Imagine you were eating at a restaurant, you ask for roast chicken, and they told you it would be several hours because they don’t start roasting until someone orders it. Unless it has a million Michelin stars you’ll likely walk out and go somewhere where they’re more prepared.

With software it can be tempting to wait until something is fully specified before starting on it. We’ve all worked on projects where the requirements changed during the process making some of the work already done useless. Unless it’s a complete change of direction there are usually things you can identify that have the most chance of being the same or easy to change as you iterate and learn.

You should get started on these when you can, even if some of them turn out to be useless. Being able to deliver earlier is usually going to have higher value, so even if you end up doing a bit of wasted work on the way it’s worth it. If you’re really not sure then refactoring the existing code to make it easier to change will at least give you a better chance of being able to hit the ground running when you do decide to start.