Do you purchase successes in your game?

since you can be successful in many areas it´s a bit of a vague question.

In Shadowrun 4 you can trade 4 dice for a success in any test. That's what I'm looking to see how many people do.

First, the rules don't say you can Buy Hits for "any test." It specifically mentions "large pools" (like a vehicle + armor) or non-stressful situation.

Second, I use it mostly for rolls in "downtime" kind of situations. Doing Matrix Searches in downtime, for example. I ask them their pool, look up the buying hits chart, and multiply that by their Skill to get their result.

Since you can figure you're turning what should be a 1 in 3 success rate into a confirmed 1 in 4 success rate, I generally allow it. 1 in 4 isn't rolling really well, after all. I've had PCs with a pool of 26 for resisting damage - it's not worth making them roll for a DV6 hit.

It doesn't usually sees use in stress situations, but Kozar has a solid point in that 1 in 4 isn't a good roll ratio. As a player I prefer to roll as normal because I almost always roll better than that. As a GM, my players have never asked to buy hits, but I allow it most situations.

I allow it in my games even in a high stress enviorment. Mainly as a way to lessen the number of rolls needed or are almost statistically pointless. Gee-you need t oresist 2S damage from drain. You have 10 dice to roll---why bother making the PC roll?

I use it in game, but GM discretion is key.
I use it to maintain pace versus excessive dice rolls (trivial things the character should be able to do easily)
and I use it to maintain pace versus excessive dice pools (largely vehicles).

Using it too much can give players with large pools immunity to dice rebellions though, so I use it sparingly.

In live games that I'm GMing, it's purely a downtime rule for me, particularly with Extended Tests. I don't want to go cross-eyed staring at hundreds of rolls and calculating how many hits show up. The only time I make an exception is when a character has Gremlins or other similar traits that affect how often a glitch occurs. Then you're stuck rolling for everything; sorry.

In play-by-post games that I'm GMing, I rarely use it even during downtime. It doesn't affect the pace of the game and unpredictability is always fun. It also leaves the burden of counting hits solely up to the player, so I don't have to go blind counting all of them up.

There are always exceptions, though. For absolutely trivial tests, like the aforementioned 2S Drain Resistance Test when you have a dice pool of 15+, I'll let a player just buy the successes. Especially if it's occuring outside of combat and there's no Gremlins-like quality at play.

I figure the only time it should be used (From player and GM perspective) is if success is guaranteed by at least 1 hit. If it only just hits the mark, it should be rolled for that chance of a bad roll, and if it doesn't hit that mark, well obviously you have to roll if you want to pass. Extended tests I would generally roll as well, due to the combination of bad rolls making it take longer (or shorter) and the possibility for glitches. And any opposed test should be rolled or else it becomes purely who has the bigger DP and luck is no longer a factor in the game.

Then of course there is gremlins. Now, if the person only has rating 1 gremlins, and a DP of 10+ a purchase might be considered, but otherwise always going to be a roll.

I try to keep tabs on the common dice pools (athletics, drive, perception) so that I know when we can skip die rolling. If everyone has 4+ dice of athletics then I can simply describe them going over a wall or other simple challenge. One guy has 12+ dice for perception tests so I don't bother to make him roll many of the tests.

For some number of dice 4k+c where k and c are natural numbers and c is one of 0, 1, 2, or 3, the odds of getting fewer than k successes against TN 5 are (2/3)^c*((2/3)^4)*((2/3)^3)^(k-1); some typical values for the best case (where c=0) are k=4->19.7%, k=8->5.8%, k=12->1.7%, and k=16->0.5%.

This isn't quite as revealing as one might think, as presumably there's some fixed number of successes that is actually necessary most of the time and so it's actually a matter of "can I guarantee that I meet that", but other than that buying successes is pretty much a sucker's game.

Here's the Python code I whipped up to do the calculation, in case anyone cares:



Edit: note that that's with true division; you either need to run it with Python 3000 or add a "from __future__ import division" before it.

EDIT2: THE ABOVE CALCULATION IS INCORRECT FOR d≥8. A fixed version will be posted below, please don't rely on the above

~J

I honestly hate relying on averages when talking about such small numbers. This mostly stems from the fact that I roll like shit. I've gotten critical glitches on a 14+ dice pool multiple times...

On the plus side I absolutely kick ass at Yahtzee.

By rolling all 1s?

The odds of critical glitching on 14 dice must be phenomenal. I mean the odds of not getting any hits alone is (2/3)^14 and then you have to factor in the odds of getting at least 7 1s....

