Martin & Koonin recently proposed that the eukaryote nucleus evolved as a quality control mechanism to prevent ribosome readthrough into introns. In their scenario, the bacterial ancestor of mitochondria was resident in an archaeal cell, and group II introns (carried by the fledgling mitochondrion) inserted into coding regions in the archaeal host genome. They suggest that if transcription and translation were coupled, and because splicing is expected to have been slower than translation, the effect of insertion would have been ribosome readthrough into introns, resulting in production of aberrant proteins. The emergence of the nuclear compartment would thus have served to separate transcription and splicing from translation, thereby alleviating this problem. In this article, I argue that Martin & Koonin's model is not compatible with current knowledge. The model requires that group II introns would spread aggressively through an archaeal genome. It is well known that selfish elements can spread through an outbreeding sexual population despite a substantial fitness cost to the host. The same is not true for asexual lineages however, where both theory and observation argue that such elements will be under pressure to reduce proliferation, and may be lost completely. The recent introduction of group II introns into archaea by horizontal transfer provides a natural test case with which to evaluate Martin & Koonin's model. The distribution and behaviour of these introns fits prior theoretical expectations, not the scenario of aggressive proliferation advocated by Martin & Koonin. I therefore conclude that the mitochondrial seed hypothesis for the origin of eukaryote introns, on which their model is based, better explains the early expansion of introns in eukaryotes. The mitochondrial seed hypothesis has the capacity to separate the origin of eukaryotes from the origin of introns, leaving open the possibility that the cell that engulfed the ancestor of mitochondria was a sexually outcrossing eukaryote cell.