blood, time, hunger, clean hands, and fast friends (on setting stakes)

(See also: mechanics for resource management, ICI doctrine, impact, attack the character sheet.)

Some crude steotypes you hear sometimes:

1. "OSR play is all bean-counting rations and torches and encumbrance"
2. "PbtA games are great if everyone is bringing the energy to make it sing, but if just one person is out of gas it falls flat"
3. "Modern D&D is just rolling dice until one is high enough that the DM shows you the next scene"
   

Some stereotypes are the opposite of true, but these all hint at something useful. And I think this has to do with the variety of consequences that the different styles work to make relevant. Failing forward and success at a cost are at the core of PbtA, and they lead to the "partial success" range being very fun if the player facing them has the creative energy to make them relevant. No resources is worth tracking in modern D&D other than spell slots and hit points, and you probably won't run out of hit points either.

For old-school play in the sense I care about, you don’t need death around every corner, or bean-counting encumbrance and gold, or strictly-kept time records (although I like some of those too.) But things like these allow the referee to attach costs to actions in a way that presents players with meaningful tradeoffs. Schematically, the conversational loop that undergirds play operates something like:

1. The referee describes a situation.
2. The player consider a response.
3. The referee communicates whether there is significant danger in this response, and if there is not, how long it would take and what other resources it might consume. ("Yeah, it looks like it would take about a dungeon round to hack your way through this wooden door") If there is no danger and no relevant time or resources to consume, they simply do the action.
   
4. If there is significant danger, then the referee either communicates that danger ("so you'll all be scaling this cliff in the storm, I'll have you roll a check at DC 20 each; for each failure you'll have to drop a backpack, and if everybody fails you fall to your doooooms") or notes where mysterious danger at least might be relevant ("okay, so to be clear, you're reading aloud from the tome, even though you don't know what it means, right?")
5. Players decide to go through with the action or consider another.
6. If they go through, referee describes consequences and hence the new situation.

Something (for instance, time) can't be a cost if it doesn't matter. The same applies to every other type of consequence:

• injury and risk of death
• encumbrance
• money
  
• rations 
• stress and mental health
  
• relationship with your patron deity
• relationship with various factions
• "are we the baddies?"
  

These don't necessarily need to be tracked quantitatively, and indeed many suffer from this. If PCs' moral status is supposed to matter, I don't think it requires a Humanity track like in Vampire and may often suffer for it (though I think in that particular game it works for it, in that it simply forces you to make a note of it when you hurt people, and the threshold for making that note of it gets more extreme as you descend further into being a monster.) But it does require that characters care about being good, and that they're thrown into situations that demand either compromise or lateral thinking. "Task: punch one million Nazis" might be an interesting challenge in any number of ways, but it's not morally interesting. 

People sometimes associate aspects of WotC-era D&D they don't like with streaming, but while to be fair tracking gold and encumbrance makes for extremely bad radio, I can think of at least one game that makes for great radio because of how obvious the risks and tradeoffs are when players face a decision: Delta Green. In DG agents have to worry about

• time (bad shit is going to keep happening or get worse without your solving it, otherwise you wouldn't have been called in)
• injury and death
• mental stress, especially how exposure to certain kinds of otherwise useful information
• the affect of all this on their relationships
• whether any particular pretty fucked-up action is morally justified by the harms they're trying to prevent

is Delta Green old-school? Whatever, it's dramatic because of (when played right) tough decision density.

Advice I'd consider generalizable: 

1. have a couple of different things you can reliably count on as a relevant cost, and keep track of it at the level of granularity and using the tools that make sense
   
2. vary up a bit what costs are relevant for any given session; give play variety by putting players in situations where time is suddenly rich where they're used to it being tight, or where rations suddenly matter even if they weren't worth tracking before
3. get your players primed to point out costs; this can align with their moment-to-moment interests if they know that you'll sometimes let them succeed at something they otherwise failed at if they propose a cost that makes sense.