Seems a little contradictory.
For example: "Constraints scale better because they don’t assume knowledge. They are adaptive. They respond to feedback. A small team that decides, "We will not hire until we have product-market fit" has created a constraint that guides decisions without locking in a prediction. A founder who says, "I will only build products I can explain to a teenager in 60 seconds" is using a constraint as a filtering mechanism."
I think sensible constraints are based on knowledge. Goals can also respond to feedback, not be indefinitely locked-in. But they do differ as tools.
The small team that decided to not hire probably created that constraint to get to some goal, e.g. profitability, and the constraint is based on a prediction about what should work best.
Similarly, the 60 sec constraint probably serves some goal. Why are goals so bad again?
The funny thing is, I think either goals or a constraint are a tool that should serve the user. Constraints that don't automatically allow the user to achieve goals they would have otherwise accomplished, and that are meaningful and important to them, are useless constraints.
I think figuring out the constraints one likes to work with can act as a great filter once someone knows what kind of success, goals, values and life they want to inhabit. Otherwise, it's as arbitrary as goal setting.
For me, I parroted other people's cool-sounding goals for a lot of my life, achieving varying degrees of success and happiness. Only in retrospect can I look at my favourite success and failure stories and consider which constraints, if I held them earlier, would have helped me narrow down to those favourite storylines from the get-go. Those constraints, I keep near and dear to my heart and attention in my daily life.
I don't think there's a way to set a meaningful constraint before practicing setting goals first. Walk before you run, etc. etc.