Stop Re-Receiving the Advice You Already Rejected
Third time in the same conversation, it suggests change data capture. You ruled that out an hour ago. The licence cost for the connector was more than the problem is worth, and you had that conversation with your finance partner in March. It is closed. But you never wrote that down. So the suggestion keeps arriving, you keep skipping past it, and you have now spent three turns being offered…
In the same conversation, the same suggestion keeps appearing, even though you previously dismissed it. The reason for rejecting the advice is clear: the license cost for the connector exceeded the value of the problem you were trying to solve. You had already discussed this with your finance partner in March, and the matter is now closed.
However, the suggestion keeps resurfacing, and you keep ignoring it. The reasoning for rejecting the suggestion is solely in your mind, as all the information you needed to make the decision is not present in the conversation. The conversation only includes your question about the problem you are still facing, and from this, change data capture appears as a sensible solution.
This issue is different from the gaps in previous discussions, which were due to missing context such as engine version, row counts, or on-call responsibilities. Instead, this is about a decision you have already made, and it appears to be something that should already be known as it is settled in your mind. Another version of this issue exists, which is more expensive.
Although you mentioned it earlier in the conversation, it was not as prominent as the initial rejection. Research has shown that models in multi-turn conversations give more attention to the first and last turns, and references to middle turns fall below 20%. Therefore, mentioning the rejection in the middle of the conversation does not carry the same weight as stating it at the beginning.
The mechanism is the same for everyone, but the reason for rejecting and why it remains unwritten varies depending on the individual. Beginners have not ruled out much yet, so this rarely costs them anything, except when it does. Intermediates have technical rejections, such as trying something that did not work or being too slow.
Seniors have rejections that are not technical at all, such as not using a vendor due to a procurement dispute or not proposing a migration due to past failures. These rejections are the ones that stay in your head, and it is not because you forgot. Writing "we cannot use vendor X for political reasons" feels like putting something on the record.
Writing "already ruled out: CDC (connector licence cost, decided March)" is a decisive constraint on a senior's options and is more likely to shape a useful answer than information about your stack. The weak objection to this approach is that it is basic prompt hygiene rather than an insight. However, cheap things are often the ones nobody does.
The stronger objection is that a list of rejections is a record of past decisions presented as facts about the present, and it goes stale without anyone noticing. If you ruled out a connector on license cost in March and it is now August, the vendor may have changed their pricing in June. You also ruled out a migration due to lack of capacity, but two new people joined the team, which changed the situation.
By carrying the block forward and dating each item, you ensure that the list remains relevant and up-to-date. This way, you can ask whether each rejection still holds and review them quarterly to see if any of them have expired. The list is only useful if it is a record of what is true, rather than a record of what you once concluded.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
