Thomas R
United States
New York
flag msg tools
Avatar
This is something that's been on my mind off and on for a while, and I figured I'd toss out what fragmented thoughts I've currently got in the hopes that others may find them a useful jumping-off point for their own thinking, and in the hopes that someone else may be able to articulate some of this better than I.

Section 1: The problem with aggregate click analysis

The fundamental thing that I'm concerned with is that we often discuss cards and options within the game in terms of undifferentiated clicks as if every click in the game has equal value. Or, at the very least, as if we could average the value of all clicks across the course of a game and derive some sort of meaningful value with that number. Ideas such as "the runner spent 15 clicks on economic actions, and earned 45 credits across the course of the game, earning an average of 3 credits per click" are the kind of thing I'm talking about.

It is, in some ways, the basis we often use for comparative click-to-credit analysis. Easy Mark lets the runner spend one click and one card for three credits. Sure Gamble lets the runner spend one click and one card for four credits. Sure Gamble raises the average credit value of your aggregate clicks more than Easy Mark does, therefore Sure Gamble is a stronger card

I think that this approach may be marginally useful in some ways, but is ultimately flawed, and probably more misleading than useful in the majority of cases due to the amount of nuance and contingency that it hides. Sure Gamble is better than Easy Mark when you have five credits in your pool already, or if you only need the burst of credits when you have at least five credits already. So in order to really understand the relative value of Sure Gamble over Easy Mark, we need to know how often a player has to click for credits in order to get their pool back up to five so that they can play the Sure Gamble.

While one might attempt to do that in an aggregate manner, simply assuming that if you regularly click for credits in order to play Sure Gamble, it will result in you having a higher number of unassisted click-for-credit actions across the entire game, but it's not clear that that's really a valid assumption. Having a higher click-for-credit count than the "average" across all games may simply mean that your only economy in the deck is Sure Gamble, for instance. So even if you never find yourself below five credits to play it, you still spend a lot of time clicking for credits.

Furthermore, if you are at four credits and want to play Sure Gamble, the efficiency calculation goes from one card + one click = 4 credits to 1 card + 2 clicks = 5 credits. But if you're at two credits when you want to play Sure Gamble, then you're looking at 1 card + 4 clicks = 7 credits. This suggests that the value of Sure Gamble fluctuates based on timing factors within a game rather than having some absolute value.

My point here isn't that doing aggregate click analysis isn't useful. I think that it can be, in a number of ways. My point is that it's fundamentally flawed, and that the ways in which it is flawed mask a lot of important aspects of the game. Which means that when we do use aggregate analysis techniques, that we need to understand their limitations, and it also means that we need to have other methods of analysis that we can use to help offset those limitations.

Section 2: What should an alternative account for?

To complicate things further, I don't think that it's a terribly controversial idea that, over the course of a game, some clicks are more valuable than others. Or, perhaps more precisely, some turns are more critical than others, and clicks on those turns are more valuable than clicks generally. For instance, if the corp installs a card in a remote and advances it twice, the runner's next turn is of (potentially) more importance than a turn where the board state does not involve a potential 3-point agenda ready to be scored on the next corp turn.

I'm pretty sure that this "some turns are more important than others" behavior within Netrunner is what lies at the heart of Hollis' thinking around "work compression" in his post My Secret Love Affair With Jinteki: Personal Evolution. The "work compression" concept strikes me as a technique, or group of techniques, focused on increasing the cost for the runner to act on the critical turn. As I understand it, Hollis is talking about the corp spending clicks, credits, and cards over a series of turns with the goal being to raise the cost to the runner of stealing an agenda (or whatever) beyond what they can afford in a single turn (i.e. the runner needs to draw two cards, gain three credits, and make two runs).

I believe that any framework that we want to use to analyze Netrunner as a game must, somehow, take into account the varying value of options from turn to turn. I don't know precisely what the correct approach to doing that is, which is one of the things prompting this post, but here are a couple of things I think we need to be able to account for:

1. The value of fast-advance strategies that don't expose agendas in remotes at all for the corp.
2. The value of "surprise" credits for the runner, a la Stimhack, that throw off calculations of available resources during a critical turn.
3. Some way to evaluate, with some amount of accuracy, when (as a runner) to invest resources in the future (2+ turns away), and when to save so that if the corp drops an agenda, you're ready to run it.

There are probably another dozen key scenarios, but this post is getting long enough already, so I think I'll cut the list there.

Next steps?

I've laid out what I think is problematic with the way we often talk about the value of a click. I've hinted around some of the problems that a more nuanced analysis technique will need to provide answers for.

So, here I am, tossing all that out there, and I'm hoping that some of the smart people here on the board will have some ideas or suggestions about where to begin.
5 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Grish Noren
msg tools
Avatar
mbmb
I think the reason aggregate click analysis as you've described it doesn't work is because good economy =/= scoring points; it is merely a tool that enables it. An economy you don't sink much time in gives you more time to score points; but you're right getting the timing down is important.

