Recommend
 
 Thumb up
 Hide
13 Posts

Leaving Earth» Forums » Variants

Subject: Outcome Cards - never be 100% certain of success rss

Your Tags: Add tags
Popular Tags: [View All]
Steve Baines
msg tools

So, I've only just got this game, and think I'm going to like it a lot.
But, straight away there is something that I feel I'd like to change...


I was considering using the following rule:

You can only pay to remove failure outcomes, not successes.
The removed failure card is immediately replaced with another randomly drawn outcome card. (So you always have the same number of outcome cards).





Reasons for suggesting this rule:


I very much like the following aspects of the outcome cards system:
1. Any particular technology has an innate level of reliability, which you don't know at the start, and which is different for each system.
2. You can spend money to improve the reliability over time.

However, I'm not keen on the way you can buy 100% reliability, and know that you have 100% reliability.

Paying to remove a successful outcome card also has the odd effect of making the technology become definitely less reliable whilst on its way to becoming perfect, and this feels very odd to me.

E.g. Suppose you have 1 success, and 2 fail cards, hence 33% chance of success, and you turn up the success card. If you buy it off, your tech (even though you don't know it yet) is now guaranteed to fail every time until you pay to remove the 2 fail cards. It is then 100% reliable, and you know this for a fact.
So as you pay off each of the cards in turn your reliability goes:
33%, 0%, 0%, 100%

If you were only allowed to buy off failures, this would avoid this - tech would only ever get better, but I can see that you'd then get the strange situation that if you started with 3 failures, then for $15 you'd be able to buy yourself a known guaranteed 100% reliable system,
but if you started with 3 success cards, you'd already have a 100% reliable system, but never be able to know that for sure.



Effects of this rule change

The deck contains 60 success cards, 15 minor failure cards and 15 major failure cards. So, ratio of 4:1:1

With this rule:
1. You cannot (and there would be no reason to) pay to remove a success card - you don't fix what isn't broken!

2. If you paid to remove a major failure, you'd have:
4/6 (67%) chance of totally fixing it - replacing it with a success card
1/6 (17%) chance of reducing the problem - reducing from major to minor failure.
1/6 (17%) chance of having had no effect.

3. Similarly, if you paid to remove a minor failure, you'd have:
4/6 (67%) chance of totally fixing it - replacing it with a success card
1/6 (17%) chance of having had no effect.
1/6 (17%) chance of making the problem worse - increasing from minor to major failure.

Consequences of the rule:
1. Although your technology can still become 100% reliable, you can never be certain that that has happened, since you can never be certain that a failure that you paid to remove wasn't replaced with another failure. (Thematically - your 'fix' didn't fix the real problem.)

2. If you have a minor failure, you have to consider whether to risk making things worse by trying to fix it. E.g. a rocket that produces no thrust is less bad than one that explodes. (Thematically - you've changed the design to fix a problem, but there's a chance you inadvertently introduced a worse problem.)


Thoughts on this idea very much welcome!

Cheers - Steve
5 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Robert Manning
United States
Sunnyvale
California
flag msg tools
badge
mbmbmbmbmb
Overall, I like the concept. But there is the practical issue of the number of success cards in the deck. If the advancements always have their full quota of cards each player could have 28 success cards across their advancements. As the success cards migrate from the the main deck to the advancement cards the ration of failures will dramatically increase to the point that only failures remain. In a four player game they could run out of cards.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Chris in Kansai
Japan
Otsu
Shiga
flag msg tools
Avatar
mbmbmbmbmb
To reintroduce an element of risk on proven tech I think I'd rather just roll a die, say a d20.

On a 1 result your rocket blows up, on a 2 you have a minor failure.

Want more risk? Use a d12 or other, or increase the number of failure results.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Larry L
United States
Stockton
California
flag msg tools
He who games with the most dice wins.
badge
I like things arnmod
Avatar
mbmbmbmbmb
It sounds like an interesting variant. I haven't played non-solitaire enough to know what that would be like. If I ever feel like the vanilla solitaire is played out for me, I might try it.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Michel Kangro
Germany
Bonn
NRW
flag msg tools
mbmbmbmbmb
How do you balance the increased risk - because you can never achieve 100% reliability?

