech workgroup or something?

If "we'll credit your account in BTC, then work out the USD coversion
and deduct that for your purchase, then you can do whatever you like
with any remaining BTC from your on-chain payment" is the idea, maybe we
should just roll with that design, but make it more decentralised: have
the initial payment setup a lightning channel between the customer and
the merchant with the BTC (so it's not custodial), but do some magic to
allow USD amounts to be transferred over it (Taro? something oracle based
so that both parties are confident a fair exchange rate will be used?).

Maybe that particular idea is naive, but having an actual problem to
solve seems more constructive than just saying "we want rbf" "but we
want zeroconf" all the time?

> > > Currently Lightning is somewhere around 15% of our total bitcoin
> > > payments.
> > So, based on last year's numbers, presumably that makes your bitcoin
> > payments break down as something like:
> >    5% txs are on-chain and seem shady and are excluded from zeroconf
> >   15% txs are lightning
> >   20% txs are on-chain but signal rbf and are excluded from zeroconf
> >   60% txs are on-chain and seem fine for zeroconf
> Numbers are right. Shady is too strong a word,

Heh, fair enough.

So the above suggests 25% of payments already get a sub-par experience,
compared to what you'd like them to have (which sucks, but if you're
trying to reinvent both money and payments, maybe isn't surprising). And
going full rbf would bump that from 25% to 85%, which would be pretty
terrible.

> RBF is a strictly worse UX as proven by anyone
> accepting bitcoin payments at scale.

So let's make it better? Building bitcoin businesses on the lie that
unconfirmed txs are safe and won't be replaced is going to bite us
eventually; focussing on trying to push that back indefinitely is just
going to make everyone less prepared when it eventually happens.

> > > For me
> > > personally it would be an easier discussion to have when Lightning is at
> > > 80%+ of all bitcoin transactions.
> > Can you extrapolate from the numbers you've seen to estimate when that
> > might be, given current trends?
> Not sure, it might be exponential growth, and the next 60% of Lightning
> growth happen faster than the first 15%. Hard to tell. But we're likely
> talking years here..

Okay? Two years is very different from 50 years, and at the moment there's
not really any data, so people are just going to go with their gut...

If it were growing in line with lightning capacity in BTC, per
bitcoinvisuals.com/ln-capacity; then 15% now would have grown from
perhaps 4% in May 2021, so perhaps 8% per year. With linear growth,
getting from 15% to 80% would then be about 8 years.