Opinions seem to be all over the map regarding both the granular matrix type and the gypsum type of soil moisture sensors. Some people (like Picta) have a great experience and others don't. It's definitely true that soil temperature will affect the readings, and one advantage to using a pre-made one like the Davis is that you can use their correction formula.
So, it seems the only way to really know is to try it for oneself. I think this summer or maybe the next I might give a nominally similar DIY version a try on some indoor potted plants:
http://www.cheapvegetablegardener.com/how-to-make-cheap-soil-moisture-sensor-2/#comment-202129
in part because indoors the temperature is fairly uniform. I imagine the Davis ones are about as good as they get, but it would be more fun if each type of potted plant, with probably a different uptake rate, could have its own sensor--and that's difficult to cost justify unless the sensors are truly cheap.
I can see how the Davis sensors would make sense for big farms having large level fields, though, because it's probably well tilled soil with a uniform crop, with each plant being nearly the same and so having about the same uptake rate at any given time. When I started this two years ago I thought the same might hold true for my lawn, which guzzles water, but I've since learned that the soil composition underneath my lawn varies quite a bit, and making it uniform would require a lot of effort. Also, I've found that sprinkler irrigation is very difficult to make make uniform after-the-fact, and the system that came with my house definitely isn't. Lastly, the amount of sun and wind and even rain exposure varies as well, because the house and fences and trees block the sun and wind and redistribute rain to varying degrees in some areas more than others. To further complicate, you have tree roots aggressively competing for moisture in some areas and not others. With benefit of hindsight, I can now see that those causes of non-uniformity all contributed to make it really, really hard to generalize from just a few soil moisture samples. Shifting to underground dripline irrigation would undoubtedly improve the uniformity and improve efficiency, but the other causes of non-uniformity would remain. So, in a nutshell, even though I had 10 wireless soil moisture sensors (with temperature compensation), it still wasn't enough to match the problem. To do it right would either require removing the sources of non-uniformity (some of which just aren't going to budge), or else a whole lot more sensors and much finer grained zoning. I already have 10 zones. It might take 30, maybe 40, to get a tight enough fit to save a significant amount of water without also killing parts of the lawn. Until there are fairly cheap, individually addressable sprinkler heads, though, or something similar for subsoil irrigation, the cost of setting that up properly renders it impractical.
On the other hand, 10 sensors was enough to keep everything green, so they are good for that. I'm just skeptical about the potential for really huge water savings.