That's why I'm not giving averages, I'm giving probabilities

(Averages are just number of dice divided by 3)

Edit: odds of a critical glitch on 14 dice are (4^7)/(6^14), or alternately (1/6)^7*(2/3)^7.

~J

I use trading for successes all the time, simply because I tend to keep running Perception(mundane & Matrix)/Assensing checks for my players.

Glitches are a very rare occurance in my game due to the prevalance of Guard, possibly only due to downgrading a Crit Glitch. The only character I can think of that is likely to have Guard (from a teammate's spirit) is the Astral Hazing Technomancer Troll.

I use it whenever it feels "right".

Such times:

• Whenever speed is important (due to the pacing of a scene).
• When an outcome is obvious.
• When no successes would ruin my narrative.

You're right - it's normally a 3.33333(repeating) chance. The extra % needed to round out to a 1-in-4 instead of a 1-in-nearly-3 guarantees the hit.

I frequently buy successes. I figure out my dicepool for soaking damage and calculate at what amount of damage received do I have to actually roll my dice, and I also calculate how many successes I can buy with my Spellcasting Tests, and what drain (per spell) I can buy my Drain Tests with, and find the most efficient Force spell with how many hits I can buy the drain of. For instance, I had built a 400 BP elven mage (back when I was worse at this) that could buy 6 successes on a F7 Stunbolt and auto-soak the drain. He could also buy 6 successes on a multicasted 2x F6 Stunbolt.

You haven't taught your players to either: A) Be honest, or 2) count successes themselves?

No. In live games when the dice are all on the table, I just naturally set about to help in order to a) speed it up and b) just be nice. And despite my apparent personality on forums, I'm not exactly some evil douchebag. I just seem to come across that way on forums for some bizarre reason. Instant messengers, chat rooms, phone calls, face-to-face... don't have that problem anywhere else. It's so odd.

Maybe on forums your ID comes to the fore? And looking at my above post it should have been:

It's amazing how changing two words like that changes the meaning of the entire sentence.

Similar to Ayeohx and the others. I allow it for downtime, no brainers and when speed is needed to keep things moving.

As a mechanical benefit, I often allow buying hits, even with dice pools of just 4 during high stress situations. This way there is a concrete trade off of benefits. A dice pool of 4 has 1 confirmed success while a dice pool of 5-7 has the same 1 confirmed success. That means you might get more mechanical benefits if you did not simply go 1/5 on Attributes and 1/4/6 on skills.

While many GMs like the statistical variation and chance of failure, I am particularly fond of buying hits when one can buy enough hits to succeed, and there is no benefit for extra hits. In that situation, making the player roll (while common) is unpleasant. They get hurt if they roll badly, and get no bonus if they roll well.
One complication is if there are not very many dice. 4 dice not only have a high chance of failure to get even one hit, but they have a noticable chance of critical glitch. This drops off very sharply at 5 dice. By 8 dice you are looking at half a percent chance of a critical glitch, and only 3% chance of getting a regular glitch. If the player can only use 2 successes, why make them roll?

Yours,
Joel

PS: SR dice rolling has a very high statistical distribution, which is why this is tricky. The odds of failure, if you make them roll, are fairly good even with quite a few dice.

Interesting point. I may consider a house rule that you can only buy successes if you're rolling at least 8 dice; Sounds like the math works out and it's a keen reward for having really huge pools.

What? Did you mean variance?

Not really. I posted above the best-case probabilities for getting fewer successes by rolling, and the probability starts quite small and goes down from there. Presumably if buying successes is even on the table to begin with the number you can buy is good enough, so that means that the odds of failure aren't really "fairly good" by any reasonable measure.

Edit: except in the four-die case, I can see a nearly 20% failure rate being called "fairly good". Still, it drops quickly; 5 dice is ~13%, 6 dice ~9%, 7 dice ~6%.

~J

Sorry Kagentenshi. It's been a long day. Yes, I meant variance or standard deviation (the square root of variance).

The effect of that high variance is that with 8 dice, you have a 19% chance of not getting 2 hits.
With 12 dice, you still have a 17% chance of not getting three hits.
Normally, this is offset (in terms of rewards to the roller) by the fact that you also have a chance of getting more hits.
But when extra hits do you no good, you still have the downside, with little upside.
(Many systems try to make sure that any roll with a downside has some upside to it.)

Yours,
Joel

Happens to everyone, just wanted to be sure what you meant.


Gah, you're right; my formula above gives incorrect answers for d≥8. Serves me right for only checking on small cases (finally a circumstance in which a form of test-driven-development might have served me well).


This is indeed the case that throws off the usual reasoning, as I noted above.

Anyway, more when I fix whatever the error was, probably while sitting at the airport later today.

~J