I find it difficult to run pre-written scenarios at very short notice, too, but the amount of work entailed in designing a scenario and hoping it works is more than outweighed by knowing the scenario should have been playtested and and at least some bug sironed out already. Printing off some handouts? Fine. Printing or designing a combat map to go with it - also fine, but odds are someone's provided one already AND it's something you have to do anyway.
I tend to run a mix of hand-built and bought scenarios in Conan but normally have an overriding patron or BBEG to spur or motivate the players. Some of my scenario outlines are sketchy, with NPCs downloaded from a range of sources, others are intended for submission or external use, so are really quite tight and I build custom NPCs.
But as for railroading? I tend not to, even if the scenario pushes that way (though I'm not sure many published scenarios are written with the intention of railroading). It is often possible to leave things open so events in a module happen as a result of what the player's do or where they go. Perhaps the plot line is extended a bit, perhaps a few other events happen in the middle, or perhaps even another short scenario gives a distraction before the original adventure rears its head again. The overarching idea is that it's a game meant for enjoyment, not a slavish adherence to what's written. It seems to work.
But I haven't run Lurking Terror, yet though.
