The agent lives in the system, receiving and executing commands. The protocol exists as a language, free for anyone to speak.
The agent is control, enforcing an external concept of truth and reporting abberations. The protocol is possibility, computers collaborating with other computers.
The agents are Microsoft Group Policy, Apple Gatekeeper, Firefox Updater. The protocols are DHCP, DNS-SD, HTTP.
Of course, agents rely on protols, and well-known protocols rely on agent-like daemons and services, but I see a difference in paradigm that attracts administrators to one or the other. Is this the divide that Neal Stephenson is talking about when he says that “Businesspeople…are polar opposites of hackers in so many ways”? Perhaps businesspeople care about expediency and economy, ends served well by the aforementioned agents, while hackers just care about tinkering, exploring, and subverting, pursuits that invite a more laissez-faire network policy. Group Policy dares to change your desktop background to a safe corporate-blue, but DHCP is just like “Hey dude! Here’s some gateway info and a rad DNS server down the street…I’ll be hammocking out back if you need anything.”
I’m writing this article because I’m conflicted about how best to administer a network. Does the posession of laissez-faire political ideas (like mine) neccessitate laissez-faire attitudes towards corporate governance and computer network administration? I can already imagine the term laissez-faire translating to “fast and loose” or “vulnerable” in the mind of an infosec professional. Is it even possible to have a secure and tinkerable network? Perhaps these questions can only be answered through long experience.