Maybe we could find a way to reduce extreme runs of luck

[peyre]

This came to mind again yesterday. My very first manned Vostok mission suffered a catastrophic failure that reduced it to 6%. At that point there seemed no point in bringing it back up to speed - it would've been too little, too late. So I put everything into Lapot since it couldn't benefit from tech transfer anyway, and on my second manned mission it too killed everyone aboard and dropped it all the way back to 6%. So frustrating! And I had done everything right, too, bringing the hardware up to Max R&D and even flying a couple dummy tests.

I realize this is what happens in a random distribution - sometimes you just get a run of very good or bad luck, and random rolls are an integral part of the game mechanics. Still, I'm reminded that most of the catastrophic failures in the Space Race occurred when the agency did something stupid --er, incautious: running a sim of the Apollo capsule with pure oxygen at one atmosphere, pushing ahead with the first Soyuz flight even when the engineers shouted it wasn't ready, etc.

The way the game works, it's nearly as likely to wipe you out whether you're taking all the precautions or cutting every corner. One of the early complaints about BARIS was that it gave you the feeling you were being punished for getting bad rolls and not because of anything you did. I just wonder if there's something we can change in the game mechanics to go easy on you if you've flown dummy tests or if you've suffered one catastrophic failure already (within, say, 8 turns or so). There could also be a flip side to it, where you're likely to suffer if you're flying people without even getting things up to Max R&D.

[smudog]

I'm not sure if its in the mechanics already - but what about getting a free failure pass for a crewed mission if you have a proceeding catastrophic failure on an uncrewed mission. That always bugged me that there seemed to be no reward for blowing up a test mission - that is typically the most educational part of an given program program! Of course, there would be the perverse incentive to fly uncrewed before you had done your R&D - but then again - I think that is exactly what is happing at Boca Chica right now....

[peyre]

That's an intriguing thought, giving an Avoid Fail card on a failed dummy test. We could get around your concern with it by giving the card only if the item is at or above Max R&D - that should prevent abuse of the system.

What I was thinking in the OP was finding some way of protecting a player who's done everything right from suffering two bad catastrophic failures back to back, especially when, not only have they done their due diligence in regard to safety, but they haven't derived any benefit from the vehicles that keep failing.

An Avoid Fail card might be helpful with manned flights as well - though oftentimes when I see this, it strikes my Mercury/Vostok, which I then abandon for a more advanced program. Also, perhaps it shouldn't get the AF card until it reaches / returns to Max R&D, again to prevent jumping ahead for a prestige boost using a new spacecraft that's not really ready yet.

[peyre]

Another possibility (which might be a little trickier to implement) would be that a component that suffered a catastrophic failure would receive an Avoid Fail card once it's at Max R&D and has flown one dummy test (Unmanned Suborbital or Orbital).

[peyre]

I just had another of these games where I just. could. not. get. ahead. I hadn't even flown a second unmanned docking mission or launched a lunar flyby, when the AI sailed over to the Moon in 1969. I think this idea of giving an avoid fail card (once the affected hardware has returned to Max R&D) is a really good one.

For an in-game justification, we could consider that after a disaster, your people will be extra careful and thorough in all their procedures - provided you're taking time to do things right (which is established/assumed by the fact that you've researched it as much as you can).

[peyre]

Here's an idea: when one side is having too much good fortune, maybe rather than suffering forced catastrophic failures they could receive negative newscasts. Conversely, a player who's doing poorly through no fault of their own could perhaps receive positive newscasts.

[rnyoakum]

Newscasts are assigned at the beginning of the game, so this would require some major work on the newscast system. It seems a little overkill.

when one side is having too much good fortune, maybe rather than suffering forced catastrophic failures they could receive negative newscasts

No. Just no. You're stuck trying to decide what is "good fortune" vs good strategy, and that's not something easily defined in an algorithm. It's also the bold move of deciding that someone is having too much fun now, so they should suffer a bit, when you've no idea what the Random Number Gods have planned.

[peyre]

Ah. Well it seemed like a good idea at the time.

