ChronA

Depending what happens with the legal status of content produced by generative AI, some contributors might not only want to but be legally required to distribute their assets under a more permissive license. We have already seen AI generated textures in at least one FM. There is precedent from the US copyright office that all such works are automatically in the public domain and asserting any kind of copyright claim about them is fraud. That includes CC-BY-NC-SA, as all copyleft licenses still depend on asserting ownership over the work in order to set conditions on its use. Even CC0 is not actually the same thing as public domain.

Personally I can see some ancillary benefits from trying to move TDM from non-commercial-libre to true-libre, beyond getting off the FOSS community's naughty list. There are a fair few indie devs who have tried to make modernized Thief-likes. None of them have done half as good a job as this community. I think that is generally down to the engine. All the detection and movement systems take a lot of time to implement, which you guys have already paid down. Imagine if those indie devs had the option to use TDM as the base for their games. More of these games would be published, and more would be successful. And this in turn would grow the public knowledge base about working in Dark Radiant and TDM. Some of those devs might make their own public FMs. Some might contribute to the wiki and documentation. Some might contribute to project maintenance or even donate new features that they develop. Personally I would call this one of the bigger things that you could do to keep the project alive. It would definitely be a big project to bring the project assets into compliance or to fork off a compliant TDM-lite. A year ago I would have said it is impossible, but AI is changing things. It can make art and it can write code, and especially when it has a working example of the thing it is recreating to learn from. It still would not be easy, but at least possible. Let this be a lesson for creators to select your licensing carefully. It is not always easy to change after the fact.

This is the same scheme the most radical voices in this debate have been asking for from the start. (Myself among them.) I thought it was settled that having extinguish on short click created a back-compatibility risk for a small subset of old FMs. I think the concern was putting out candles that are needed as a light source to progress, and then not having flint to relight them. That's why the current (mechanically and cognitively sub-optimal) compromise was selected. Am I remembering right? If so, let's just all reread the thread history rather than rehash this argument over again.

My friend and I came up with a pitch for you. I think we would also be interested in helping you with dialogue and read-ables writing regardless of the story you chose.

With that many maps what a lot of people are going to want is curation and quality control. If you want to know why Thief FMs are still pulling more eyes, that would be my guess. It's not even that the quality of the best Thief FMs are better than ours (IMO they are not), or that they have better tools for recommending them. The big advantage is that Thief has an popular and established lexicon for having those discussions. People know their expectations and can communicate them; and when they communicate their preferences they can reasonably expect to have them understood and satisfied. An excessive example (not biographical but a plausible hypothetical): The relevant point is that playing TDM is a major investment to the biggest demographic of people that might be interested in branching out. It seems like some people here don't recognize that dynamic. Some of us think that there are a bunch of quirky or vestigial irregularities in TDM that don't add much value to the established community and could be homogenized to to lower that barrier to Thief transplants. We are confused why you guys see it as so threatening. Personally I still feel like I haven't got a single straight answer. All that said on this specific mantling issue, I am actually sympathetic to the reactionaries. I love that the new speed was calibrated off of parkour videos. That was super clever. However the opposition has a point that our play characters are usually hauling like 25 kg of loot, gear, and body armor during a typical mission, judging by the in game models. And they are generally not in any big hurry. It it were me I would split the difference between the old and new speeds, or maybe go 70-30 in favor of parkour.

The core TDM community is in a bit of a testy mood at the moment. A few weeks ago some vocal commentators felt that a major proposal went much too far in catering to the naive expectations of new players about how a certain feature should work, instead of respecting many long time players' desire for continuity of function. They don't have much patience for the argument you are putting forward right now. They think you should just learn to use the game system the way it was intended. In this case I agree with them. What you want is to be able to load your last save with a single key press, and the system in place now allows you to do exactly that. You just need to get used to the idea that this requires your last save to always be a quicksave. That's because you aren't supposed to use named saves to save or reload progress in TDM; that's what quicksaves are for. Named saves are just for bookmarking major milestones.

