Thanks for the link to the info about the Kenyan Archives, at least the articles are stored in vaults. I would have been highly suspicious if the building had been destroyed by fire during the recent election unrest. Sounds like a trip to hell to even navigate the archives let alone to actually find out information. I did note that some of the papers from the death of Tom Mboya may be released, which is interesting since his family has been campaigning since 2003 for a full and complete investigation of the matter. I imagine such an investigation will be put on hold now that his widow has passed away(Jan 27/09), as has one of his sons, Peter.(2004 in a motorcycle accident).
Having done quite a bit of research of Irish records, I wouldn't be that suspicious. Many Irish records were destroyed during periods of civil unrest which continued well into the 20th Century, or through unintentional mishaps, and fire was the most frequent source of the destruction. As far as the accuracy of the records go, I'd say it depends.
Yes, the British placed great emphasis on record-keeping, but when it came to registration of births, marriages and deaths, it depended on native self-reporting, supported by the negative incentive of fines for non-reporting. Again, in periods of unrest, or in the event of a personal family tragedy, events went unreported. Add to that the fact the local registrars were British, so the spelling of family and place names were often mangled in the records. As a final difficulty, in predominantly rural societies such as Ireland (and Kenya) in the the 19th and early 20th centuries, age was of little consequence. (You were young, middle-aged or old, and that was about it.) As a result, you find the birth, marriage and death records (if you're lucky enough to find all three for the same person) vary in age by an average of 5 years, sometimes more.
It wasn't until government entitlements such as old-age pensions were introduced that accuracy in age became imperative. To a lesser extent, the same applies to the US. Most states or large cities didn't have a centralized vital records system until the advent of Social Security registration in the late '30's. In fact, one of the major WPA projects involved manually creating a name index for earlier US census records, since these were frequently used as proof of age and/or citizenship in the absence of birth certificates.
I don't know what kind of registration system the Brits or the Kenyan government set up in Kenya, but I suspect, from the scenes described by that article about the National Archives, that the surviving records aren't all that complete, so you find a lot of research into ancillary materials such as old publications, lists, wills, land records, etc.