But in that case, it would be best to do a subscription model of like $10-ish per year.

This would involve an agreement not to sell data, to collect only data with a use demonstrated to be critical to the operation of the service, and a plan to dispose of that data within X amount of time. This also needs a written contract stating that the cost of subscription won’t go more than X% above the user’s pro-rata share of the demonstrated cost of providing the service, consisting of certain very specific purposes (building, servers, ISP, employee revenue, etc.) to avoid cheating for more profit. A subscription service, protected against privacy infringement and price gouging (a profit limit).

If it’s ever going to work this would need to be a government-mandated privacy act. I usually hate government intervention, but this is very much a necessary evil to prevent price gouging.

I only suggest this over donations, because realistically, after upscaling to a global audience, only 5-10% of traffic would be users that choose to donate, increasing costs to around $10-20/month, which yet again lowers the number of people who choose to donate. It stabilizes, but at such a low percentage that it’s unsustainable at a large scale without millionaire donors, and a very small percent willing to pledge $20-30 ish per month throughout the entire product lifetime…