This is a serious question and not me trying to be snarky: if knocking everyone out and leaving them roughly where they fall is you preferred play style, then why is being able to manipulate limbs so important to you? It sounds like you are not too worried about the corpses you leave being found, and by your own testimony shouldering is more useful than fine manipulation for getting at loot. So why are you apparently routinely moving bodies short distances using limb frobbing? This does not compute for me.

Thanks for sharing. As a Krita user and a generative AI enthusiast this will be a fantastic resource for me. I'm excited to try it out.

I have certainly done my share of awkwardly shoving bodies into dark corners and behind furniture in TDM. I don't think I've ever successfully put them inside or under furniture. TDM bodies are (rightly) pretty rigid in my experience. However, @AluminumHaste, you have repeatedly shown yourself to be the top authority and wizard on what is or isn't possible in TDM. If you want to put up one of your meme-worthy videos I will gladly eat my words on this point. I am ready to be dazzled by the dozens of ways TDM lets you hide bodies. Otherwise slow experimentation and experience must convince me. I think the reason you need the current system is that TDM's un-shouldering is deliberately way more restrictive than Thief's, and the reason for that (circa 2010) was to force you to use the fancy new ragdoll dragging gimmick. You have Stockholm syndrome. That is why I think this matter must be debated. The nays have successfully argued that we can't have easy and maximally-discoverable candle pinching because there are two or three missions where you need to move a lit candle, among dozens where you need to pinch them, and you-all will continue to argue that easy and discoverable body shouldering is a mistake because it makes life marginally harder in the occasional situations where manual body dragging is needed. But what we should be debating is a patch to make unshouldering less restrictive. Or maybe I'm off my rocker. Unfortunately I can't prove a negative, so it is up to you to provide counter examples where these systems create fun gameplay challenges. Show me the receipts and you win.

Show me some receipts then. Cool features are cool because they create cool moments and stories. For example, mantling is super cool. If a guard is chasing you and there is a low roof nearby you can mantle up to make a clutch escape. And maybe then you find a hidden vent or something and your entire approach to the level changes. That's a cool moment and a cool story. Similarly, pinching out candles is super cool because it can create moments where the player needs to rush across a room and pinch out a candle before an enemy rounds spots them, changing how the player will approach the room. And I will freely admit, stacking boxes and planks with TDM's physics objects system is super cool. It opens up a lot of platforming possibilities. But where is the equivalent for the ragdoll bodies or being able to pick up and turn over every apparently-not-silver teapot? Ragdoll bodies could be cool if, for example, you could stuff bodies into chests or under beds to hide them. However I don't think anyone does that. Usually it won't work because the body won't fit and you will make a huge racket in the process of trying. In fact it is super-lame because it seems like TDM should allow that kind of gameplay with the ragdolls feature, but it doesn't. Same with object manipulation. What does picking up and manipulate every (non-valuable) plate, hat, cup, and apple actually let you do? You can throw them to distract guards, but that would also be true if they just went into your inventory, like bottles in other stealth games. There could also be loot or important items hidden under them, except in practice it can't be anything important because TDM's object selection is imprecise and its collisions tend to send objects flying or clipping into the world. So what we actually gets is coins and purses hidden under hats... gripping. Edit: Note, I'm not saying these gimmicky features should be removed though. That would break back compatibility with older missions, which no one wants. But we should stop giving object manipulation place-of-pride over more impactful mechanics like shouldering bodies in our tutorials, FMs, and default control bindings.

