Second, it's worth keeping in mind that sometimes it's useful to have multiple small programs instead of one big coordinated program because small programs can experiment with different approaches to see what works best instead of being stifled by a single big bureaucracy. It's also the case that sometimes it's actually more efficient to have multiple programs. A homelessness program aimed at helping municipal governments is probably best run out of an agency that already deals with municipal governments and doesn't have to reinvent the wheel just to figure out who the players are and how to do outreach. Ditto for programs aimed at church groups, nonprofits, law enforcement, etc.Really good point. I'm far from convinced that anything was gained by the last major shakeup in the organization of the federal government -- the creation of the Department of Homeland Security (and that's outside of the horrible name, but that's another subject).
Streamlining is a worthwhile goal, but it's wise not to get too caught up in box-drawing exercises. They're the favorite solution of management consultants and CEOs with no better ideas at hand, but it's rare that reporting structure is really the key problem in an organization. If you have 20 different programs that provide assistance to 20 different kinds of clients, it's simply not obvious whether you should organize things by program type or by client type. On a more general level, sometimes centralization is good, sometimes decentralization is good, and sometimes it's good just to shake things up. But the real problems usually lie elsewhere.
On the other hand, the electoral incentives involved in this sort of thing aren't pretty. The rewards (in positive press attention) are almost certainly found in uncovering the problems, not in solving them. Which means that from the point of view of politicians, it's a good thing to have some abuses; otherwise, there's nothing to uncover. And of course individual bureaucrats usually have an incentive to keep their office open and well-funded, even if it would be clearly better to merge with another agency.
Now, you can take that too far -- just because there are some political incentives for waste in government doesn't mean that all or even a large portion of government spending is wasteful. It's a cost, not necessarily a reason for action (and anyone who believes large private sector organizations are inevitably very efficient because of the profit motive has presumably never worked in one, nor have they thought about the incentives faced by many individuals within those organizations). But still, the occasional push for reform is probably a useful corrective, even if many actual reforms might turn out to be counterproductive.