Science is a very conservative effort, as we seldom can completely toss out what we've found in the past to be true. Instead we have to keep adding and modifying our understanding, constantly improving it. Consequently, it goes against the grain to have definitions of words changing. It's difficult enough to get a consistent definition in the first place (despite the efforts of IUPAC and other standard creating organizations), so to see subtle changes occur over time [*] is disconcerting.
Such is the case with the term "thixotropy". As has been reported in a review paper (open access) by Barnes, the term was first used nearly 100 years ago to describe the situation with certain gels that would become liquidlike when shaken or otherwise disturbed. Upon being allowed to rest, the gel would be restored. (At a molecular level, what was occurring was that the gel network responsible for the solid characteristics was disrupted and/or destroyed by the mechanical forces. At rest, the network would restore itself.)
Somewhere along the lines, a more proper, less operational definition has arisen, which is that a thixotropic material is one where the viscosity of the material at a constant stress rate decreases over time. It certain captures the original definition, but is broader in that it is not restricted to gels.
It also is unfortunately the rheological term term that I personally think is the most abused. The confusion of the term's meaning that I've encountered is endless. First, there are the people who confuse it with shear thinning (a.k.a. pseudoplasticity), a term which instead refers to the viscosity decrease seen when the shear rate increases. This is totally different than thixotropy, as a shear thinning material at a constant shear rate will have a constant viscosity over time. The decrease in viscosity only occurs when increasing the shear rate.
Second, there are the people that think it has something to do with thickening, since "thick" sounds like "thix-". This problem is compounded by all the "thick" gels created by adding a thixotropic agent; and of course these gels show thixotropic behavior.
Clearly rheology would be better off when sticking to the most current definition, but there is quite a bit of embedded infrastructure that needs to be replaced, and the characteristic time for that reaction is that of an engineer's career. Just remember that for exponential decay, the characteristic time is the time for a 63% reduction from the original concentration, a value that seems fairly correct for this situation.
[*] Now on the other hand, I find etymology - how the definition of words change over time - to be a fascinating study. I love that languages are alive, constantly changing entities, with English probably being the most vibrant of all.