This was a bit of a joke from me by the way. (Maybe in bad taste). My purpose was to point out incrementalism has value in itself. No one is completely happy with the compromises in the pseudo final version we have arrived at, but it seems to be good enough. So let's all stop whining about the parts that didn't go our way before Daft Mugi keels over. I think an argument could be made that Skyrim's controls might be theoretically more optimal than TDM's adapted Thief scheme. Skyrim does have potions, readables, consumable items, and real time inventory management; if not in the official game then in some of its million mods. It seems to work well and about 10-20 million gamers are deeply familiar with it. But it would be a gargantuan task to completely rework TDM to fit that template, and it would alienate a lot of us who are used to the traditional ways. Better to take a few pointed lessons where they are most applicable and work slowly to optimize what we have inherited. But neither should we rest on our laurels. As I alluded to before, "gimmicks" like TDM's ragdoll bodies and physics object manipulation were still kind of cool in the early 2010s, but now day-they feel positively retro--and not always in a good way. It wouldn't hurt to have a hard think about what features actually contribute to the enduring appeal of TDM, versus which ones might be holding it back from charming an even wider audience.

Hold your horses everyone! We are going about this entirely wrong! Consider: What is the most popular and genre defining immersive sim of all time? Obviously the answer is The Elder Scrolls V: Skyrim! So how does TESV handle the functions we collectively call Frob, Interact, and Manipulate? Skyrim uses a single button called Activate/Use. A short click of the button always activates the target object's primary function if one exists (i.e. shoulder body, put out candle, put item into inventory, etc.), and holding down Activate/Use for a long click always makes the PC manipulate the object (if it's not too heavy). This is our solution! We just need to get rid of the manipulate button and this archaic "Frob" terminology. Everything can be handled consistently through just a combination of short and long clicks. Short click to Activate/Use objects and long click to manipulate them. That is the control scheme that will be the most accessible to the greatest number of potential players. It will make Wellingtoncrab and me happy by placing the primary interaction mode for every object on the easiest and most discoverable control input. And wesp5 and friends can be happy because the control scheme will go back to being 100% semantically consistent according to their interpretation of that term. I bet Daft Mugi can knock this out in a jiffy. Then we can start working on bring the rest of TDM's controls into line with the objectively correct Skyrim standard! ...Unless maybe some of us are not being honest about what we really want in a control scheme...

Okay, now this is a super cool idea! Social roleplay always felt a bit underdeveloped in immersive sims of Looking Glass descent, but that's a thoroughly understandable compromise. Social roleplay potentially makes the NPC decision tree logic and sound/animation requirements explode in complexity, such that not even AAA game makers generally want to touch it. But there is new technology around now that I think can overcome a lot of those difficulties quite economically. This is something I really hope will be picked up and start rapidly evolving for TDM in the near future. But for now some initial reactions: It's quite sensible to just use headgear to simplify matters with the player model. Maybe as this evolves more the guard disguise could be a helmet plus sur-coat/tabard. That would be plausible for the PC to quickly slip on and would not change his/her arms, but it would make a more believable disguise. And it could be even more plausible by introducing light-armored variant guards wearing sleeves and gloves like the PC! A similar civilian set might be a wide-brim hat and an over-coat or great-cloak. I unfortunately don't have time (or skills) to dig into this now, but I'd love to know more about the implementation: Does it it make you more visible to NPCs that are against your purloined team? That seems like a reasonable balance concession. Does it take time to put a disguise on or off, and do you need to be concealed for it to work? Does doing thief-y things like stealing (from nobles) or mantling around like a damn-acrobat make your disguise degrade faster? I wouldn't expect that level of complexity with only 3 days work, but it would be cool. Speaking of complexity, it seems like you could evolve this further to give each NPC an individual suspicion meter for each player disguise. Then you could do away with the artificial suspicion meter and have more sophisticated roleplay situations. I am excited to see this develop, and I really hope it can soon find a home in some future FMs!

