That part where you mentioned the good RNG at the end... if you don't have crew with high enough max proficiency, then you won't be able to make those checks at the end regardless of RNG. That's why I mentioned the special staffing thing. There's a difference (though subtle) between maximizing for the longest likely voyage and the most potential to hit 12 hours. That's why I say it would need to be more nuanced than what you described. Though I certainly thank you for the explanation! That was some insight that I needed.
That's exactly what I wanted. Let you pick between longest average length, longest likely length(TBD define likely), or highest chance of hitting X hours. All datacore needs to do is give you that option and change what it optimizes for. The question of high/low proficiency and higher stats vs trait matches will be handled automatically as it tries to pick the best crew given what you are targeting.
great work and thanks for sharing this! Excellent write up as well.
What I really like about this, is that it adds an extra element of gameplay to the game. A more tactical element where you have to consider what risk you want to take and what goal you want to achieve.
To the developers I would suggest to share these mechanisms as it would be an easy way to make the game more interesting.
@Bylo Band you suggest to avoid crew with a wide proficiency range for a more reliable outcome. But if I understand it well, you should target crew with the highest minimum score (i.e. base + min prof.). As that is what determines your ‘guaranteed‘ outcome.
More specific, a crew with a high minimum and a wide proficiency on top of that (like Braxton) is still better than a crew with a lower minimum and a narrower proficiency range.
@Bylo Band you suggest to avoid crew with a wide proficiency range for a more reliable outcome. But if I understand it well, you should target crew with the highest minimum score (i.e. base + min prof.). As that is what determines your ‘guaranteed‘ outcome.
More specific, a crew with a high minimum and a wide proficiency on top of that (like Braxton) is still better than a crew with a lower minimum and a narrower proficiency range.
This is undoubtedly true. I am fortunate that I was in a position bench crew such as him to test this out on my own by chasing 10 hour voyages and still have reserves strong enough to get the job done.
There is a part 3 or sorts. I do not recall what the original Part 3 was, although it may have morphed into the next installment of this voyage project. Part 3 though will be structured much more like an essay/lecture, similar to this one, and incorporates a third thing to keep track of for voyages, and will conclude with a breakdown of the new voyage system I developed recently that has been achieving extremely promising results from the limited number of people who I've shared it with to date
As far as actually designing a test to get hard data for this Part 2 stuff, it will be very tough, and a big part of why we kept it informal in our "lab" was we couldn't agree on a way to proceed, and if a handful of people couldn't agree, there seemed little hope in opening it up to hundreds of people. It would have been chaos! Under perfect circumstances, the best test method we devised would have been to send an identical voyage twice, with one of them going with a lot of High Proficiency crew and then the second with Low Proficiency crew, and then repeating this dozens of times. This of course has numerous problems!
The first problem is pretty pedestrian but important nonetheless, there is no way to lock in a voyage to ensure you get the exact same setup twice in a row. This often happens naturally, but if a voyage comes back after the voyage reset time there is always a chance the parameters will change, at which point you run the risk of having lots of single voyages without a second voyage for comparison. The second problem was trying to control for varying levels of voyage strength. There was no way to send out the same voyage twice (ie same feature skills, ship trait, and seat bonuses) with different crew and keep the scores the same, so then how would you even go about comparing the results? That is when we hit on the idea to try and compare the results with pre-voyage duration predictions but that was also problematic because we couldn't find a universally accepted method for doing that...
...and that was assuming we could even come up with lists of crew to fit our definitions! This is basically how it went, we had the ideas, but couldn't really figure out a methodology that we felt would pass muster here. So we did what we did, started paying attention to proficiencies a lot more, taking a lot of screenshots, comparing the pre-voyage numbers to finish times, and the pattern started to become apparent. At that point, we didn't really feel we needed to understand the formula to explain why the apple fell from the tree, we were content to be aware that the apple fell from the tree. if that makes sense. I stopped using crew like Caretaker, Red Angel, Locutus, Gary Seven, Captain Braxton, Guinan etc, and my voyages stopped crashing early. We learned in Part 1 that it is most efficient from a resource harvesting perspective to recall immediately after the last dilemma you can achieve (wording it that way since for some that is only 2 hours, for others 10, and everyone in between) so I shifted focus from trying to run my voyages as long as possible to running them to 10 hours then recalling, and found that once we started this project and I no longer used these High Proficiency crew, my voyages rarely failed to hit 10 hours, and that was the lightbulb moment for me.
