If we have at least two weeks of ready ready backlog of meaningful work to create and we have vetted that backlog within the last couple of weeks, why should we validate it one more time? Ideally you are in a continuous deployment cycle of build, measure and learn with learning and feedback happening daily. And this fast feedback loop is allowing you to optimize your plan as it unfolds. This means that what was assumed to be correct two weeks ago may be wrong today, or done differently today, and that the backlog that is only a week old may no longer be right. To make sure that nothing has changed or been learned that could invalidate the assumptions for a card in the backlog the person who picks up a card to work on should at the same time grab their next card, make sure they understand it completely and that it makes sense to them, and they should add to the card how they are going to address the problem. What their implementation strategy will be. Once confirmed and updated the card is put up to their team to verify their understanding and implementation plan. This pre-validation of just in time backlog inventory should be happening on a constant basis to confirm that the work to done is still meaningful.
Ship your current card > Pick the card that you pre-validated last > Pick another card from the backlog and queue it for pre-validation > And repeat.