[peyre]

I see that implementing this idea would entail setting the variable for avoiding newscasts, then inserting a check into news_suq.cpp for that variable and rerolling certain cases, then clearing or reducing the variable. Tricky. I think the avoid fail idea is probably a better one.

[hweimer]

Well, not all catastrophic failures were because of stupidity, just think of Soyuz 11 or STS-107. There were also a couple of close calls that were essentially random events (Gemini 8, for example). What one could do, would be to increase the chance of a catastrophic failure the more the die roll surpasses the safety factor of the step involved. But that would affect the balance of the game by quite a lot and make the endgame much less thrilling.

[peyre]

Ok, well as usual, I'm casting out ideas. I think of good ones and bad ones. I certainly don't want to give additional advantages to Gemini.

My thinking with the 2 points for those missions are that they're essentially dedicated dummy missions. The point that they could effectively halve the time a Gemini/Voskhod needs to get back up to speed hadn't occurred to me.

[sflynn850]

Do we have a spreadsheet or anything that illustrates cost+risk to reward ratios of various platforms and missions? That might help us decide how to balance some of these things.

[hweimer]

Leon, I totally appreciate your ideas. I think it is really hard to find solutions that are a) balanced, b) fun, and c) historically realistic, all at the same time. Nudging people towards dummy tests certainly helps with c), although we will always have less missions than there were in real life.

I once started such a spreadsheet, but then gave up at some point, because you would have to include large parts of the mission logic (e.g., how likely it is for a failed roll on a given step to be catastrophic or lead to the mission being scrubbed). I also made some quick calculations on how much money it costs to get the entire equipment stack to 90%. Perhaps not too surprising, Gemini came out on top, but this didn't include things like the additional complexity of doing a joint lunar landing mission. And there are some things that are extremely hard to evaluate in such a setting. For instance, EOR or LOR? Sure, EOR has less mission steps, so less chance for failures, but LOR actually has one manned mission step less than EOR (no earth orbital activities) and also doesn't require boosters on the manned launch (thanks to the lighter kicker), so less chance for a catastrophic failure.

[peyre]

Thanks. I'm pretty good at thinking of things that might help with b) fun or c) historically realistic, but most of them turn out to conflict with a) balanced, or even with each other (b & c).

For my money, the extra manned mission step on EOR is a lesser issue than the several extra unmanned steps with LOR. Sure there's no chance for catastrophic failure on those steps, but on the flip side there's a much greater chance of something going wrong on one of them, than on the single extra manned EOR step, and a failure on an unmanned step always fails the mission. I suppose for purposes of your calculations there you could consider EOR vs. LOR a wash.

[hweimer]

Okay, now I got really curious to look at the numbers. LOR has two more unmanned steps (TLI and lunar orbit ins burn), so that's close to 20% increase in failure probability at 90% safety. A failure will delay the lunar landing by one turn (assuming that you have scheduled a second attempt), i.e., a delay of 0.2 turns on average. EOR's additional earth orbital activities step has about 20% chance of being catastrophic, and 40% chance of scrubbing the mission. Again, at 90% safety, that's an average delay of 0.1 * 0.4 + 4 * 0.1 * 0.2 = 0.12 turns, if we assume that bringing Gemini back to 90% safety takes 4 turns for R&D and dummy tests. So, yeah, you are right that EOR is the better one.

[peyre]

A related idea might be for U. Suborbital and U. Orbital to always raise the capsule's Safety and Max R&D by a point even if it doesn't succeed on any steps of the mission. That would encourage flying dummy missions...though it might also invite abuse by someone flying it at, say, 23%.

The rationale would be that that mission was chosen specifically to iron out kinks in the hardware.

Edit: possibly, the Max R&D going up could apply only if the capsule is at Max R&D already, or perhaps at 75% (the top threshold for gaining technology transfer).

[peyre]

I just had another of these! After my first suborbital I just couldn't make any headway! Everything failed. Castastrophic failure followed by newscasts repeatedly slapping down the Safety of components I was already having trouble researching because my budget sucked. I couldn't even fly a Duration B until Spring 1969 and never got my Ranger up to speed. When the computer swept in with the Moon landing in Fall '69 it was kind of a relief to be done with that game.