Of course the reverse is also true. For people chasing 12 hour voyages I think this idea is equally powerful, as it sort of unlocks a big portion of the RNG and will let folks tinker with their crew to try and get the best chance for a long voyage, and that is exciting too!
Just remember when you start looking over your cards with this idea in mind, a portion of every card's base score is hidden in the proficiency range, so be mindful. For example:
Locutus of Borg (Diplomacy): 847 + (399-783).
Locutus' base Diplomacy is not 847, it is actually 1,246, because a minimum of 399 will be added to any Diplomacy check. So for RNG purposes, his stats are better read like this:
Locutus of Borg (Diplomacy): 1,246 + (0-384). So even though his max proficiency number looks scary, his actual range is only 384, which IMO is still on the high side, but it is not as severe as it appears at first glance.
This is an excellent read. Thank you for doing such a detailed breakdown.
The ability to have different options and tinker with my crew usage is part of the reason why my main goal is to collect all possible crew in the game. Well, that and the fact that I'm one of those 100% completionist kind of people I love the chance to be able to figure out a setup that gives me an edge.
It's stuff like this that gets me excited about the potential for this game. I'm so sick of everything being so cookie cutter, be it walls in the gauntlet, the same builds in the arena, or just the wash, rinse, repeat nature of a lot of the events. Give me strategy, tactics and something that forces me to think and perform analysis like this any day over the normal autopilot gameplay.
I have to add when you fail a hazard you lose 30 AM. if you succceed you get 5 AM. The heavy weighting for failing will mean there is more cost to having a wider Range because of proficiency . So it is not an even one to one exchange. I would say the base stat is 5% more important than the proficiency average. But this is just my opinion. Trying to calculate this really is dependent upon the crew and voyage choices.
But this weighting i am talking about really is only meaningful if you do not intend to extend the voyage. If you are going to extend to the point you exceed the range of your crews proficiency, then you will get the full benefit of your crews proficiency.
Edit: with my second paragraph, it then implies the weighting should only matter to the primary and maybe the secondary skill. Depends if your voyage will exceed the range of that second skill or not.
That part where you mentioned the good RNG at the end... if you don't have crew with high enough max proficiency, then you won't be able to make those checks at the end regardless of RNG. That's why I mentioned the special staffing thing. There's a difference (though subtle) between maximizing for the longest likely voyage and the most potential to hit 12 hours. That's why I say it would need to be more nuanced than what you described. Though I certainly thank you for the explanation! That was some insight that I needed.
That's exactly what I wanted. Let you pick between longest average length, longest likely length(TBD define likely), or highest chance of hitting X hours. All datacore needs to do is give you that option and change what it optimizes for. The question of high/low proficiency and higher stats vs trait matches will be handled automatically as it tries to pick the best crew given what you are targeting.
Is there a way to do that, though? Like if you put in 10 or 12 hours for desired voyage length, does it run a bunch of simulations to see what combination has the best chance to get there?
If datacore is currently optimising for average score, all it would need is 2 additional options: 1 for optimising for minimum score (base + min prof.) and 1 for optimising for max score (base + max prof.). The latter should give you the chance of the longest voyage possible with your crew.
I would just like to add in here that it was wild working on this. I feel like the forum has been ALL around this idea for over a year, finding dots and connecting some but that the final dot was missing. It was really thrilling to be a part of the group that found the dot and started making new connections
Regarding Part 3, work has already begun on compiling information for the write up. The final system is already basically done, we just need to go back and fill in the temporary steps we used to get there with ones more stable More work needs to be done to accurately explain and communicate the final piece of the puzzle, how it all relates with everything else we already know, and then to wrap it all up in a neat package. We have begun to craft a resource/tool that we hope will be both useful for folks (either directly or inspire somebody to take it and make it better!) but also to help explain how it all works.
That part where you mentioned the good RNG at the end... if you don't have crew with high enough max proficiency, then you won't be able to make those checks at the end regardless of RNG. That's why I mentioned the special staffing thing. There's a difference (though subtle) between maximizing for the longest likely voyage and the most potential to hit 12 hours. That's why I say it would need to be more nuanced than what you described. Though I certainly thank you for the explanation! That was some insight that I needed.
