With respect to India's colonial experience, Razib wrote: "[W]ould anyone be willing to entertain the possibility that some groups benefited vis-a-vis other browns because of british patronage? i am thinking of, for example, dalit soldiers who attained higher status because of their service in the british army, and who boldly walked into "high caste" areas of madras/chennai with their weapons and dared to the brahmins to kick them out. also, as someone of muslim background i am aware that many muslims were skeptical about transferring british christian rule for the rule of indian elites who would predominantly be hindu (ergo, pakistan)."
That may well be so, but the problem is that the above view takes the colonial subject (I mean the subject of the colonial regime) as a given, and does not recognize that colonialism itself might have played a role in manufacturing such a subject and such a consciousness. Take the Muslim example. It is undoubtedly true that large numbers of Muslims might have, by the third and fourth decades of the twentieth century, preferred British rule to "Hindu rule." But as vast amounts of research have powerfully shown, it is precisely in a post-colonial India that the question of the two-nation theory arises. To put it another way, the Western-style nationalist, for whom all nationalism is modeled on European models, and for whom "one people", "one nation", "one religion" are the signs uber alles of nationhood, is a product of colonialism. Colonialism's own complicity in the manufacture of identities who were in turn "protected" by the colonial dispensation cannot be ignored.
I take just one example, but one the destructive impact of which was remarkable: separate electorates on a communal basis, instituted by the British in the early twentieth century. Under the scheme (in place till independence, and re-instituted in Pakistan in the 1980s) Hindus could only vote for Hindus, and Muslims only for Muslims; moreover, there were no seats reserved on the basis of any cross-communal categories (e.g. no seat reservation on the basis of Gujarati-speakers, or Marathi or Bengali-speakers; no seat reservation based on occupation; and so on). Can there be any doubt that such a system incentivized the manufacture of communal sentiment? If the politicians, voters, the very political reality that is produced by this system is skewed a certain way, I for one cannot accept the resulting preferences at face value. Specifically, while one can (and should) always interrogate foundational assumptions, the choice (where Muslims are concerned) between "British" and "Hindu" rule is itself a choice presented by the colonial prism (as is, of course, the conception of India as a "Hindu nation," or as reducible to "Muslim" and "Hindu" nations).
The Dalit example is a complex one. In particular, while there can be little doubt that Dalits fared far better in the armies of the East India Company than, say, in the realm of the Peshwas, a fascinating realm of study is the ways in which the British used the census, as well as disciplines such as anthropology, to standardize the caste system. As you note perhaps a weblog is no place for a detailed discussion, but I highly recommend the book "Castes of Mind" by Nicholas Dirks, which is in large part addressed to precisely this issue. One point Dirks highlights is how British administrators were relentlessly focused on classifying Hindus into the four-fold varna scheme of Manu even though, as their writings and records themselves testify, Hindu reality stubbornly refused to accord with the Dharmashastra (the exceptions were the groups the British could most easily place in the varna system: Brahmins and untouchables, both of whom had some sort of pan-Indian presence). The tensions came to a head with the first colonial census, with forcible British characterization of various jatis and groups as belonging to one or another varna. The result was the institutionalisation of a system where the varna system became the measure of social classification-- anything else was just not recognized by the state. Thus if one was a Reddy or a Naicker or whatever, it was no longer enough to just be a Reddy or a Yadav or a Naicker, but one had to be situated in the varna scheme in order to count when it came time for giving out jobs, political representation, etc.
The point is not that the caste system was created by the British (and as noted two ends of the varna system were rather recognizable in a pan-Indian context, namely Brahmins and untouchables), but that under the British caste became the primary organizing political and social principle of Indian society (just as religion became the organizing principle of a broader Muslim and Hindu division).
Finally, there has also been much scholarship on the fact that what constituted untouchability (and more generally "low" status) varied from region to region, and even within regions. The Dalits of the regions of the later Marattha confederacy were among the most oppressed in Indian history; but the same may not necessarily be said of the "low" millions of Bengal. That untouchability is always unacceptable no matter what I accept as a sine qua non; but it is hard for me to accept uncritically the (self-interested) notion that endorsement of the colonial imperium was the only way out of the "dead end" of "native" tradition. I do not blame Dalits who accepted this choice, perhaps in their situation more could not be expected; but the choice remains a false one.
Might one entertain alternative possibilities? Why not? Might one not read even in Ambedkar's late turn to the Buddha, that is to say the Buddha of protest, the Buddha of dissent and disappointment, a recognition of the problem inherent in proffering colonial ideology alone (even in its liberal self-representation) as the antidote to the injustice that so haunted the great soul, and a recognition that the rhetoric of resistance and liberation could not be coterminous with a conquering idiom?