Oh, sure. Absolutely. I’m just honing in on three points:
One, the fact that 2D4-4 results in a spread of 24 to -12. And that’s a pretty big gulf.
Two, so if you end up with some characters with 15 point builds while others have 4 point builds, you’re going to have some characters who are super successful at checks and attempts and others who aren’t. We’ve seen enough concern over the speed quiver around here to know that great disparity in damage, for example, makes the less fortunate chafe a little. I suspect it would be the same here. While Mary constantly crushes through every task, Joe might sulk a little. Also, a huge gulf makes it harder on you as the DM to assess a universal target number.
Three, if there is chafing at your table, you may have to regulate it as the DM and feel compelled to play a little “catch up” with your less fortunate characters.
Like anything, I’m not knocking disparity or your idea @RH1N0. I love rolling 3D6 straight down the line. And I think 2D4-4 will work. But from a DM or a game design standpoint, when you end up with a spread of -12 to 24 for stats (not attributes that mitigate a little with a range, but stats/modifiers), it helps to know the cost/benefit analysis: what you’re gaining and what you lose when you change things up.
As we talked through this yesterday, @James_Horn had a good solution for the vast gulf, and that was to ensure all builds are at least 6 point builds, or some variant. But, if you’re good with -12 to 24, rock on dude. If that works for you and your table, press.
The idea of rolling 2D4-4 (also 1D4-1) has been around since the google+ days of the forum, but I never played with it much until you posted it again. And the anecdotal, off-the-cuff rolls I got surprised me. It generated some great conversation about the pros and cons of rolling random stats. But that’s all it has been for me. Conversation.