Among the many conference/journal discussions, one systems person said the reason they don't publish in journals is that their work is very dependent on current technology. A few years in the future the technology will change and this research is no long relevant. The best systems research develop ideas that transcends technology, but much research in systems have a limited lifespan of relevancy.
In theory, if you prove a theorem that theorem will be true forever. In fact that theorem was always true, we just didn't know it. So it is important to have refereed long-term archival write ups of these results. A theorem could be supplanted by another result or get less interesting over time but it always remains true. Theory results transcend technology, though most theory result have a zero lifespan of relevancy.