Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[pondicherry]

Your work and commintment are so so impressive. Have a great 2017!

[Ultima Ratio Regum]

Your work and commintment are so so impressive. Have a great 2017!

Thanks pondicherry - you too!

[Ultima Ratio Regum]

This last fortnight has been extremely productive for finishing off the development of URR 0.8. It has been really great to properly get back into the swing of things and see, and be able to play, major changes to the game at the end of each night. I’ve produced three substantially new things this week – the first initial steps towards NPC personalities, the set of answers to questions that have a range of distinct options rather than only one possible reply with words being switchable in and out, and lots of extra detail for insults and compliments. I’m also about half-way through the establishment of a comprehensive baseline for conversations, meaning that the player can now successfully have a Q&A-esque conversation with any NPC they encounter on any topic, which is the biggest and most important first stage towards the full conversation system that’s being developed here. That will be finished by next week, and will be focus of next week’s blog post! For now, however, read on…

(Warning: due to the nature of this update being entirely programming, the adding of new content, and developing elements that aren’t yet finished and ready to show off, I’m afraid there are no images. Hopefully some screenshots will be back next week!)

NPC Personalities

Firstly this week, in the process of filling out the “options” responses (see below), I needed to actually come in and add some of the personality modifiers for NPCs. In some cases these apply to all NPCs irrespective of whether they are important or not, and in some cases they apply only to the important NPCs that game tracks independently as the player moves around the world, and then in other cases they are relevant only to important NPCs of particular classes (for example, only a gladiator needs to have a fully-formed opinion about the crowds who watch gladiatorial combat). There is now a pretty large set of personality traits that NPCs possess, which affect their actions and their responses – and as with everything, should give the player hints about their origins, backgrounds, allegiances, and so forth. There are definitely too many to look at them all in depth here, but there’s a few particularly interesting ones which I’ll recount here. These include:

like_of_other_countries: This personality trait, somewhat obviously, determines what individuals think about other countries beyond their own. As with many of the traits here, this trait is modified by a range of factors. These include the NPC class and background of the individual person, and the wider ideologies of their nation, and their religion, alongside a small random component throw in to ensure that two NPCs with the same demographics will not always match up exactly, but will still generally be within a logical variation on either side of a set of beliefs. For example, an “explorer” from an “internationalist” nation is likely to think very positively about the rest of the world; a “jailer” or “officer” from an “imperialist” nation is likely to think very negatively about the rest of the world; and so on across all the NPC classes, and potential modifiers for national and religious preferences. Taken on a broad scale, you’ll be able to identify commonalities and overall feelings in a culture, but individuals will still vary significantly according to their individual life experiences.

like_of_art: This trait determines what kind of interest the person has in artistic outputs (paintings, sculpture, etc – there is an equivalent for “literature”, which will cover books, poetry, etc). This is affected once more by the kind of NPC you’re talking to, and to the ideologies of the nation in question. For example, a nation with a strong cultural interest in aesthetics will naturally produce those who like art a lot more; whereas a nation with a strong intellectual interest in mathematics or mechanical engineering will likely be less interested in works of art. This will affect how much people are willing to tell you about the artwork of their homeland, how much they know about it, and give you some hints about the place of artwork in that culture and therefore where (and what) artwork you might be able to find, which might yield clues in your central quest.

religious_zeal: This is a trait affecting quite a range of responses. This will affect how NPCs respond to you if it becomes apparent that you belong to a different religion, what NPCs think about heresy, how friendly and well-disposed they are towards inquisitions and other religious rules and strictures, how they act towards priests, what kinds of money or resources they give to their church, and so forth. Although most obviously living in a theocracy will boost the average religious zeal, this still varies a lot between individuals, in large part from their status in society, their contact with other nations and religions, and their personal history and relationship with the religion in question. There’s a wide set of speech replies that draw on this particular trait, and I’m very happy with how these have all turned out.

policy_acceptance and X_preference: There is a set of nine related traits: the first is policy_acceptance, and the others are X_preference, where X is foreign, military, leadership, and so forth, for each policy grouping in each culture/nation. The first of these refers to the overall contentment of the individual with the general policies of their homeland. Leaders and regents will, naturally, be extremely positive about the policies that they themselves have implemented and oversee; nobles and lords will generally be very positive, but may express small amounts of concern about particular elements of policy; and so on and so forth across the full set of NPC classes, with some classes having much higher chances to have serious issues with the policies, and some classes having particular issues with particular policies – a jailer will almost always think building prisons is a great idea, a prisoner will almost always disagree, and so forth. The second of these, the set of eight preferences, refers to what policy the NPC would like to see implemented instead of the current policy in each of the eight areas. The number of “other policies” an NPC likes is dependent on their overall policy acceptance, and then what alternatives they like vary according to their NPC class and a range of other elements. For now these just lead to a wide range of interesting conversation replies, but in the future I’m hoping to do much more with these personality traits and individual/personal preferences.

leadership_like: This trait refers to how much the NPC likes the leadership of their nation. This is not to say the leadership policy of their nation, as above – theocracy, monarchy, etc – but the individual personality/personalities of the person/people at the top. There are a lot of elements which go into this particular decision for each NPC, and as with the above set, I’m hoping to later tie this into the potential for social movements, conspiracies, and the like…

fellow_soldier_opinion: For those who are within the military, this determines what they think of their fellow soldiers. This varies by rank, by leadership, and by the individual histories of particular soldiers. I’m not quite sure what else this variable will affect yet – beyond a couple of possible conversation replies – but I think it could be a nice way to build up a sense of how different military forces function in the URR world.

There are many others beyond these, but these should give a good idea of the kinds of personality traits that URR NPCs have. As with much of the game, these numbers will not be explicitly visible to the player, but rather should become apparent by the behaviour of the NPC, which – hopefully – should be rich and detailed enough that one can actually draw these kinds of conclusions, and then use this kind of information to make informed strategic decisions about your relationship to that NPC. In turn, all the sentences that NPCs can say which draw upon these elements have been finished, and offer a massive variety of comments and observations that NPCs can make through drawing on their perspectives, understandings, and past experiences.

Insults and Compliments Revisited

Secondly this past fortnight, I took the feedback from several people on-board about the insults and compliments, and decided to revisit these. Although the greetings and farewells vary substantially in length and detail – and, of course, one will never see lots of these in quick succession as we do in these blog posts – the same wasn’t quite true for insults and compliments, so I’ve adjusted these. There are now a range of shorter and snappier insults and compliments, and these have been added appropriately to the game’s databases of possible statements.

I also this week took all the farewells, greetings, thanks, insults, compliments, and threats out of the demonstration file and implemented them into the main game. This took a while because these sentences are generated in a unique way to give a particularly high amount of variation compared to other sentences (because they are so common) and they need to vary both overall between cultures/religions, and in individual moments of speech, so that two people from the exact same background will themselves offer different farewells at different times. This seems to be all in place now, however, and NPCs can now give these statements at appropriate times!

Option Responses

The third major body of work completed this week was what I’ve taken to calling “option responses”. Some questions are easy to answer, since the answer will always take the same form with a word or two exchanged – these are “basic responses”. Other questions are harder to answer, which split into “option responses” (where responses are very different depending on the nature of the answer) and “list responses” (where a response will always take the form of a list). This week I’ve been working on the option responses. Some of the questions that have these kinds of responses include:

What do you think of the leadership?
What do you believe your foreign policy should be?
What do you think of your culture’s art?
What are the religious policies of your nation?
How widely spread is your religion?
What is your job?
What is the history of this monastery?
In all of these cases the game can’t just take a default sentence and then vary it, but it has to instead select a sentence from a wide set depending on the data available, and then create that sentence anew each time. There are substantially more possible “base” responses for option questions than there are for all the basic sentences combined, which gives some idea of the kind of variation that some of these need to have. With this fortnight finished, I’ve now finished these off, and I’m very happy with the kinds of sentences they create – they’re varied, detailed, and will take far longer than the basic sentences before the player will ever come around to seeing the “same” sentences again.

Conclusion

This fortnight has seen some major progress in sentence generation and the conversation system, and we’re almost at the point where the player should be able to have a full – if thus far a little basic – conversation with every NPC you encounter. Stay tuned!

[Retropunch]

Looks fantastic.
I especially love the way you're developing NPC Personalities - I honestly don't think it's been done before on an individual NPC level, and I can't wait to see it in action.

In terms of traits; will there be more 'individual' traits as well (like 'selfish', 'drunk', 'generous', 'talkative' etc.)? The ones listed seem mostly to do with their thoughts on certain issues rather than their personal quirks. I can imagine these may only properly effect gameplay later on (selfish vs generous shop keepers for instance) but it'd be interesting to see these more individual traits come in.

[Ultima Ratio Regum]

Looks fantastic.
I especially love the way you're developing NPC Personalities - I honestly don't think it's been done before on an individual NPC level, and I can't wait to see it in action.

In terms of traits; will there be more 'individual' traits as well (like 'selfish', 'drunk', 'generous', 'talkative' etc.)? The ones listed seem mostly to do with their thoughts on certain issues rather than their personal quirks. I can imagine these may only properly effect gameplay later on (selfish vs generous shop keepers for instance) but it'd be interesting to see these more individual traits come in.

Excellent! Yeah, I'm really happy with how these things are looking at the moment, and everything is coming along so nicely. It's so exciting to ask ANY question (well, almost any question, right now) and see a logical, sensible, and grammatically-correct (normally) reply come out. Those kinds of quirks: interesting, and I do like the idea, but definitely not in 0.8. Going for only the absolute essentials to get the blasted thing out asap!

----

Lots of major developments (and some minor ones) in URR this fortnight! Firstly, all the content I added last week meant that a whole new set of words and phrases had to be added to the lexicon; the best part of a thousand in total, if I’m counting correctly. I’ve now implemented all of these, and the game can correctly vary the words from all these new sentences by drawing on these new additions to the central lexicon. Secondly, I’ve now almost finished what I’m calling the “standard” or “basic” conversation system, which is ensuring NPCs can reply to every question they are asked; this should be done by this time next week. Thirdly, a wide range of further variables for individuals and the cultures they hail from have been implemented, and are now having an effect on what kinds of things NPCs will say. Another very, very text-heavy post this week – although next week’s will contain a lot of screenshots, this week continues to be lots of under-the-hood programming and content addition that cannot, yet, be reflected in screenshots, but is getting pretty damned close. Read on!

More Variations for New Words

Firstly, as above, there was a huge set of new words and phrases that needed to be varied for each culture, as with all the others. These entered the game because I was writing the new “option” answers (see last week’s entry), and needed to be written up. This was one of those tasks that isn’t especially intellectually challenging or needs a lot of programming experience, but mostly just involves adding a massive chunk of “content” which the game will then deploy in the appropriate situations. With this finished, I’m very happy with those look, and the sentences read really well and really nicely. Here’s a screenshot from the new set – I’m sure you can see how/where a lot of these fit in…



I also went through the existing words, and decided to statistically bias some of them back towards slightly shorter variations, and therefor slightly shorter sentences, as a response to the feedback I regularly get about some of the sentences being too wordy. You’ll see the same in the earlier example, where we have some sets using the same short word twice to boost the chance of that word being selected (this is of course not an especially elegant way to do it, but let’s be honest: my programming is not known for its elegance). This should ensure that sentences will tend to be just a little shorter and a little less wordy, and I’m going to continue this trend of chopping out irrelevant words whilst maintaining sentence variety – though this is a tricky balance to strike.

Nearing Completion on Standard Conversations

The development of the game towards what I’ve been calling the ability to have a “basic conversation” – i.e. the traditional question-and-answer session that one gets in most games, where the player asks something, the NPC responds, and this pattern continues until all conversation options have been exhausted – is now extremely close. NPCs respond correctly for all the “basic” questions (about 1/2 of all questions), and for the “options” questions (about 1/4 of questions), and are in the process of being programmed to respond appropriately to the “list” questions (also about 1/4 of all questions). This is the final step to them being able to ask any NPC any question, and get a response. For now these responses are all truthful, and they always answer, but this will still be a huge milestone once finished. As such, by this time next week, I’ll be able to show off the basic conversation system working universally, for all possible inputs and outputs, even if a few placeholders like “[holybook]” will still be in place, rather than the appropriate text itself. From my current trials with it, it looks amazing, and the experience of being able to select any question from this gigantic list and get an appropriate and sensible response is really something. It has been a long time coming, but I’m very confident that 0.8 will be worth the wait.

New Individual and National Variables

This week I also found myself needing to implement a large number of additional variables for both individuals, and in some cases for nations and religions and cultures, which NPCs would need to draw upon when they give their responses to particular questions and sets of questions.

Mercenaries

A range of new variables were implemented this week for mercenaries, along with quite a complex formula to decide on how much mercenaries cost to hire. There are four elements here – how much training a mercenary has had (counted in years, generally as a soldier, or in an arena, or they simply became a mercenary immediately and learned on the go), how much experience they’d had (how many years they’ve served as a mercenary, and also some specific stories/information about what they’ve done during their tenure), and then a list of benefits and conditions each mercenary brings. For each mercenary, the game develops a possible list of each according to their  background. For instance, a particularly zealous mercenary might refuse to fight their own religion (a condition) but be especially keen to fight other religions (benefit). This system is naturally comparable to a lot of modern roguelike games where you have PCG characters with selections of “traits” (or an equivalent term) that mix positives and negatives, and task the player with interesting strategic decisions. The game then figures out how much “true money” a mercenary costs to hire (a secret number converted into in-game currencies whenever it’ll appear on screen), which goes by a formula I’ve developed. Broadly speaking, better-trained and more-experienced mercenaries will naturally cost more, and the more benefits they bring the more they cost, but the more conditions they bring, the less they cost. There’s a bit more to it than that, but some initial testing has shown that this attends some very satisfying results.

National Voting Rights

Here’s an interesting one – the question of national voting rights. At the moment when each nation generates, a die is rolled to decide whether women, men, or both are allowed to hold the throne (or whatever the equivalent of the “throne” is). It then considers who is allowed to join the military – if only one sex is allowed to hold the throne, then that sex will always be able to serve in the military, and sometimes the other one will be too; if either is allowed to hold the throne, then both will be allowed to serve (generally). I’ve now extended this to voting right; if either can hold the throne, then both can vote, but if only one can hold the throne, then normally only that sex is allowed to vote, but sometimes the other sex is allowed to vote at a “reduced vote”, e.g. one quarter of the other. There is no real-world bias here so it’s totally randomised between F/M/either, and all variants are equally likely, although certain ideologies make “Only one not the other” decisions more likely than others (Imperialist nations are more likely to restrict, democratic nations are less likely to restrict, and so forth, although this is not absolute or guaranteed).

Nomadic and Tribal Relations

There are now variables to determine how much particular NPCs like tribal and nomadic nations. This draws upon the general feelings of individuals about other nations, which in turn of course draws on a range of ideological and historical factors, but also then modulates this further according to particular ideologies (if these are a close match with lots of tribal/nomadic states), and the individual classes of the NPCs. These will therefore be generally close to a more general feeling about foreign lands, but also quite different in particular contexts, and will affect what NPCs say about their nations, what they’re willing to tell you, what they themselves have “bothered” to find out (if they care), and so forth.

Delegates and Parties

This week I also implemented the first half of the system that will determine what kinds of political parties exist in democratic nations, which parties hold power, and how these parties will have shifted over time. To assess the political leanings of each nation, the game first goes through all of their ideological preferences and develops a set of overall political leanings for the nation, going by some axes that the majority of real-world political parties go by: are they primarily globalist or nationalist, are they liberal or conservative, are they religious or atheistic, are they collectivist or individualist, and so on and so forth. For instance, the “Isolationist” ideology will naturally add a lot to nationalist, and some to conservative, and a little bit to conservative; the “Aesthetics” ideology will promote individualism, but also nationalism, but also a little bit of globalism, and might also support religious iconography… and so on and so forth. Then, if there is a strong specific religious belief in that nation, it checks what political leanings that religion has (is it very peaceful towards others? Does it enjoy holy wars? Etc) and adds that into the mix where appropriate. The game then ranks these overall national political leanings, and then selects a number of parties, before going through each party and having the central tenet of that party be one of the leanings, starting with the biggest leaning and working down. So a nation that is first and foremost “liberal” will have its biggest party adopt a broadly liberal position, with smatterings of other political tendencies lower down the list for that nation. Once that is done, the game then looks at whereabouts delegates come from in that nation – do they come from districts of the capital, from towns, from monasteries, from farms – and distributes delegates appropriately to each party, depending on its political leanings (and generates a procedural name for the party, although this isn’t finished yet, but I’ll show some off soon). This party affiliation will allow delegates to comment on the current political situation according to their own allegiance – which was the only reason I implemented all of this now! But it’s nice to have it in place, and it makes the democratic nations just a little more fleshed-out, even if it’ll be the small 0.9 release where we really see this more visibly.

There are also about another thirty new variables added in this week alongside the ones listed here, but I decided to mention these four as they struck me as being some of the more intriguing examples from the selection. As noted last week, these variables are entirely hidden (or rather, the numbers in each variable are hidden) and are always going to be represented instead by who people are, how they act, how they dress, what they say, where you find them, who they worship, who they serve, and so on and so forth…

Next Week

The completion of the standard conversation system, all questions/answers for all possible questions, and a lot of screenshots to show it off! It has been a while since we’ve had any proper in-game screenshots, so it’s definitely time to actually give you all a look at how (incredibly neat) everything is looking now. See you then!

[Retropunch]

All extremely interesting, you may well achieve the first game ever to completely avoid repetition in ProcGen!!

National Voting Rights

Here’s an interesting one – the question of national voting rights. At the moment when each nation generates, a die is rolled to decide whether women, men, or both are allowed to hold the throne (or whatever the equivalent of the “throne” is). It then considers who is allowed to join the military – if only one sex is allowed to hold the throne, then that sex will always be able to serve in the military, and sometimes the other one will be too; if either is allowed to hold the throne, then both will be allowed to serve (generally).

I find this a bit of a weird one. Whilst I know you're not going for a real world analogue, the vast majority of civilisations that have had female rulers have still only allowed male soldiers. Take our own country for instance, we've had tons of Queens whilst not allowing women to serve in the military.
All previous civilisations (Egyptians and so forth) have had similar.

Again, I realise you're not going for a parallel to real life, but it seems strange to have the norm being that the throne holder's gender can always serve.

[zaimoni]

It's strange [allowing women in the military] because such civilizations are preparing to win the Darwin Award.  Keep in mind that before WWI almost all wars were limited wars, and it was nearly an annual thing before gunpowder (no extended peacetime).

That said, low fantasy is rife with dying or nearly-exterminated societies where this is appropriate.

[Retropunch]

Yeah, from a medieval/renaissance time period perspective, allowing women in the military isn't particularly wise - hence why it was nearly never done.

If your army has pretty much zero force multipliers, you just need large numbers of people to fight, hence you want as big of a population as necessary. If you have 1 in 4 children dying in childbirth (and a mother 1 in 100), you kinda need to make sure that you have women around to procreate and not fighting in wars and dying/being injured. Also, many women of young to middle age were pretty much CONSTANTLY pregnant during their life times, with a year or two apart at max making them not particularly effective soldiers.

I can imagine cultural/religious reasons allowing it/enforcing it, but for even basic time period accuracy it seems a bit odd for it to be a 50% chance. My main issue though is just that you'll be ending up with around 30% of civilisations having a woman leader with women fighting and men not. Again, whilst I realise it's not looking for historical accuracy, it's biologically not sensible.

[NJW2000]

Unless men can get pregnant in certain cultures.


This is getting pretty Ursula LeGuin...

[Neonivek]

Yeah, from a medieval/renaissance time period perspective, allowing women in the military isn't particularly wise - hence why it was nearly never done.

Sometimes they were trained for militia service incase they either need to defend themselves (Not intended to enlist) OR as an emergency (Emergency enlisting).

This wasn't common and was specific but it did happen.

Still remember that person who wrote about a pregnant woman basically trying to fight off enemy soldiers with a sword because the fight was so hopeless that they basically had to field entire villages.

Though in terms of a society not under immediate existential threat having a regular female army or at least regular female soldiers? Very rare... to the extent that any examples I might have could even be fictional or myth. Well maybe as an inner guard or a personal guard.

[Ninteen45]

Yeah, from a medieval/renaissance time period perspective, allowing women in the military isn't particularly wise - hence why it was nearly never done.

If your army has pretty much zero force multipliers, you just need large numbers of people to fight, hence you want as big of a population as necessary. If you have 1 in 4 children dying in childbirth (and a mother 1 in 100), you kinda need to make sure that you have women around to procreate and not fighting in wars and dying/being injured. Also, many women of young to middle age were pretty much CONSTANTLY pregnant during their life times, with a year or two apart at max making them not particularly effective soldiers.

I can imagine cultural/religious reasons allowing it/enforcing it, but for even basic time period accuracy it seems a bit odd for it to be a 50% chance. My main issue though is just that you'll be ending up with around 30% of civilisations having a woman leader with women fighting and men not. Again, whilst I realise it's not looking for historical accuracy, it's biologically not sensible.

Note that in URR's earliest posts there was stuff like Minotaur generals of armies and the acceptable number of squares for a dragon to take up being casually discussed. Even if it doesn't look that way, we are in the realm of high fantasy. I'd say it's acceptable to say childbirth isn't usually fatal in fantasyland.

[Retropunch]

Note that in URR's earliest posts there was stuff like Minotaur generals of armies and the acceptable number of squares for a dragon to take up being casually discussed. Even if it doesn't look that way, we are in the realm of high fantasy. I'd say it's acceptable to say childbirth isn't usually fatal in fantasyland.

From many posts since then Mark has strongly re-situated it in a much more realistic setting and said that it is definitely not high fantasy. He's said a number of time that there won't be any magic or similar, and has on this forum (politely) rebuffed many people asking if there will be magic. As a caveat, there are many religions, and some may believe in magical powers, but that won't be something in game. At the very most, the artefacts that are at the centre of the game's quest may have some sorts of powers, but only in a vague sense (not, fireballs or whatever).

It's not even about childbirth being fatal though, it's just about not wanting to expose the primary way your civilisation grows to danger. Even if Mark takes a very loose attitude to medical realities in URR, it just doesn't make any basic economic or societal sense to do so. Sure, there may be a few cultures that see women fighting as a cultural necessity (only tested warriors may have children etc.) or something, but they'd be few and far between.

[Dorsidwarf]

I'd definitely say that a culture restricting only women to the millitary is very, very strange.

[Ninteen45]

I'd definitely say that a culture restricting only women to the millitary is very, very strange.

Baby shields. You wouldn't swing a sword at a baby, would you?

[Ultima Ratio Regum]

I was going to quote all the previous messages on this topic, but now I think it's easier to just give my general rationale/thoughts.

Basically: I'm trying to generate a world with major differences from our real world, within the realms of physical possibility: why restrict ourselves to what the real world has had? Why limit what can come up that way? That's just going to massively reduce the amount of interesting things the game can generate, and I'm very much against that. The idea that just because female soldiers might not be "as effective" as male soldiers (whether true or not) means civilizations wouldn't field them, implies everything that every culture does is "rational". And, I think, the history of culture demonstrates that every culture on earth does all kinds of things that aren't "rational" or "optimal", and I see absolutely no inherent reason why this would never happen. We have or have had cultures that kill a non-trivial number of their own children, infect themselves with diseases, hurt themselves or mutilate themselves in various ways, and so forth; cultures don't just do things that would be "best". I just think the game is just vastly more interesting this way, by allowing cultural ideas and preferences that didn't arise in the real world, but where there's absolutely no reason for those cultural ideas to not arise - as real-world cultures do far, FAR more striking things than some of the stuff that's in the game so far.

From many posts since then Mark has strongly re-situated it in a much more realistic setting and said that it is definitely not high fantasy. He's said a number of time that there won't be any magic or similar, and has on this forum (politely) rebuffed many people asking if there will be magic. As a caveat, there are many religions, and some may believe in magical powers, but that won't be something in game. At the very most, the artefacts that are at the centre of the game's quest may have some sorts of powers, but only in a vague sense (not, fireballs or whatever).

Yep; that model only lasted a year or so until I figured out what the heck the game was actually going to be, and then all the fantasy elements vanished. If there is anything that isn't 100% physically realistic, it will be a tiny fraction of items that have truly minor "powers" (akin to some of the items in Borges short stories, for instance), and even then... they might not. We'll see. I'm just focused on getting 0.8 out now, and a very quick couple-of-months 0.9 to further develop the speech system, then taking stock.