[peyre]

Having yet another of these "piece of sh**" games where it just keeps dumping on me. The AI's at level 3, yet it can do no wrong and obstructions keep being thrown my way. The AI flew an orbital before I could even get a satellite up, it flew a manned lunar pass back in '65 or '66, and in Spring '67 it looks to be a step or two away from the landing. Meanwhile the newscast (seemingly "strategically", if I believed that sort of thing) damages my equipment and then when I'm back up to speed it points and laughs, "HA, ha, no rockets for a year! Sucks be to you!"

It really would be helpful if there were a way to have the game avoid negative newscasts when there's a huge discrepancy in game score or something.

Edit: sure enough, the AI cruised to an easy victory in Spring 1968, before I had even done a lunar flyby or Duration D.

[rnyoakum]

Newscasts are assigned at the beginning of the game, and while some event cards are restricted to certain years, they're otherwise random. Best I can tell, there's no attempt to balance things, other than to prevent repetition of events. That means some games are not going to be in your favor, which reflects the random nature of the board game. That's not restricted to the proportions of good/bad/neutral events; early events have a much greater impact than events in the late 70s. Also, event cards can't recur, for you OR your opponent, so if you're getting bad events your opponent may statistically benefit from your misfortune.

But Race Into Space can be anything it wants to be. Here's a hypothetical approach that would 'balance' things:

• News events are chosen era by era.
• In each era, all the available event cards (those not chosen) are collected together. 'Expired' events restricted to previous eras are removed from the pool, and the newly available events for that era are added to the pool.
• The card pool is sorted into 'good', 'bad', and 'neutral' piles.
• each player draws a certain number of cards from each pile, proportion determined by difficult level, to represent the events of that era. These cards are shuffled to create a random order of events for that section of the game.

EDIT: Integral to the event system I described is being balanced for both players, though potentially biased by difficulty level. In a game with two human players, that's important; it wouldn't be fair to perform a sleight-of-hand on the newscasts if one player's advantage is owed to superior strategy. For a game against an AI, blessed subterfuge is a legitimate option.

[peyre]

That sounds like a reasonable approach! Most of the time, the game is reasonably balanced, but there are times when one side constantly rolls snake eyes, and it would be good if we could moderate some of the excesses.

[peyre]

I had been thinking about some kind of system that would prevent (or make less likely) receiving 2+ negative newscasts in a row, but your system there would largely take care of that.

[peyre]

Another idea that's come to mind is, maybe the game should dock your budget less if the component that failed was at least Max R&D, as a sort of recognition that you at least weren't being reckless with it. But...that might sort of contradict the whole point of the prestige system...

[hweimer]

A quite different approach would be to have (un)lucky breaks not to have such a large impact on the game to begin with. Right now, the game rewards you if you are successful and penalizes you if you are not. I am not sure that this is always a realistic model. For example, following the Sputnik shock, the Western world has invested an incredible amount of resources into science and technology, i.e., quite the opposite of what happens in-game right now. Similarly, I recently had a game where I got a spring half-off sale with something like 170 MB cash on hand, which allowed me to land on the moon in the fall of 1966. I can't imagine a real-world congress saying "yeah, we said we would go to the moon in 1969, but we have no problem of giving you even more money to get there a few years earlier and make the president look like a moron that didn't know how long this would take".

Maybe one could extend the VP review to create a government inquiry after each catastrophic failure. If you didn't take any excessive risks (not flying at max R&D, no dummy tests, etc.), you are cleared of any wrongdoing and your budget gets restored.

[peyre]

That sounds like a very complicated solution to implement, but also very fair. It would certainly help in games where you're suffering from a string of setbacks! Those are extremely discouraging.

[peyre]

My son had a good idea this morning. If we were to find a way to smooth the "lumpiness" of runs of luck, maybe that smoothing could be lessened at difficulty level 2 and not done at all at level 3.