Extensive customization options are appropriate during the pre-release testing or early roll out of a new feature. The people testing the feature can't know yet what configuration will work best for them, much less the people making the patch. My expectation is that by the time you have to make a final decision about including this patch in the next official update of TDM, there will be a rough consensus about the optimal parameters for the new mechanics, and you will only need one toggle for the new behavior. Or, preferably, there would be no need for an option at all if we can only get over our OCD about preserving arbitrary semantic grouping in our key-bind allocations; and embrace a superior control scheme that's not designed around highlighting an old gimmick feature whose trendiness expired with the Half-Life 2 era of FPS. But that's a pipe dream. Maybe instead we can just all agree that the general>gameplay options are getting a bit crowded. Perhaps we could split some of them off into an "appearance" or "accessibility" tag. As more people make their own mods and we modernize our accessibility options it is going to become a problem regardless. Best to get ahead of it. I agree, and that is one of TDM's strengths. It borrows and combines many of the best features from the three beloved Thief games, but it also gives FM creators tools to expand their creative vision in new directions. To some Thief purists that will never be an enticing proposition. But I don't see this feature as throwing a bone to those Thief players specifically, but just player in general who are bouncing off our game because it deviates from some genre conventions for no obvious reason, which makes them think this is an unpolished product. It is true that Thief was what set these conventions, and (shockingly/s) it is mostly Thief players who are interested enough in TDM to give it a try... but our responsibility to address these problems (if we can) is the same as if AMD or Linux players were having technical problems with TDM. They are people who we want to be part of our community because we never know if one of them might make the next FM like Iris, or do something crazy-innovative with TDM that we can't even imagine!

Seems like an inoffensive suggestion to me. It would not provide much of a challenge though since all one need do is let go of the mouse and voila: perfect stillness. For that reason, to work as advertised I feel like it would need to be tuned to be a non-factor 98% of the time, such that even veteran players would forget it exists until suddenly, surprise! here's an edge case where it matters. Like: only movements greater than 90 degrees per second trigger the effect. Basically it would just be an in-game jump scare mechanic. If something makes you as the player character involuntarily jump or flinch then that has a chance of being spotted. In the hands of a devious FM author that could make for some extremely memorable scripted moments. That's all just in principle though. In practice, this is an idea that would cost development energy to implement, and we want to make sure that investment is committed wisely. Remember how the community flipped the f*** out when Hazard Pay had the audacity to restrict saving, going so far as to make a mod specifically to negate the author's intent, and basically ensuring that no FM that's not a deliberate troll will ever use that feature again? Great use of resources that was. /s

On the off chance this is a case of ignorance and not stubbornness... Imagine if every object in the game only had one interaction mode. What would you want that interaction to be? For candles and lanterns would you rather be able to put them out or pick them up? For bodies would you rather be able to shoulder them or drag them around by their limbs? Now imagine for whatever you decided, the game only let you do the other thing. For new players who don't realize TDM supports multiple interaction modes, that is how the game feels right now. The essence of the proposal is that whatever interaction is the most indispensable--the "primary action"--that should be mapped to the default short frob that everyone is familiar with. The less important "secondary" action gets mapped to the new long frob, which we hope will be more discoverable than the use-while-frobbing combo. We might not all agree yet which actions make the most sense as the primary vs the secondary, or what should happen in all of the obscure edge cases, but I hope we at least have a consensus that this idea in principle is a worth deploying as an accessibility concession.

I just want to say, kudos to @Wellingtoncrab for so steadfastly championing the pro-change side of this discussion! The people who will benefit most from this change are people whose voices we would normally never hear. It is courageous of you to lend them a voice. Also, at risk of beating an already dead horse, I'd like to raise an objection against the anti-change argument earlier that the current set of control complications are all explained in the training mission, and therefore anyone who doesn't find them has only themselves to blame. That requires the training tutorial be fit for purpose, and according to modern standards it is simply not. The TDM training mission is technically competent, but it fails at the critical requirement of giving new players motivation to slog through the learning curve in order to play the game proper. This is an area where Thief inarguably did it better even 20 years ago. TDM's tutorial is too long. It is bloated with too much non-essential information, and lacks any teasers of the world building or gameplay loops that would motivate a player to try the game proper. To anyone with modern gaming sensibilities, the training mission says this game does not respect your time. People today expect game mechanics to be discoverable in the course of gameplay, which largely means clinging to genre conventions as much as possible, and teaching the rest through minimal, immersive tutorialization within the main gameplay loop. A New Job is a big step in the right direction as a replacement, but unless I am misremembering, it is only a partial one. It wisely focuses on the essentials of moment-to moment play, leaving the advanced stuff for the player to discover on their own. Except the whole problem is these advanced features are not at all designed to be discoverable. If your argument is that something is fine because anyone who played the whole training mission would know how to deal with it, that in itself is a sign of a major problem.

