It's like trying to decide how to build a system/business while also trying to work out what to build/what problem you're solving. At some point, to actually do anything, you have to pinch off the space of all possible problem requirements/business objective/etc.—not permanently, but long enough to start building! You certainly can't coordinate a large team to do it when the goal itself is changing. You have to draw a line around something and saying "we're doing this", and then focus on how to do it.
In ML terms: If you don't commit to a goal for a while, then every time you encounter uncertainty in whether you're accomplishing your current goal, it will "backpropagate" to increase uncertainty in the goal itself. This can easily become unmanageable. To make any progress, you need a series of forward passes (attempting to do things) and backward passes (revising your idea of what you're trying to do.) Balancing these is a skill, and is highly contingent on what you're doing, runway, etc.
Applied to debating with critical theory, this means: if you constantly cast doubt on the entire framework of modernity, you can never come up with a useful answer about any particular thing in the immediate present. Every potential harm or pain backpropagates to indict the entire system. These harms are rightly considered indictments of the system—but in the real world, unlike in a debate, you still have to act.
I haven't done debate myself, but these same problems come up in internet discourse all the time, and it seems clear that you need to some equipment to handle this tendency to make any progress. What's necessary is the ability to designate "responsibility", i.e. leadership.