It is no accident that the 'science' of statistics arose from the efforts of gamblers to predict, and thereby to profit from, the 'odds.' Science submerges into engineering when prediction becomes an instrument of control. The purpose of engineering is control. So, when Foucault said that the purpose of the "human sciences" was 'surveillance', he was viewing 'surveillance' as the tool of social science applied to social engineering, the prediction--and direction--of human behavior. No matter the claim of the apologists to the contrary, the reason for 'science,' for milennia, has been 'prediction' for purposes of 'control.' Foucault, of course, translated this into 'surveillance in the service of power'.
That Foucault said it is in and of itself not warrant to believe it, without the evidence of your own experience. the relevance of the observation stems from its (uncomfortable, to some) truth-to-life.
'Human' science disappeared into 'social engineering' without a ripple. The wholesale surveillance state has been a "Liberal" dream since Bentham. Its genius being that since, in the perfect Panopticon, one might be at ANY time under observation, one NEVER knows when one is under observation, and so constrains one's behavior to conform to the expectations of the watchers whether they be watching or not. The inclination of the State to perfect the Panopticon is undeniable, and virtually every apparently democratizing tool developed in the culture of the State always already is designed for use by the State to observe, and thereby to enforce discipline upon, the unruly, untrustworthy, unreliable populace.
Agreement with the main point is not necessary, since experience indicates that this is indeed the direction in which society and culture are 'developing' (it can scarcely be called 'progress'), and that this impetus is exaggerated, not impeded, by the practitioners of the "human sciences," their apoligists and philosophers as well...