In essence, how do you make testing cheaper?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Steve Baines
msg tools
@rmanning:

A very good point, thank you!

Something that I didn't mention was that I was considering writing a small app to handle the outcome cards for this game.

I was thinking that this would be particularly useful with this variant: Since you can never remove the outcome cards, you always have to go through the shuffle and choose process, and I can see that that is going to get tedious - especially for solo where it's harder to shuffle 2 or 3 cards convincingly for yourself.

A helpful consequence of such an app is that rather than simulating the deck exactly, if using this rule I could just keep a 4:1:1 ratio on the card likelihoods (effectively simulating an infinite deck), which would solve that practical issue.


So, due to the point you raised, then I think my suggestion is indeed broken for multiplayer games, but maybe still viable for solo play. With an outcomes app, it should be workable for all.

(Note that in general I am against getting apps involved in board games - I'd rather be playing a game than fiddling with a phone/tablet, but for the case of handling outcomes I'm sure it would be a good time saver)

 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Steve Baines
msg tools
@mideg

I wasn't planning to. So, this would make for a harder game.
If it turned out to make it too hard then maybe an increase in budget to allow for more testing per year?

Note that I'm thinking mainly from the standpoint of solo play, as that is what I'm going to be doing most often.
Since the solo play rules already suggest going for a hard or very hard game (in terms of missions), then an increase in the general game difficulty could help make easier missions more relevant in single player.

Edit:
To be clear, you can still achieve 100% reliability, but you can never be certain that you've achieved it. The idea is to get away from "another test would be a waste because I know it is now perfect" and more towards "It seems to be reliable, but do I really trust it yet?"
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Barry Miller
United States
Saint Charles
Missouri
flag msg tools
badge
Avatar
mbmbmbmbmb

It'd be interesting to discuss why Joe designed the Outcome mechanic the way he did. I'm not entirely sure, but my guess is that the current mechanic is a way to abstract a nation's ability to achieve statistical "GO" reliability. Or, IOW, the ability to increase program reliability to a level where it statistically mitigates the cost of risk, is what I think he captured in an abstract way.

Determining the cause of, and fixing a failure to be reliable in the future is easy. That's why failures cost only $5 to get rid of. The challenge though, is finding the reliability issues that haven't failed yet. Now, that's expensive. You need to fix those issues (the components and systems that haven't failed yet) in order to achieve the level of reliability I discussed above. Again, getting your program to this state isn't cheap, which is why removing a success costs more than removing a failure (or at least that's my supposition).

The end goal of course, is achieving that level of reliability which can be counted on, statistically. What do I mean by that? Well of course you can never achieve 100% system reliability. Either form, fit, or function will fail at some time. Just as it's mathematically impossible to achieve a 100% probability when evaluating the outcome of multiple events, it's physically impossible to achieve 100% reliability. (But you CAN possibly get to 99.9%).

So, what should the level of reliability be, that our program should expect to enjoy after sinking millions and millions of dollars into reliability research?
Let's look at the Space Shuttle, as perhaps the best known system with enough occurrences to draw relevant statistics...

135 Operational Missions
5 Test Missions

The failures:
2 Loss of ship after launch (STS-51-L and STS-107)
3 Launch Aborts (STS-41, STS-68, STS-133) (STS-38 & STS-39 could be considered aborts as well, though their problems were caught before launch day)
2 Mission Aborts due to various life support problems (STS-2, STS-5)
1 Mission Abort due to mechanical failure (STS-83)

Let's leave the Test Missions out of the discussion for now (though to get pedantic, they should be included).

So first, let's consider the two obvious and tragic failures, STS-51-L and STS-107... The loss of both Challenger and Endeavour would be, in game terms, considered "Major Failures". Thusly these failures resulted in a 98.5% Reliability Rate against a Major Failure. (135 missions w/ 2 Failures).

