"Rules [of thumb] are made to be broken". But "you have to understand why the rule exists so you can effectively choose to break it". This is what makes them different from Laws and Regulations.
To answer your question straight up, it would be rare to see an expert whose weak 2 agreements in all positions are "2/top 3" (even 2/top 3 or 3/top 5); in fact, it would be rare to see an expert whose agreements in any position (maybe 2nd seat unfavourable) is 2/top 3. This conservative rule of thumb has been shown to be less effective than a more liberal style.
It's easier to play a conservative style when it comes up; it's trivial for partner to know when to bid game, and when to double the opponents. It's not that KQ or AQ vs QJT is a trick, it's that partner knows that top-honour-doubleton will solidify the suit (for max 1 loser, 90% of the time), when counting tricks for game. You can use your limited asking room very effectively.
But it's much easier to oppose a conservative style when they pass and your teammates/the rest of the room doesn't. The whole point behind preempting is a gamble that because you're weak, the opponents own the hand, and we should take away their room.
Third seat "preempts" have always been some flavour of "he has 13 cards and some of them are spades". With no worries of preempting partner, hands that are clearly too dangerous to open in case partner has the big hand are clear openers, as are hands that are "too strong" and would be opened at the 1 level if partner could have a hand.
As far as systems go, I submit EHAA. From the linked synopsis:
An EHAA two-bid shows six to twelve high card points, and a five card or longer suit. There are no restrictions on suit quality (xxxxx and AKQJxxxx both qualify).
Is it a good system? Well, nobody would play it in a world championship, so no. But it got us to the second day of a National event, because it puts the opponents to a lot of guesses the rest of the room doesn't have, and if they get enough wrong, even underdogs can do well (of course, our first final session was 35%; if they get enough right, it doesn't matter how good you play). Obviously, all the system you have over a more normal weak 2 goes out the window, replaced by things that work when partner could have an 8- (or even 6-) high suit.
Less spectacularly, my regular partnerships:
- play Precision, and we've agreed "2/top3 or 3/top5 second or vul". Remember we open most 10-counts at the 1 level, especially 10-counts with a 6-card major;
- play "anything goes" in a 12-14 NT context. First seat favourable, J9xxxx is an expected minimum (and could be worse), second seat all white probably minimum KJ8xxx or QT9xxx. We play Ogust so we have a hope the few times partner has the hand; we lose on average when she does; we win on average when she doesn't[*];
- expects to be the best pair (or second) in the room, and therefore avoids gambles. We play disciplined first two seats, trusting our defence to get back some of the advantage we're giving to the more freewheeling pairs, knowing we're getting their bad results "for free".
[*]: This is the partnership that plays EHAA when we're bored, challenged to, or when the opponents complain that "we only play all that weird stuff to confuse us". No, this is what we play when we want to confuse you...