My 2 cents remains that if you want to do click analysis, your scoring to clicks rate is probably more important for judging how good your deck is. How effective are you, and how much do you decrease your opponent's effectiveness, creating those turns you're talking about where its too late to stop your plan.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Dave Sutcliffe
United Kingdom
Manchester
flag msg tools
Avatar
I think you're back to front in section 1. It's not Sure Gambles fault you only have 3 credits - your game state doesn't influence what the card offers you in terms of incremental value, it's just a limiting factor on whether you can realise that value or not.

It's not helpful to say 'sometimes Sure Gamble is 3 clicks to gain 6', it's always either 1 click to gain 4 or a card you can't play.


In section 2 the timing issue is important and makes it difficult to rate cards that aren't pure economy without knowing the game state, but I think the answers to your 3 scenarios can be found in aggregate click analysis.

1) fast advance strategies replace the click/credit cost of creating a secure server by other means. The value of what that secure server would have required fluctuates depending on factors like the runner's ability to break servers, and the click/credit saving may be huge. This is strictly work compression as well if you can compress clicks you would have had to spend over be next three turns to create the server into an immediate fast advance window that also saves clicks.

2) stimhack creates an opportunity to gain 9 clicks worth of value, the rewards of which are already in front of you - the cards you played while you weren't gaining credits to run.

3) requires the two-tier click analysis I mentioned in the theory thread. Daily casts is a short term reduction in aggregate clicks that puts you behind, but with longer term rewards. It's often a catch 22 though - if you don't play Casts the Corp won't feel he has a click advantage to score an agenda, so you had the window to pay Casts. But if you play Casts into he window you close the window by offering the Corp a short term advantage.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Drake Villareal
United States
California
flag msg tools
Avatar
mb
I think you're talking about opportunity cost, mate.

However, when deckbuilding, we go "worst case/best case" and try not to worry about "fringe case". That's why our traditional analyses are working fairly well.

EDIT: I'm using "we" very lightly here.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
João Almeida
msg tools
Avatar
mbmb
anathomical wrote:
My point here isn't that doing aggregate click analysis isn't useful. I think that it can be, in a number of ways. My point is that it's fundamentally flawed, and that the ways in which it is flawed mask a lot of important aspects of the game. Which means that when we do use aggregate analysis techniques, that we need to understand their limitations, and it also means that we need to have other methods of analysis that we can use to help offset those limitations.


It is not a flaw to limit a model of evaluation in order to make it more simple as long as you're aware of the flaws (as you seem to be).

I might have misunderstood you, but it looks like you want a "complete model on how to evaluate cards", and there won't be such a thing because it would be too complicated to be useful.

You should use more than one method, because each one has its problems, but trying to put them all together in a "super model" would only make it worse.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
msg tools
mbmbmbmbmb
You might have seen this, but hey:

http://teamcovenant.com/flaminghito/2013/06/26/a-new-model-f...

It's a fascinating read, might be the solution for your problem.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Thomas R
United States
New York
flag msg tools
Avatar
gumOnShoe wrote:
I think the reason aggregate click analysis as you've described it doesn't work is because good economy =/= scoring points; it is merely a tool that enables it. An economy you don't sink much time in gives you more time to score points; but you're right getting the timing down is important.

My 2 cents remains that if you want to do click analysis, your scoring to clicks rate is probably more important for judging how good your deck is. How effective are you, and how much do you decrease your opponent's effectiveness, creating those turns you're talking about where its too late to stop your plan.


Clicks-to-points is an interesting point of analysis, but it feels somewhat incomplete. Possibly because it's still aggregate. And it's also probably incredibly match-up dependent. Are you getting your points from R&D primarily? Or HQ? Or remote servers? Archives? Or, if you're the corp, are you even primarily aiming for points at all, or are you a flatline-focused deck? Again, I think clicks-to-points is probably a useful bit of aggregate analysis, but I don't know that it's helpful at the lower-level point where we make tactical decisions.

Magicdave wrote:
I think you're back to front in section 1. It's not Sure Gambles fault you only have 3 credits - your game state doesn't influence what the card offers you in terms of incremental value, it's just a limiting factor on whether you can realise that value or not.

It's not helpful to say 'sometimes Sure Gamble is 3 clicks to gain 6', it's always either 1 click to gain 4 or a card you can't play.


In section 2 the timing issue is important and makes it difficult to rate cards that aren't pure economy without knowing the game state, but I think the answers to your 3 scenarios can be found in aggregate click analysis.

