That has come up in a few older threads in years past here.

For me it varies a-lot.

Usually I just have a basic map and some place names and a general idea of each. Usually some manner of start town and some common knowlege stuff for the areas local, and some rumours if the PCs ask around. Very low prep as with those I tend to let the players hie off wherever. I only prep more as I get a more solid idea what the players and their characters goals and current heading are. Initially I tend to not flesh out the common knowledges and rumours  until someone starts poking one or more and then flesh it out if they start taking more active interest in one. And in the background keep mental tabs on anything that might advance on its own if the PCs do nothing. eg: Theres rumours of gnoll bandits in the swamps, and its common knowledge that lizard men trade stuff with the local druids. I have a base idea that over time if not dealt with the bandits will raid a specific merchant passing through, and, the lizard men will if not helped with a problem of their own, miss their next trade meeting.

Other times I like to use random gen to see what get and work with that on the fly.

Or I might have a module and use that as a loose frame for the area.

I heard that the best thing to do is require players at end of session to tell you what they want to do next.

That only works if you end a session on a major decision point.

From experience most of the major player decisions are near the start. This is nearly allways where they hit on a course and set out to follow it. From there it is mostly a matter of working with that choice.
Did they head off to beat up bandits? Or did they wander out to investigate a request from the lizard men to deal with a problem? etc.

You can go two ways with this (potentially blending them):

1) Create a system that generates events procedurally, using things like weather, natural disasters, factions acting along predetermined (if not interfered with) goals, and then spinning out the consequences of such events throughout the game world, and then representing that to players so they can act, and then inject their actions back into the system so it can ripple and cycle through again to repeat the process, or

2) Create the agendas in play and just decide when something would happen based on what feels right, and what they "should" do. Then let the players interact with that and repeat.

It's not that different than a megadungeon or west marches.

Quote from: PencilBoy99;1125204I heard that the best thing to do is require players at end of session to tell you what they want to do next.

Quote from: Omega;1125215That only works if you end a session on a major decision point.

From experience most of the major player decisions are near the start. This is nearly allways where they hit on a course and set out to follow it. From there it is mostly a matter of working with that choice.
Did they head off to beat up bandits? Or did they wander out to investigate a request from the lizard men to deal with a problem? etc.

Maybe we just have different concepts of what constitutes a "major decision point", but my version of asking players at the end of a session what they want to do next (which I strongly endorse for any kind of sandboxy play) is to wrap up a session with "So, next time, do you guys want to head off to beat up bandits, wander out to investigate the request from the lizardmen to deal with their problem, or is there something else you'd rather do?"

This is also effectively a part of those Sine Nomine games where the players control a faction (such as Darkness Visible or Starvation Cheap), which suggest ending each session with a faction turn, then asking the players to take a look at all the faction-level events affecting their faction that turn and decide on which one they want their PCs to be directly involved in.

Didn't see anyone else post this absolute gem of fantasy sandbox advice. So, there it is!

I see your Hex Crawl and ante Path Crawl.

Quote from: Greentongue;1125284I see your Hex Crawl and ante Path Crawl.

Sir, I am in your debt. Yoink!

Quote from: PencilBoy99;1125121In terms of thin, I'm just trying to explain why, to date, prepping for a session by creating a Front has never been helpful. If you buy a published scenario, it will have tons of detail, possible scenes, NPC goals, etc. When I use something like that, read over the materials a bunch of times, and make notes and changes to it over a week on a daily basis, I can run a very effective session. When I've tried to wing it using a Front (which in most iterations of PbtA are just a few sentences including clock steps) I (not you) struggle into turning it into anything useful at the table.

Are there parts that you can improvise better than others?  If you have a couple of NPCs, can you extrapolate their minions?  If you have a location and NPC, does that prod an event?  Or any combination like that?  

The key to effective improvisation is to analyze which things you improvise well, which things poorly or not at all, and which things somewhere in the middle.  Then spend your prep where it counts.  

I'm not very good at doing certain parts of NPC improvisation (names, description, title, etc.) but I can do goals and plans off the cuff.  So I'll partially prep NPCs to have a list of shells to which I can attach more details as needed.  Doubly critical for my style, because I need to set the goals and plans as soon as the players encounter anything related to it, instead of waiting until they meet the NPC.  (Players find some minions raiding a caravan.  I know that they were hired.  I need some details right now for organization that hired them, because the players are likely to take a prisoner and extract hints.)

Quote from: Steven Mitchell;1125315The key to effective improvisation is to analyze which things you improvise well, which things poorly or not at all, and which things somewhere in the middle.  Then spend your prep where it counts.

Excellent point.

Well this inspired me to start work on a new sandbox - a hexcrawl sandbox.

[ATTACH=CONFIG]4237[/ATTACH]

Neat!

I def need to prep NPCs since I'm not great at improving them from nothing.

Quote from: PencilBoy99;1125342Neat!

I love Arr-Kelaan hexmapper! :cool: 25 years old & still the best!
http://breeyark.org/aar-kelaan-hexmapper-and-more/