The Pamir Times reports that as of lunchtime today the flow through the spillway had increased to about 850 cubic feet per second - i.e. 24 cubic metres per second, giving a discharge - time graph that looks like this:
For the last 72 hours or so the increase in discharge with time appears to have been linear, which is interesting. This suggests that erosion of the saddle has yet to initiate fully. Fortunately so far the in situ (i.e. not excavated) lacustrine silt and clay has been more resistant to erosion than some had speculated, although the disturbed (excavated and dumped) material has proven to be very erodible. However, the inflow remains more than three times greater than the outflow (seepage plus spillway flow), and retrogressive erosion will in due course steepen the channel, which will increase the erosive potential.
In the meantime the level of the lake is continuing to rise at 50 cm or so per day, with the upshot that further flooding is occurring upstream. The Pamir Times has a tragic (but beautifully shot) photo piece (plus here) on these continued impacts. The level of impact up there is all too clear: