Tuning the game

I admit it; I am getting very frustrated with this game. How do you do the tuning for loot and resources? Must we wait for a patch to alter tuning each time?

Have you noticed that on youtube, there are almost no new extended games past about 5 days? What is the average life expectancy in this game?

I find it very frustrating to invest hours into a game and then gradually die a miserable death in the cold freezing storms. Maybe on Coastal there are more resources to extend the game play however new users must start on Mystery Lake. Finding the rifle there has become extremely difficult; I don't know if there is even a rifle spawned on this map anymore. It is extremely difficult to travel to all of the spawn points on the map due to the weather. Moving around is very slow at times and you have to stop to make fire. The wind direction changes so rapidly that fires seldom burn more than an hour outdoors even in a sheltered spot.

I've noticed recent changes to the wolf conflict interaction; it seems we can fight the wolf all the way to its death now and it won't necessarily flee. I think that makes the game more survivable however the need to get out regardless of weather due to starvation is a killer. I would say the problem is a shortage of fishing lines (or cured guts) The two means of gathering food, fishing and snaring, are not adequate to get us through the early stages of the game. The calorie consumption seems too high however a consumption of 3000 calories per day of hard work outdoors is not unreasonable. I think there should be more meat on carcasses and animals to be realistic if we are going to make the calorie budgets realistic too. If there were more other resources that we could gather, plants and edibles, that might help a little.

I think we need to make sure that getting past "the wall" is a matter of skill, not luck. I call it "the wall" because there is soon a point where food and clothing resources force us outdoors regardless of weather and we have to take huge risks just to have a chance to survive. That part is ok to an extent however when nobody is making it, we have to think there is a technical tuning issue here.

Are you gathering statistics on the games? You might consider this an enhancement (it certainly is) however I submit that this game has to have proper tuning to become more popular. In other words, I think it's a critical functional need. We need to see those extended videos past day 10 or so. Most newbies don't seem able to get past about an hour or so of game play on Youtube.

I love this game; I love the depth of the strategy; I hate getting snuffed for no good reason. :-(

Please leave this post in the technical issue forum. It will have more visibility here.

A lot is dependent on how you play the game... Even Stalker mode is relatively easy when you plan out travelling and have backup plans.

Some of the frequent TLD streams for Stalker mode you might want to check [they use fresh starts each major update, rather than using old saves which can be exploited when new settings come out]:

Kraelman - Plays a lot of Mega Hard Mode [self imposed limitation such as only playing outdoors, or even playing blindfolded]

LMG (LordMasterGaming) - one of the longest running streamers for The Long Dark, and does frequent 24-hour marathon streams for charity playing TLD. He usually dedicates every Tuesday for The Long Dark play.

PocketsLLP - Another frequent Mega Hard Mode player... one great challenge he and Kraelman sometimes play is pretending there is a child who requires daily supplies as well.

There are many others as well, but I recommend them as a great start... You'll see various east techniques used to survive and handle different elements, and they're extremely open to answering questions and showing players how to accomplish each task. And, as mentioned, they play fresh rounds with each major update, so you'll know the approaches are accurate for the most current version without using older version saves.

You can also check their previous broadcasts if they aren't streaming live.

Here is a link to a specific tuning proposal I made earlier. It didn't seem to garner much interest nor even a developer response.

Life Expectancy Feedback Heuristic

Automated heuristics might not be the way to go: a) they are complex b) complexity introduces opportunities for innocuous and insidious bugs

A manual heuristic is where humans have to be involved in the feedback process. You have to assess the technical skill sets of your developers; don't put in too much "magic". It can be tricky.

Kraelman - Plays a lot of Mega Hard Mode [self imposed limitation such as only playing outdoors, or even playing blindfolded]

LMG (LordMasterGaming) - one of the longest running streamers for The Long Dark, and does frequent 24-hour marathon streams for charity playing TLD. He usually dedicates every Tuesday for The Long Dark play.

PocketsLLP -

I found it difficult to get the videos to play. What I did see what a lot of looting on Coastal Highway. Again, this is not the same path as a new player would follow. They could perhaps unlock the other maps on Pilgrim mode and then switch to Coastal to loot and start on the latest updates. I am currently only starting in Mystery Lake since this is the path all players must follow.

I don't think the game became greatly harder after the last patch? Measuring player success by the number of youtube videos seems a bit odd.

I don't think the game became greatly harder after the last patch? Measuring player success by the number of youtube videos seems a bit odd.

Did I say the gross number of youtube videos? I did not. I was specific about the life expectancy aspects of the videos.

Have you noticed that on youtube, there are almost no new extended games past about 5 days? What is the average life expectancy in this game?

Just checked and there were quite a few...

Plus YouTube isn't the only source of videos - a lot of players on Twitch are well over your 5 day mark [check the past broadcasts and highlights]... in fact many are 50+ and 100+ days even in Stalker mode... Not everybody posts all their runs on YouTube, so that would be a rather limiting deciding factor if that's all you went by.

As for life expectancy, it depends on the individual player, their mode, and their play style... You can get players with runs lasting under 10 minutes, and you can get players with 100+ days survival.

Ok, measuring player success by the life expectancy in youtube videos seems odd. Maybe most runs end quickly, but I believe that has always been the case?

Bad decisions, rather than bad luck, have been the reason for my death in almost all of my runs, some (most) of which were certainly less than 5 days--and that was before the update.

Edit: let me try to be more constructive,,, Besides fishing and snaring, there is a lot of lootable food available in each of the maps, plus hunting and scavenging wolf-kills. Perhaps you could try focusing more on food collection methods that give a better rate of return on effort to start out with -- i.e. looting, and if no rifle, you can always try scaring a deer towards as wolf.. then scaring off the wolf.. (carefully!)

This isn't a bug. This is more feedback to how the game is tuned. I'm moving this to the Feedback Section of the forums and you guys can continue to discuss this there.

Jeremy

My question is "How do you tune the resources?"

The presence or absence of loot seems to be random or pseudo-random. The last run had a LOT more loot available from the Mystery Lake camp area and I finally discovered what seems to be a new spawn point for the rifle. Weather improved significantly also. This could be just random "luck" or it could be by design.

It does not matter how I gather data; I still want to know if we wait for patches to change the tuning or how it works. I don't need coding details. My data sets were based on careful observation. I wish you would not disparage my methodology. I don't care if it seems strange; I do know about statistics and the HUGE problem is that to make accurate assessment, I would require a huge amount of data. I am observing trends that may be random or may be as a result of the game changes. Whatever; my question remains.

Thanks for the assistance; really, only the devs can answer this for me.

Perhaps it would be better to be more specific if you want a specific answer.

Do you mean:

How do you (as the player) tune the resources? (Answer: you choose a difficulty level)

or

How do you (the developers) tune the resources?

Your question could be interpreted either way, though now I assume you mean the latter based on the additional info

Perhaps it would be better to be more specific if you want a specific answer.

Do you mean:

How do you (as the player) tune the resources? (Answer: you choose a difficulty level)

or

How do you (the developers) tune the resources?

Yes, the latter. I'd like to know how the developers tune the resources of the game because it seems so variable. Specifically, are resources controlled by the server and the developers can tune them between updates or the game itself tunes resources but the parameters of the pseudo-random allocation of resources is fixed for each update.

The reason that this is important is because as the game has evolved, especially with the new harvesting mechanisms, it affects availability of resources. I do not know what other factors about resource distribution probability have changed (if any) with each new release.

[glow=red]Warning: TECHNICAL[/glow]

Let me insert a reference here without getting too technical:

Probability Distribution Function

Relationships amoung probability distributions

Stochastic Process

Suppose F(x) is a uniform linear pseudo-random probabilty distribution (i.e C function rand()); there is a function G(y) which is used to map F to a new pseudo-random pdf such as a normal distribution (not suitable) or a Weibull or other pdf.

The Weibull distribution has three parameters

Nu scale parameter, or characteristic life

Beta shape parameter (or slope)

Chi location parameter (or failure free life)

In the case of TLD, we want a discrete probabilty distribution with a relatively narrow range. Too much range or width of the probability curve, means a lot of variability. The Chi value is like the mean or median of a normal distribution. This is the primary tuning of any given probability distribution which means that on average, the player should receive a sufficient amount of resources to achieve a certain "life expectancy". (confusing since same terms used in different places)

So if we look at the player's game life expectancy, this is the indirect result of all other parameters as well as player skill (which should increase over time with experience) I am suggesting that we look at the multi-variate function that determines player life expectancy from a statistical point of view. Then we can use each of the parameters to see which ones are significantly affecting life expectancy. This is known as a Monte Carlo statistical study.

The process of tuning all the independent parameters is a stochastic process. That is we have to apply feedback to achieve a desired result. A surplus of any given resource will not yield a greater player life expectancy whereas a shortage can show a significant effect. For example, if no axe, knife or other vital tool/weapon is available or is not relatively easily discovered (perhaps by learning the spawn points), the player can die of starvation or be forced out of doors in bad weather or become the victim of wolves while on a desperate quest for food. When you add new spawn points or change them, it gets harder to find the necessary resource (tool). At this point in the game, I have to guess that the critical resources are: 1) clothing 2) cloth 3) weapon 4) good weather (vs bad weather) 5) tools 6) found food. Of course weather is also a multi-variate function having parameters snow, fog, temperature, wind speed, duration and variability. If wind direction changes several times per hour, it becomes very difficult to locate a wind shielded fire spot. So there are literally hundreds of parameters which can affect the game. Tuning is thus not trivial. I think we might be following an ad-hoc process based solely upon forum feedback; this is problematic. That is why I strongly recommend gathering data on mortality of the players including as many of the parameter measures as possible (shortages specifically).

I would recommend not getting too hung up on theory but gather data and see if it can help the ad-hoc process.

Now I don't really need to know HOW Hinterland does the pseudo-random probability for resources; I do want to know if the developers are tuning the parameters and I would be happy to offer suggestions for how to run a better simulation that achieves our objectives which are: maximize player enjoyment and challenge.

I hope this makes sense. If not, please ask specific questions.

This is not intended as play feedback; it is purely a technical issue.

It is possible to build a completely self-tuning stochastic process for resources. I think we have to have a handle on the overall big picture rather than making it too complex. Ad-hoc is good but still needs good data.

So I don't get it, are you just curious, or are you offering a solution to a problem? And if the latter, can you state the problem in one sentence without making any assumptions?

So I don't get it, are you just curious, or are you offering a solution to a problem? And if the latter, can you state the problem in one sentence without making any assumptions?

I would say I am more than curious. :-) I have proposed two approaches:

1) gather stochastic information about player life expectancy and the parameters that affect it (player experience, games played, and more; resources, weather etc.

2) heuristics (ad-hoc or automated) which is the process for tuning resources and other parameters which affect game play

and I guess a third suggestion would be to give us a non-technical overview of the tuning process. Is it clear what I mean by tuning process and parameters?

To state the problem:

The game is a multi-variate simulation of hundreds of independent parameters. Tuning the parameters is non-trivial. The process of tuning is assumed to be ad-hoc guided solely by player feedback so that would be an assumption. I don't think there is any need to be defensive about using ad-hoc tuning however we can more rapidly understand the effects of parameter changes by using stochastics (statistical data gathering)

I'm aware of the various confounds of gathering data solely upon my own observations. I don't understand the process of how the game is tuned. My observations could be due to the variability of the resource probability functions or they could be significant trends (assumption). I feel strongly that gathering more data will help streamline and optimize tuning. Understanding the tuning process will help me to provide more useful feedback either technical or just observations useful as data.

