Unless you've got a very good reason to suggest a lead, I would say this: never interfere over an invitation. My reflections on this subject arose particularly from one hand at EMBA's A/X Swiss this last weekend. I was sitting East and and heard partner open 1♦ as dealer. RHO passed and, eschewing the 2♣ call, I instead bid 2NT, showing 11-12 high card points. As much as I love and value extra aces, this hand looks most likely to be playing in notrump where the Work point count is more accurate. And I didn't feel I could add anything extra for the threadbare club suit.
LHO now bid 3♥ which was doubled by partner, ending the auction. We collected 500 (the double-dummy result) which looked perhaps like a 3-imp pickup over 3NT (yes, it is possible to make 10 tricks in notrump if you're fearless and brilliant). In fact, something very similar happened at the other table. My counterpart bid 2♣ but otherwise the bidding was the same.
Of course, South was very unlucky to find partner with nothing useful on offense (but plenty of defense). And, given the possibility of a misdefence, there was a good prospect of gain.
On the other hand, there is no more dangerous "death seat" than this particular one. RHO has invited in notrump and we will be on lead if we end up defending (thus no advantage in making a lead-directing bid). Furthermore, LHO knows to within half a point her partnership's total assets. What if North had possessed both the J and T of diamonds? It would then be impossible to make 3NT and the potential loss from bidding would be either 380 points (9 imps) or 550 (11 imps), depending on whether the opponents actually bid the game.
The same argument applies to suit contract invitations, also. The invitee will know to within two or three points his partnership's total assets and moreover will have a pretty good idea whether their game will make and also how much the bidder is likely to go down. Essentially, you (the potential intervener) will be making the last guess, always something to be avoided if possible.
A similar argument applies after the opponents have issued and declined an invitation, as I discovered the previous day (surely, I should have known better!). Playing pairs against Adam Grossack, who now has two gold medals from world junior/youth play, I decided to compete after 1♦ (2 or more diamonds, 11-15) by me, 1♥, pass, 2♦, pass, 2♥, pass, pass. This is kind of like balancing after 1♥, 2♥, right? And in this case, partner might have a fit for my six-bagger in diamonds but not want to support. But no, it isn't like that. The opponents have exchanged all of the information they need at this point. At the club, most opponents would have simply gone on to 3♥. But I had forgotten whom I was playing against! Adam doubled and soon I was looking at -500. Yes, they had game all along - but I had made the fatal last guess. Details of hand suppressed to save me some embarrassment.
Finally, in a variation on the theme, we received a happier result from the second session of the same pairs event, where, having established a major suit fit and (a bit reluctantly) declined a game try (in clubs), my opponent took one last shot with their suit (diamonds). I didn't think we would be getting rich with a double (they can make their 4♦) so I bid the game anyway ("you talked me into it"). Very bad discipline, I know, but it worked out. I was able to take 11 tricks for a clear top on the trump lead (a diamond lead would have set me one).
BTW, if you're wondering about the title, it's a fairly lame reference to papers with such titles in the computer science academia, especially the first such: GOTO Statement Considered Harmful by Edsger Dijkstra (Communications of the ACM, 1968).