Keep on the Borderland is a sandbox so adventures can be written that are sandboxes.
Both those examples cannot be "true" sanboxes, though, because you can't (typically) say "giants are dumb, let's go somewhere else". Or "that temple has been there forever, we can come back later".
<snip>
Implicit assumption is GM provides something interactive for the players
The last time I GMed KotB it was more like what mcbobbo describes in the final quoted sentence than what Emerikol describes in the first quoted sentence.
The players had both built (multi-class) thief PCs. Of all the classic D&D classes, in my view the thief actually provides the clearest hooks for the GM: players who build thieves want to engage in skull-duggery.
So, I narrated them as being in the Keep, and then framed them into conflict with the evil priest (it was a long time ago, so I've forgotten the details). The campaign ended up being about the cult (which I ran as a necromantic death cult), and in due course the focus of adventure moved from the Keep to a nearby town which had also been infiltrated by the cult. The PCs never went to the caves (I can't remember if they ever explored the "wilderness").
So I did provide something interactive for the players, drawing on the material provided by the module. But, contrary to the paragraph from mcbobbo in the middle of the quotes, the players didn't have to take their PCs into the caves. All of us (players and GM) followed the adventure where it led.
Either your DM allows for you to leave the area and ignore Giants/Temple (playing a sandbox campaign) or forces the issue (railroads the campaign)
Similar to my discussion with [MENTION=20323]Quickleaf[/MENTION], I think it misdescribes the range of options to treat sandbox and railroad as two extremes on a spectrum. There are other approaches. For instance, if the GM describes the Keep being under attack by hobgoblins from the Caves, then that is "forcing the issue", but - provided the GM is actually framing the PCs into a situation of interest to the players - then they are not just going to have their PCs leave the area.
But this goes back to the issue of D&D adventure design, raised upthread by [MENTION=6670763]Yora[/MENTION] and [MENTION=23240]steenan[/MENTION]. D&D modules have a tendency to be very weak when it comes to the situation. So instead of suggestions for forcing the issue by dropping the players into the action ("You are in the Keep when hobgoblins assault it - how do you react?"), they tend to either set out a rather static situation (static, at least, as far as the PCs are concerned - eg KotB, GDQ, etc) or else set up a "hook" which the PCs have to follow if the adventure is to go anywhere at all (countless examples could be given, but Dead Gods and Expedition to the Demonweb Pits are two that come straight to mind).
That's one reason why I'm fairly choosy with the modules that I use.
A railroad gives that emotional narrative and provides a strong direction, but sacrifices player agency. With more active or critical players it can lead to moments without a strong sense of motivation and a sort of "why do I care?" attitude.
I think the tension in this paragraph brings out my own objections to railroading - they purport to give dramatic narrative but in fact frequently fail to do so, precisely because of that "why should I care?" problem. The player is, in effect, being prescribed a pre-given emotional response by the GM.
This can work in certain media - cinema is particularly good at it - but in my experience RPGing typically isn't such a medium. And the quality of the stories (plot, theme and especially actual, real-life pacing) is such that they don't work just as stories on their own terms. They work by way of buy-in. And the safest pathway to buy-in is co-creation.
Which rules out railroading.
- Ensures the players have agency throughout. (sandbox)
- Provides an emotionally gripping experience and strong direction. (railroad)
- Creates/encourages strong motivations, i.e. for the players to know what they're after and to have a reason to care about it. (indie)
Currently I'm approaching #1 by...
- Building open ended encounters that can be solved in multiple ways.
- Providing indexes to the various sites & scenarios encountered in the adventure, for a DM to use in a plug-and-play manner.
- Giving multiple hooks for each adventure.
- Tying encounters to each other, noting how the outcome of one can affect how others play out.
- Writing multiple endgame states to each conflict, and how the players' decisions/actions might lead to each.
- Trusting that boundary conditions to the adventure don't diminish agency.
I'm approaching #2 by...
- Making the main story arcs rugged enough to adapt to player abuse.
- Presenting the adventure in discrete chunks that are thematically connected, then choosing a default presentation of those chunks according to recommended experience level.
- Framing opening scenes to provide strong direction.
- Spending more time developing compelling NPCs and how their plots would unfold without player intervention, as well as notes on how they adapt to changes.
I'm approaching #3 by...
- Creating a choice at character creation (and beyond) for how individual PCs connect to the various story arcs and themes. Likewise, if there are types of characters (not) suited to the adventure, listing those up front.
- Referencing those choices at different points in the adventure, particularly when framing the scene for new conflicts.
- Giving players opportunity to see firsthand the unique outcomes/consequences of their decisions and actions.
I'm not sure what audience you are writing adventures for. Are you talking about commercial publication? I think that certainly imposes some fairly stringent additional constraints on adventure design.
When I am preparing adventures for my own game, I focus on open-ended encounters, and strong framing based upon PC build and subsequent PC play. Well-conceived situations that hook onto the players' preferences (as expressed via build and play of their PCs) means that I don't really need a "main story arc". Good situations lead to engaged play, and with engaged play the story arc will take care of itself.
Here's a
link to the Burning Wheel session I ran last weekend. To run that session I needed the players to build PCs with some clear motivations, plus my map of Greyhawk with its geography around the Wooly Bay. The rest "wrote" itself.
I'm not sure how I would present that as a module: maybe maps and some basic geography; NPCs with motivations and connections; some ideas for possible conflicts/developments; and advice on what sorts of PCs the players might build to hook into all this.
I think the 4e Neverwinter Campaign Guide is a possible model for this sort of thing. So is the Penumbra d20 module Three Days to Kill (though at a smaller scale).