You should check in the Carey Sundberg part B. In the 4th edition, the first chapter is dedicated to the alkylation of nucleophilic carbon intermediates, which turns out to be mostly about enolates. A key sentence is: "Ideal conditions for kinetic control of enolate formation are those in which deprotonation is rapid, quantitative and irreversible".
Using LDA (a bulky strong base) on butanone you should be able to deprotonate quantitatively the least hindered carbon atom, which turns out to be the terminal methyl group. However, this is rarely perfect and some deprotonation of the most hindered carbon atom occurs at the same time.
However, you need to use a slight excess of strong base for this to be true. In the case where the ratio of LDA: ketone would be slightly below 1.0, you would end up with a mixture of the kinetic enolate and some leftover ketone. The kinetic enolate could then deprotonate the leftover ketone to give a mixture of kinetic and thermodynamic enolates (remember, the deprotonation is rarely 100% selective) + the resulting ketone (since you protonated the enolate during the process). This newly formed enolate could then undergo the same reaction, until you reach a point of equilibrium where the ratio of the two differen enolates is determined by their respective stability (thermodynamic enolate).