If you have a thing that will bite at some point, you have a choice to make.

Do you want to make it bite hard, at some indeterminate time in the future?

The bite, if it comes, will be hard, and possibly existentially dangerous.

But most of the time, you can simply not think about it.

Or do you want it to bite persistently but softly up front?

This makes it unlikely to cause unexpected problems down the line.

But it also means the user has to contend with a continuous, never-ending tax.

And also, maybe the user never would have gotten the hard bite in the first place?

The very first version of Google App Engine had a very odd development paradigm.

The framework and datastore felt wildly unlike the traditional LAMP stack at the time.

The benefit was that if you built your apps in this particular way, your app could scale to insane heights if necessary.

… But the vast majority of webapps that someone might build will never get more than a few dozen users anyway.

The tradeoff was to have a persistent, non-trivial tax, which made it less likely you'd ever bother to make anything worthwhile anyway.