When folks speak hypothetically about speed, they almost always assume identical outputs. Which is pretty contrary to at least my experience, if not reality.
Apple is the worlds most valuable company, and they're first and fastest in essentially nothing. So it really depends on the market you're going after.
Anecdotally, (and backed by language learning research iirc), when learning is involved iterations are important. Learn a little, practice a little, get feedback a little, repeat frequently.
But afaik it all comes down to the cost of failure. Public iteration on your encryption algorithm that's guarding billions of dollars is probably a bad idea.
Apple is the worlds most valuable company, and they're first and fastest in essentially nothing. So it really depends on the market you're going after.
Anecdotally, (and backed by language learning research iirc), when learning is involved iterations are important. Learn a little, practice a little, get feedback a little, repeat frequently.
But afaik it all comes down to the cost of failure. Public iteration on your encryption algorithm that's guarding billions of dollars is probably a bad idea.