That's exactly what I wanted. Let you pick between longest average length, longest likely length(TBD define likely), or highest chance of hitting X hours. All datacore needs to do is give you that option and change what it optimizes for. The question of high/low proficiency and higher stats vs trait matches will be handled automatically as it tries to pick the best crew given what you are targeting.
Is there a way to do that, though? Like if you put in 10 or 12 hours for desired voyage length, does it run a bunch of simulations to see what combination has the best chance to get there?
Yes. It runs a bunch of simulations now, averages the results, and uses that number as the length to compare against other crew options. It could also run a bunch of simulations, look at the number that passed 10 or 12 hours, and use that to compare against other crew options. In general you can't try to optimize for 2 things at once, but if you want to optimize for highest chance at X hours rather than best average length, that's relatively easy to change.
Comments
That's exactly what I wanted. Let you pick between longest average length, longest likely length(TBD define likely), or highest chance of hitting X hours. All datacore needs to do is give you that option and change what it optimizes for. The question of high/low proficiency and higher stats vs trait matches will be handled automatically as it tries to pick the best crew given what you are targeting.
What I really like about this, is that it adds an extra element of gameplay to the game. A more tactical element where you have to consider what risk you want to take and what goal you want to achieve.
To the developers I would suggest to share these mechanisms as it would be an easy way to make the game more interesting.
@Bylo Band you suggest to avoid crew with a wide proficiency range for a more reliable outcome. But if I understand it well, you should target crew with the highest minimum score (i.e. base + min prof.). As that is what determines your ‘guaranteed‘ outcome.
More specific, a crew with a high minimum and a wide proficiency on top of that (like Braxton) is still better than a crew with a lower minimum and a narrower proficiency range.
This is undoubtedly true. I am fortunate that I was in a position bench crew such as him to test this out on my own by chasing 10 hour voyages and still have reserves strong enough to get the job done.
This is an excellent read. Thank you for doing such a detailed breakdown.
The ability to have different options and tinker with my crew usage is part of the reason why my main goal is to collect all possible crew in the game. Well, that and the fact that I'm one of those 100% completionist kind of people I love the chance to be able to figure out a setup that gives me an edge.
It's stuff like this that gets me excited about the potential for this game. I'm so sick of everything being so cookie cutter, be it walls in the gauntlet, the same builds in the arena, or just the wash, rinse, repeat nature of a lot of the events. Give me strategy, tactics and something that forces me to think and perform analysis like this any day over the normal autopilot gameplay.
But this weighting i am talking about really is only meaningful if you do not intend to extend the voyage. If you are going to extend to the point you exceed the range of your crews proficiency, then you will get the full benefit of your crews proficiency.
Edit: with my second paragraph, it then implies the weighting should only matter to the primary and maybe the secondary skill. Depends if your voyage will exceed the range of that second skill or not.
Maximise chance to get to x (10) hours ( not max) and minimise chance of failure.
Here you want to balance/minimize lower proficiencies.
Get to that elusive 12 hour voyage:
take risks with more high proficiencies.
Abandon if not at certain boundaries in earlier dilemmas.
Improve your longest voyage without taking huge risks or abandoning when you can get to next dilemma.
Each needs its own different staffing strategies.
And behold/cite strategies
Is there a way to do that, though? Like if you put in 10 or 12 hours for desired voyage length, does it run a bunch of simulations to see what combination has the best chance to get there?
Regarding Part 3, work has already begun on compiling information for the write up. The final system is already basically done, we just need to go back and fill in the temporary steps we used to get there with ones more stable More work needs to be done to accurately explain and communicate the final piece of the puzzle, how it all relates with everything else we already know, and then to wrap it all up in a neat package. We have begun to craft a resource/tool that we hope will be both useful for folks (either directly or inspire somebody to take it and make it better!) but also to help explain how it all works.
Yes. It runs a bunch of simulations now, averages the results, and uses that number as the length to compare against other crew options. It could also run a bunch of simulations, look at the number that passed 10 or 12 hours, and use that to compare against other crew options. In general you can't try to optimize for 2 things at once, but if you want to optimize for highest chance at X hours rather than best average length, that's relatively easy to change.