Hi @edtimm Ahh, very good question. We’re all long-time windsurfers and kitesurfers too. Most of us are engineers and meteorologists who specialize in wind measurement…so anemometers are very dear to us. 
In this case, we think you might be getting fooled by the temporal resolution of your graph view. The graph you posted above is displaying 3hr data. Please see Data archive buckets explained for a more complete explanation of data binning. Each 3hr data point for wind includes the avg wind speed over the 3hr duration, the lowest recorded instantaneous value during that 3hrs is displayed as the lull, and the highest recorded instantaneous values during that 3hrs is displayed as the gust. With such a large timeframe, you can easily see why the avg would not appear roughly the mid-point between gust and lull. Rather, at the 3hr resolution the data present actually is telling you the highest and lowest obs during that time and the avg over the duration.
Try this: zoom into the 1min temporal resolution and you will instantly see what you perceive to be a more “sane” range / correlation between gust | lull | avg. Here is your station from today 4 Oct as example:
To echo what @WFstaff said (a degreed meteorologist on WF staff) … if you compare a quality sonic anemometer side-by-side with a traditional spinny cup anemometer…you will likely notice differences due mainly to the inertia in the spinny cups. A sonic is more capable at measuring the actual instantaneous wind speeds without the false influence of inertia, old bearings, etc. It’s challenging all of us to re-cast how we evaluate wind speed data.