As discussed privately, the following test cases should raise TypeError
(currently ValueError):

self.assertRaises(TypeError, Decimal, (0., (4, 3, 4, 9, 1), 2) )
self.assertRaises(TypeError, Decimal, (Decimal(1), (4, 3, 4, 9, 1), 2))
self.assertRaises(TypeError, Decimal, (1, (4, 3, 4, 9, 1), 0.) )
self.assertRaises(TypeError, Decimal, (1, (4, 3, 4, None, 1), 2) )
self.assertRaises(TypeError, Decimal, (1, (4, 3, 4, 'a', 1), 2) )

It's a bit ambiguous:  the object being supplied to the Decimal constructor does have the right type (tuple), but the wrong contents (i.e., value).  So you could argue either way on this, I guess.

A ValueError seems more appropriate to me.

From both practical and purity points of view I think I prefer TypeError for the cases Stefan gives: if you're passing a float for the sign, then you've somehow ended up with the wrong type for that sign, and that's a TypeError rather than a ValueError.

I could see a TypeError for the outer tuple a tuple wasn't passed-in.  If the contents aren't what we expect, I believe that calling it a ValueError is the Python way.

Given the lack of agreement here, I propose that we just leave the current code as it is;  it doesn't seem like a big issue to me, and we should resist messing with code that isn't clearly broken.