Now, the six aborts would be again, in game terms, considered "Minor Failures". These failures resulted in a 94% Reliability Rate against a Minor Failure. (135 missions w/ 6 Failures).

-----

So where am I going with this? I believe Joe, through the Outcome Card mechanic and rules, is trying to abstractly give the player a way to achieve reliability rates that are 95% or above - a statistical "GO" - if the Space Shuttle Program is any measure. Otherwise, you'll always be testing and getting nowhere. You'll note that as the Space Shuttle program matured, the failures were occurring less. For instance, after STS-68, the "Minor Failure" reliability rate was 97%!

Again, achieving that 97% rate cost lots of money! It required far more effort than simply correcting problems discovered during earlier flights (hence again, why a "Success" costs double to get rid of). So once your program gets to that point in the game, how do you model the real-life confidence factor? IOW, how should Joe model the fact that you should rightfully be comfortable in attempting your mission when you've reached a certain level of program reliability (the statistical "GO")? Especially reliability that you've paid dearly for?

I apologize for my rambling as it's difficult for me to capture what I think Joe is abstractly trying to achieve with the Outcome Cards.

To summarize, I think the Outcome Card mechanic is to provide a means of building reliability into your program - at a cost, and with some failures along the way. And if you achieve the highest level of reliability, then you should be abstractly rewarded with a level of comfort that your missions are going to succeed. From the Space Shuttle experience, that level of comfort became 94%.

Sure, in our games we could make it so that the reliability level is always uncertain, or low enough to be uncomfortable. But I think doing so will counter what the Outcome mechanic is trying to achieve, and is also counter to what real-life reliability programs strive for.

On the flip side, one idea mentioned above was to use a D20. I could actually see that... I.e., roll a D20 after you've discarded all your Outcome Cards. A "1" results in failure. A "2" - "20" results in success = 95% reliability rate. Gee... how about that?

My two cents, FWIW.

P.S. I'd like to hear what the real reason is, for the Outcome Card mechanic. All the above rambling is only my take!

And Oh yeah, all the Space Shuttle stats I tossed around above are taken only from the notable occurrences. I'm sure there are many other failures, but the stats above reflect the notable failures that best mirror the "Major" and "Minor" failures of the game.

6 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Michel Kangro
Germany
Bonn
NRW
flag msg tools
mbmbmbmbmb
So, to summarize the text above, with extensive costs a reliability of about 95% or higher should be achievable?

Well, if I test vigurously (extensive costs) and after some time, I got many successes and then one failure. I do then replace that failure with a new card.

I will then have a very high (let's make it a 100% just for the math here, any number arbitrarily close to that could be achieved given enough tests/money) confidence that the other two cards are successes. Adding a failure with 67% chance of success to that gives me an overall chance of success of about 90%.

That's close.

Then again, once an app is introduces anyway, we could just change the system completly:
Any tech starts with a reliability of 67%. For each success drawn, the chance of failure halves. For each failure, the chance of failure halves if the player pays for it. In both cases, additional money may be spent to half it again.

Exchange "half" for any number below 1 to fine tune that apporach. You will never reach 100%. You will reach comfortable numbers. Disadvantage is the clear knowledge of the probabilities, then again, in principle, we do know the probabilities of the cards lying there being successes right now, too.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Steve Baines
msg tools
Thanks for the comments, some interesting points coming up here!

bgm1961 wrote:
And if you achieve the highest level of reliability, then you should be abstractly rewarded with a level of comfort that your missions are going to succeed.


I see this as the key issue - the difference between 'a level of comfort' and 'complete confidence'.

bgm1961 wrote:
Sure, in our games we could make it so that the reliability level is always uncertain, or low enough to be uncomfortable. But I think doing so will counter what the Outcome mechanic is trying to achieve, and is also counter to what real-life reliability programs strive for.