Assumptions:

Assumptions are always there; identifying the assumptions during analysis is a good approach my S/W dev teams have used in the past. I guess it is the assumptions about the tuning process that are bedevilling me because I don't know how tuning works. Perhaps an overview of the process would help me and others?

I guess I was unable to state the problem in a single sentence. Oh well.

Have I answered what you want to know Octavian? I'm trying to restate succinctly. Good question.

You want to know how the game decides what stuff goes where and who or what alters that decision process and how and when.

You're suggesting that tracking what decision was made beside player success could yield data describing some of those decisions actually limit player success and then suggest elimination of those outcomes occurring in the probability.

I finally discovered what seems to be a new spawn point for the rifle.

Pray do tell, I'll have a look next time.

I finally discovered what seems to be a new spawn point for the rifle.

Pray do tell, I'll have a look next time.

[spoil]under the rail bridge near Carter Dam beside the burnt out fire[/spoil]

AFAIK this has not been mentioned in the various FAQ.

I finally discovered what seems to be a new spawn point for the rifle.

Pray do tell, I'll have a look next time.

[spoil]under the rail bridge near Carter Dam beside the burnt out fire[/spoil]

AFAIK this has not been mentioned in the various FAQ.

That rifle spawn location has been there since about November 2014

You want to know how the game decides what stuff goes where and who or what alters that decision process and how and when.

You're suggesting that tracking what decision was made beside player success could yield data describing some of those decisions actually limit player success and then suggest elimination of those outcomes occurring in the probability.

In a way. The developers mention that patience is necessary as the game evolves since they have to "tune" resource allocation algorithms. What I want to know about is how that process happens. Tuning. We know that the game uses pseudo-random algorithms to place items on the map for each play. So we have to assume that those items have some pdf functions. (probability distribution functions) Also there are algorithms for weather randomness and algorithms for wolf combat and algorithms for wolf pursuit/hearing and so forth. Pdf functions have parameters (constants) that determine how narrow or wide a spectrum as well as the median for that spectrum. This is technical. For example, the pdf might say there is a probability of an axe being placed on the map. There might be 0, 1, 2, 3, 4... axes. There could be a probability of 10^% for one axe 20% for two axes, 30% for three axes, 10% for 4 axes, 0% for 0 axes and 0% for greater than 5 axes. When you draw a histogram of probability (it's discrete) you might discover the PDF closely resembles a Weibull distribution probability curve. The Weibull distribution has three parameters that determine its shape. It is easy to map a continuous pdf to a discrete pdf by simply truncating the fractional part of the generated value.

There are a lot of things that can be tracked. I don't know the internals of the game but things like how often the player is starving or freezing is something that could be tracked too. We can't track decisions or thinking but we can observe if A leads to B which leads to death. I have to talk about this in the abstract.

So for any given death on the game, there are number of data points we could hypothetically collect to determine the influence of the various game parameters. My goal here is not to explain how to run Monte Carlo simulations or other more enhanced simulations (such as this one where user decisions also play a great influence) You can read about that here: Monte Carlo Simulation Methodology

The developers understand this far better than (most of) we players.

There is one other point I want to make (again extremely technical)

We need to study the boundary conditions which means the conditions where there is just enough of something to get by. If you are a player and you don't realize that you don't have a statistical hope of breaking through a "wall" that is you will inevitably run out of something at some point, it will be frustrating. The tendency is to self-blame. That might not be the case. Does this make sense?

So the first step is probably to determine the boundary conditions for any given item like an ax or rifle. No ax and things get grim quickly. But what are the significant parameters? My guess is that weather is the wild card, clothing wear out rate, availability of clothing and so forth. So data for those things is useful. But if you are a dev, you will already know the pdf function used to allocate something so you don't need to determine that items pdf by statistics. But cumulatively, things play together. More clothing and warmer clothing mitigates weather. I sure hope this is making sense to someone. :-)

No.

I can only assume from your posts combined what you think the problem is because you never state it directly.

What you're saying is that the game could be better balanced than it is now, so in other words, in your opinion, the game, now, is not as balanced as it could be, correct?