rationalskepticism.org

One of the reasons that the Porifera is chosen for such discussions is that the literature on molecular phylogeny points to the Porifera as having had the most distant shared common ancestor with all the other multicellular animal phyla. Even more compellingly, genes that share homologies with genes for more advanced bauplan development in other phyla can be found in modern sponges, suggesting that those genes had a very ancient origin, and that while sponges didn't take the cellular differentiation route of other phyla, they inherited the genes from the common ancestor that permitted cellular differentiation and specialisation in those other phyla. The Cnidaria is the modern clade that contains evidence of what the first instances of cellular differentiation may have looked like in the past, and other clades that arose later exhibit additional features that arose when they diverged from their common ancestors with the Cnidaria etc.

Basically, the Animalia diverged into the Porifera, the Cnidaria and the Bilateria (this is a simplified rendition of the tree in question, but not essentially wrong for being a simplification thereof), and the Bilateria gave rise to the other major animal phyla. At the Bilatera node, you then have a basic division between Protostomes (which gave rise to most of the invertebrate phyla courtesy of a split into Lophotrichozoa and Ecdysozoa) and Deuterostomes (which gave rise to echinoderms and vertebrates). Tracing the phylogeny of various gene families, including HOX genes, Pax genes, and the various signalling genes such as bmp, wnt and the hedgehog signalling genes, more or less establishes the aforementioned clades as being the clades that arose early in the history of multicellular life.

An apposite paper that reveals why sponges are important from the standpoint of understanding the early history of multicellular life is this one:

Mitochondrial Genome Of The Homoscleromorph Oscarella carmela (Porifera, Demospongiae) Reveals Unexpected Complexity In The Common Ancestor Of Sponges And Other Animals by Xiujuan Wang and Dennis V. Lavrov, Molecular Biology and Evolution, 24(2): 363-373 (2007, with advance online publication November 7th, 2006) [Full paper downloadable from here]

Wang & Lavrov, 2007 wrote:Homoscleromorpha is a small group in the phylum Porifera (Sponges) characterized by several morphological features (basement membrane, acrosomes in spermatozoa, and cross-striated rootlets of the flagellar basal apparatus) shared with eumetazoan animals but not found in most other sponges. To clarify the phylogenetic position of this group, we determined and analyzed the complete mitochondrial DNA (mtDNA) sequence of the homoscleromorph sponge Oscarella carmela (Porifera, Demospongiae). O. carmela mtDNA is 20,327 bp and contains the largest complement of genes reported for animal mtDNA, including a putative gene for the C subunit of the twin-arginine translocase (tatC) that has never been found in animal mtDNA. The genes in O. carmela mtDNA are arranged in 2 clusters with opposite transcriptional orientations, a gene arrangement reminiscent of those in several cnidarian mtDNAs but unlike those reported in sponges. At the same time, phylogenetic analyses based on concatenated amino acid sequences from 12 mitochondrial (mt) protein genes strongly support the phylogenetic affinity between the Homoscleromorpha and other demosponges. Altogether, our data suggest that homoscleromorphs are demosponges that have retained ancestral features in both mt genome and morphological organization lost in other taxa and that the most recent common ancestor of sponges and other animals was morphologically and genetically more complex than previously thought.

EDIT : Oh, and another apposite paper can be found here.

EDIT 2 : Found another one, though that one is behind a paywall.