Of course, in real-life we are striving to achieve 100% reliability, but we know we will never get there.

bgm1961 wrote:
On the flip side, one idea mentioned above was to use a D20. I could actually see that... I.e., roll a D20 after you've discarded all your Outcome Cards. A "1" results in failure. A "2" - "20" results in success = 95% reliability rate. Gee... how about that?


Using a D20 is undeniably an easy way to introduce a small chance of failures, and thus remove the certainty. The limitation it brings is that you know the failure chance is 5%, and there's nothing you can do about it.

With my suggested rule, you can still achieve 100% reliability (as in the standard game), but the difference is that you can't know you've achieved it. Every success will increase your confidence, and even when an unexpected failure occurs, you can buy it off, the odds are that your system got more reliable again.

So, the more you test, the more confident you'll become, and the lower the likelihood will become that you'll see a failure in future. But you can never be certain that you won't get a failure.

mideg wrote:
Then again, once an app is introduced anyway, we could just change the system completly:


Indeed, any system at all for modelling failure could be used. But I would rather not stray too far from the original game. The idea for an app for this game was purely as a convenience tool to make outcome handling faster. After the issue of card migration was pointed out (due to the limited deck size), it occurred to me that the app could very easily avoid this problem by essentially having an unlimited outcome card deck.


mideg wrote:
Disadvantage is the clear knowledge of the probabilities, then again, in principle, we do know the probabilities of the cards lying there being successes right now, too.


Well, we don't really know the probabilities, and I think this is the important point.
E.g. if you've had 10 successes in a row, you can be pretty 98% sure that you've actually got 3 successes, in which case all future runs will also be successes. But there's that 2%... You might actually still have a fail in there. If you do, your next run has a 33% chance of failure (as did your last 10).


When people involved in the Apollo missions, especially the astronauts, talk about safety, it was pretty clear that they knew they were taking some big risks. There certainly wasn't certainty!


 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Josh Zscheile
Germany
flag msg tools
Avatar
mbmbmbmbmb
After having thought about this topic since my first play of the base game, over time and while reading more and more suggestions, I got pretty confident that I do not want any changes to the outcome mechanism. This is a game after all, and the way it works now you can either take risks or spend resources and time to lower them bit by bit until there is no risk left.

I do not want my Saturn Station mission to fail in 1982 because I rolled a 1 on a Juno rocket whose technology I used for decades and dozens of times before. I do not want that kind of randomness in the game, because I fully realize I am a Euro gamer and not one that takes this kind of event as fun part of the story (I'd get really angry at that die...).

I do however understand people that want more 'realism' in the game. I am not saying you should not try different additional mechanisms. I just know this will not be for everybody.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Michel Kangro
Germany
Bonn
NRW
flag msg tools
mbmbmbmbmb
sjbaines wrote:
mideg wrote:
Disadvantage is the clear knowledge of the probabilities, then again, in principle, we do know the probabilities of the cards lying there being successes right now, too.


Well, we don't really know the probabilities, and I think this is the important point.
E.g. if you've had 10 successes in a row, you can be pretty 98% sure that you've actually got 3 successes, in which case all future runs will also be successes. But there's that 2%... You might actually still have a fail in there. If you do, your next run has a 33% chance of failure (as did your last 10).


I disagree. Probability is a lack of knowledge, in most cases. Given that I've had 10 successes and thus I am 98% certain that I got 3 successes, then I KNOW that my probability of another success is about 99,3%, more or less. (Either I do have 3 successes or I have the chance to draw one of the successes.) I do know my chances exactly.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Frazer Eden
United Kingdom
Cheshire
flag msg tools
badge
Avatar
mbmbmbmbmb
Perhaps if you get one success you can add another success to your deck for 10K? This would forever increase the chance that you will have future successes.

Thematically this works quite well too, as you invest in safety measures but there may still be that hidden issue you haven't spotted yet.

You couldn't play it as more than a 2 player game or you would need two copies due to the lack of cards/
3 
 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.