[hweimer]

Right now, I am playtesting PBEM (more on that later), so I play against myself using virtually identical strategies. What I've noticed, is that the course of the game strongly depends on who comes out first during the Mercury/Vostok age. Once the low-hanging fruits are taken, it is basically impossible for the player who has been on the short end to buy and research Titan/Proton for the lunar flyby or the two-seater because the budget is too low.

So, I'm inclined to think that it's not runs of luck that are the problem, but rather luck at a pivotal point in the game. In games against the AI, this is less noticeable because the AI can catch up by cheating, and the human player can catch up by playing better.

[peyre]

Ryan, here's another visual that might help show what you're talking about above. In this graph, the Soviets were first to achieve Satellite and Suborbital, but the Americans then achieved Orbital and Spacewalk.

[hweimer]

@peyre:

How do we enable PBEM, by the way?

Run cmake with "-DPBEM=ON"

[peyre]

Thanks!

Edit: Funny, I ran it with that option and the game compiled, but Play By Email didn't appear in the main menu and pressing P doesn't open a PBEM game. This is the command I ran to compile the game:

cd ~/ && sudo rm -r raceintospace-build && mkdir raceintospace-build && cd raceintospace-build && cmake -DBPEM=ON -DCMAKE_BUILD_TYPE=Debug ../raceintospace && make && sudo make install && cd src/game && ./raceintospace

(This, of course, was my own fault, a 10T error. It's not going to work if you enter -DBPEM instead of -DPBEM ! )

[hweimer]

Actually, the "pivotal moment" thing might be due to a bug. I just discovered that in most situations, prestige awarded for missions (both positive and negative) was counted twice, see #694. This was not very noticeable because the prestige summary in the admin building showed the correct score. But the part where it matters the most for the game, i.e., the budget, used the wrong values. So you could get a much larger boost from early successes than how it should be.

[peyre]

No kidding! So that might reduce the impact of this sort of thing. Must start testing it this weekend - maybe even tomorrow, since I have the day off.

[peyre]

Another thought is that it might be nice if the newscast made some effort to spread out hits to your R&D, so you won't have a series of setbacks just getting something up and running, like:

Especially when it's immediately followed by a rocket above Max R&D exploding on the pad:

And then my capsule (also above Max R&D) burning up on reentry. It's Fall 1963 and I can't even fly a manned mission without killing whoever's aboard.

So I take the hint that Vostok's just not going to work in this game, and build up my Soyuz, manage to do a suborbital, but get swatted for trying anything more ambitious. It's Spring 1966 and my people can't spend a few days in space without paying for it with their lives. When I did get the program up and running, I managed to fly a duration C just before the computer won the game with Apollo III.

[peyre]

On the flip side of course, it would be good to find some way of counterbalancing when someone's getting all good rolls, especially if his opponent is having a bad time. In this last game, I played the US and it seemed I could do no wrong, and the AI just couldn't get ahead.

[peyre]

Aaand, another one. EVERY SINGLE TIME I try to fly a manned mission, it fails - usually catastrophically. Mercury gets trashed so I abandon it in favor of Apollo. I watch the AI get two, three prestige firsts per turn while I'm struggling with research.

Finally in Fall 1965 I'm ready to fly my first g--d--ned Suborbital mission and my Max R&D capsule burns like a wad of tissue paper. Then our friendly newscaster tells me I haven't had enough misfortune yet, so I also get to pay extra for more Apollos. "That's a nice space program you have there. Be a shame if something happened to it. --Oh, it did!"

[peyre]

I just had a thought about something that might reduce this issue. What if we did something like (maybe after your first catastrophic failure), further cat fails--as long as the hardware was at or above Max R&D--cost you half as much prestige, to reflect that you're at least doing your due diligence as regards safety?

Or something more graduated, such as reducing the penalty by 40% if you're 1 or 2 points from Max R&D, 50% if you're at Max, or 60% if you're 2 or more points above Max?

[peyre]

Another possibility occurs. One of the most frustrating parts of this sort of thing is that multiple cat failures tank your budget, which makes it almost impossible to catch up to your opponent. How about if the usual hit to your budget occurs only if yours is currently higher than your opponent's?

Maybe your government doesn't vindictively slash your budget when it can see that it's already spending less on the project than the competition.