Oddbean

i'd say stick to the 5 hours and keep your effort as an OTS backbone; hopefully OTS will at some point catchs some usecase for a party for which the cost are marginal enough they will set up a calender that stamps more frequently.

Note that the OTS infrastructure works better if there are fewer calendars, not more (to a point at least). The issue is that OTS clients create incomplete proofs by default, to avoid waiting for a confirmation. So to verify a proof, the calendar data needs to be available. More calenders means more data for the whole community to backup, and more chances something gets lost. If someone, eg, wanted timestamps once a block it would be much better if they just paid me to do timestamps once a block than setting up their own calendar. Which incidentally is exactly what was done for the Guatemalan election project: the company doing it paid me a few hundred dollars to pay for fees of doing almost constant timestamps.

Ok let me rephrase it; ideally at some point enough interest (be it commercial, or public, academic or whatever) is there for some entity to run a commercial/'professional' callender where businesses, states, whomever, pay that entity for this service. Which subsequently leaves your efforts as a 'community' backbone. Or do you see this differently? Sir, are you mad for power to become the sole timestamping overlord of the universe?

The nature of time-stamping is that it doesn't make sense to duplicate efforts beyond what is necessary for failure tolerance. Indeed, I should add a pay-with-lightning scheme to trigger an immediate time-stamp transaction for anyone willing to pay for it. The calendar server software is of course open source, so it's trivial for others to run their own calenders. But there isn't a good reason to do so.