By "manipulative" I assume you mean hand texturing. With experience this can be very accurate. It lacks precision, in that it will indicate the category (e.g. clay loam), it won't tell you the percentage of clay. But great precision is not always necessary, and hand texturing can save a huge amount of time, especially in the field. My approach is usually to hand texture in the field and bring back a few samples to "calibrate" my field results with laboratory methods, or where the textural category is borderline.

Hand texturing can be very accurate, depending on the skill of the person that does it. Often, it is even more reliable (although more categorical) than the lab as a lot can go wrong with the pipette or hydrometer method: e.g. in calcareous soils it can be really difficult to adequately remove all carbonates, in volcanic soils it can be difficult to disperse all the clay, in saline soils the Na content can be too high, etc. I always check field texture of my samples by hand before sending them to the lab.

A good idea is indeed to 'calibrate' your fingers by doing the exercise for soils with a known texture. Always wet the soil though, to get accurate results. The USDA has a really comprehensive technique, which you can see nicely demonstrated here:

http://youtu.be/GWZwbVJCNec

Please bear in mind that texture triangles and classes vary between systems, so make sure you use the correct one.