1) fast advance strategies replace the click/credit cost of creating a secure server by other means. The value of what that secure server would have required fluctuates depending on factors like the runner's ability to break servers, and the click/credit saving may be huge. This is strictly work compression as well if you can compress clicks you would have had to spend over be next three turns to create the server into an immediate fast advance window that also saves clicks.

2) stimhack creates an opportunity to gain 9 clicks worth of value, the rewards of which are already in front of you - the cards you played while you weren't gaining credits to run.

3) requires the two-tier click analysis I mentioned in the theory thread. Daily casts is a short term reduction in aggregate clicks that puts you behind, but with longer term rewards. It's often a catch 22 though - if you don't play Casts the Corp won't feel he has a click advantage to score an agenda, so you had the window to pay Casts. But if you play Casts into he window you close the window by offering the Corp a short term advantage.


I think I may agree with you about the Sure Gamble analysis, but perhaps for different reasons. I'm going to spend a bit of time thinking over your points rather than snapping off a quick response because I think there are some important nuances you're missing, but I'm not yet sure if those are real, or just bad intuition on my part.

OddCrow19 wrote:
I think you're talking about opportunity cost, mate.

However, when deckbuilding, we go "worst case/best case" and try not to worry about "fringe case". That's why our traditional analyses are working fairly well.

EDIT: I'm using "we" very lightly here.


Actually, and I'll admit that this is a minor quibble, we should probably use "most common case" rather than best/worst. Or, at least, some combination of them. I don't think you're disputing that, of course, but I figured I'd point it out.

Further, though, I feel like the "cases" can get complicated, and that's largely obscured. Because a scenario you could face is "I need to draw two cards to survive hitting a Fetal AI, and I need two additional credits (going from 4 to 6) to pay for stealing it after I get into the server, and I have to run to take it this turn or the corp will score it for the win". It's very specific, but it's going to be a case where having Sure Gamble is likely worse than Infiltration. Unless you have a Quality Time or Diesel in your hand, or you started your turn with a Wyldside in play. Or, or, or. It's all highly contingent.

Hraklea wrote:
anathomical wrote:
My point here isn't that doing aggregate click analysis isn't useful. I think that it can be, in a number of ways. My point is that it's fundamentally flawed, and that the ways in which it is flawed mask a lot of important aspects of the game. Which means that when we do use aggregate analysis techniques, that we need to understand their limitations, and it also means that we need to have other methods of analysis that we can use to help offset those limitations.


It is not a flaw to limit a model of evaluation in order to make it more simple as long as you're aware of the flaws (as you seem to be).

I might have misunderstood you, but it looks like you want a "complete model on how to evaluate cards", and there won't be such a thing because it would be too complicated to be useful.

You should use more than one method, because each one has its problems, but trying to put them all together in a "super model" would only make it worse.


This is a good point, and one I'm cognizant of. What I'm looking to do isn't to create a super model, as you put it. It's to create a model that excels at the types of analysis click aggregation models are bad at. I'm not looking for replacement systems so much as complimentary ones.

kissgz wrote:
You might have seen this, but hey:

http://teamcovenant.com/flaminghito/2013/06/26/a-new-model-f...

It's a fascinating read, might be the solution for your problem.


I remember reading it when it first showed up, but the link and reminder is appreciated. It's still a good read. I think that it might be in the direction of a solution, but isn't one in itself. It's really too generalized an idea, which I think is the point: it's a foundational framework, but it doesn't really help you make strategic or tactical decisions in and of itself. Or, really, to do solid analysis. It's a large list of things that affect the value of any given click, but no real actual valuation. Which is fine, but does mean that it's, at best, a start toward a solution.

2097 wrote:
I think I did account for these alternative values in my "scrambled rambling". Notably, I included (and praised) cards that had a CE cost, such as Self-Modifying Code. It is saving a lot in the long run by preventing drawing and digging.
On the corp side, it can be interesting to compare Biotic, SanSan, Psychographics and ToL.


Actually, that last comparison you mention really could be a good place to kick things off. Relative strengths and weakness of the major fast-advance facilitators. What situations is one better than another in? What situations is one worse? And when do they break even?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Calvin Moore
United States
Washington
flag msg tools
mb
kissgz wrote:
You might have seen this, but hey:

http://teamcovenant.com/flaminghito/2013/06/26/a-new-model-f...

It's a fascinating read, might be the solution for your problem.


Wow, thanks for this link. One of the more interesting things I've read in the last few weeks. This game continues to blow my mind.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
John F
United States
flag msg tools
mbmbmbmbmb
I think Flaminghito nailed it with that (and his other articles are great, too).

Raw math does matter, but context matters, too.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Richard Linnell
United States
San Diego
CA
flag msg tools
I really need a new badge, avatar, and overtext. GM me with any ideas....
badge
This is the seahorse valley
Avatar
mb
Can someone sum up the article? Can't read it at work...
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.