I won't start a thread on it but we are at the edge of a big change in term of SNP data. The 1000 genomes project collected many new Y SNPs and many are still not included in the ISOGG tree. Geno 2.0 the new chip used by the genographic project is using an unprecedented set of Y SNPs from the 1000 genomes project , from Chris Taylor Smith, From some chinese sequencing and from some Sardinian sequencing (+ those already known). A lot of data a presently accumulating but very little is known on forums such as this one. just to give an idea : at the root of I1 there are presently 24 SNPs (it's a specificity of I1 to have this long list of synonymous SNPs at the root) but with Geno 2.0 this figure should go around 72 SNPs (from partial results) ; at the I2 root there are only 2 SNPs on ISOGG but Geno 2.0 knows 12 (or about -to be confirmed), quite a change.
Why is it related to paleolithic human groups ? SNPs will allow a better estimation of the age of the branches in the Y tree. Presently, I believe that several ages are very inaccurate. By the way I also have some suspicion against the many ancient DNA haplogroup designation.
So, I consider as a possibility that [IJ] was the only Y haplogroup in Europe. As the glaciation isolated Europe IJ had an "in situ" evolution to I (and to J in the middle east). I subsequently split into I1 and I2. This scheme is similar to the mitochondrial U --> U5 --> U5a and U5b with the possibility that U also evolved in U2.
What's above isn't a demonstration of what occurred but it's a demonstration that a "simple" scheme is possible. An in situ evolution of Y-DNA I / mtDNA U would explain a specific european type.