Maybe I'm missing something here about Tilt and its connectivity, but when you are accessing data, it is either connected or not, no? It is probably not poorly connected and thus sending you erroneous quality data. I think the issue which I have seen, which may be related to the above problem, is that the Tilt can get krausen and/or CO2 bubbles glommed on the device, and then it may break off. When I have what looks to be a spurious data readings, I slightly rock my conical several times, and then after the liquid movement subsides, I often get representative data points again. I think this can happen especially with high gravity beers and high krausen-generating yeasts.