This is a good point that I had not considered. Can anyone remember any specific missions that require this contrivance of having the player carry a candle as a light source to progress? I feel like maybe "In Remembrance of Him" used it, and it was one of many factors that made that FM almost unplayably frustrating, despite having very nice art. If there are others that used it better, I might change my position. The counter argument would be that needing to move an extinguishable candle for light was always a problematic mechanic. If the player needs to emit light, then level makers should give them a lantern. Plus for those rare cases where an extinguishable candle needs to be carried for light, you can always quick-load if you put it out by mistake. Quick-load excuses many sins.... I'm less persuaded by this argument. As AluminumHaste pointed out, the difference between 1 keypress vs 2 or even 3 (because you also need to put the candle down) is only a few milliseconds. If that routinely makes the difference of whether you are caught, and that thrill of walking on the razor's edge is what you live for, then bravo, you are an exceptional player! But I don't think most people are playing TDM that way, and if we give them this I'm confident you can find other ways to keep the game challenging.

It's not hard. Nor would it be hard if you had to simultaneously hold down 3 different key modifiers, or enter a sequence of 5 directional inputs timed with key presses like a fighting game combo, and yet I bet you would agree that would be ridiculous. For any software there is value in streamlining the most common interactions thathas nothing to do with accessibility. When most players click a body 95% of the time they don't want to nudge around the limbs, they want to pick it up and move it. When most players click a candle 99.8% of the time they don't want to move it, they want to put it out. It doesn't make sense for these most common interactions to require 2 actions when they could require just 1. That's not to say that this is a obligatory change that will massively improve the quality of the game. You are right that it is a very minor thing, but if the willingness exists to create a patch to change it then it seems like a no-brainer to make it the default. That's my interpretation of this thread.

On the balance this seems like a good quality of life proposal. No functionality would be lost. The only complication I can see would be needing to make some minor alterations to the tutorial mission.

I don't want to take away from the significance of projects like this. It is amazing and an indicator of the marvels to come. But I must stand by my previous assessment. That's based on the date on the video (which predates their first demo video with GPT-3 by a month), the fact that the sequence of responses shows no processing delays at all, and the fact the description frames it as a demo of new actuators, besides my experience with how GPT LMs think (which to my knowledge are the first and only AI so far that show anything approach that level of self awareness). If the reaction is genuinely spontaneous it is a far more significant breakthough than anything else this group has done... like why would they be wasting their time with robotics if they had those kinds of insights about machine learning? I mean the alternative hypothesis is that these guys have invented some sort of facial recognition and emotive response prediction algorithm that can: Recognize its own reflection as a real-time image of itself. Be surprised to see itself, implying an ability to form expectations and a sophisticated theory of world. Realize it can use its reflection to test its own actuators. Do all of this in real time. That is an insane list of achievements, any one of which on its own would be able to raise millions in venture capital in a heart beat. I would love to be proven wrong about this, really, but for extraordinary claims I need extraordinary evidence. In the meantime, for fun here is how GPT-4 reacts to a surprise mirror: https://chat.openai.com/share/ae22e99e-ef8d-4397-9cf6-55c8f8abc526 (It's a little slower to catch on.)

GPT-3. However I don't think that video was generated by autonomous reactions. Rather it looks to me like the developers having fun with a pre-scripted sequence of expressions. If you did this experiment for real I don't think it would play out that way. GPT-3/4 cannot react with genuine surprise in my experience. Surprise requires having an expectation and then finding it subverted, but these LLMs don't have the neurological hardware to form those sorts of impressions. They have no continuity of non-verbal memory and limited options for introspection. Plus I have a hard time believing image recognition would be able to recognize the robot's reflection as a robot, or convey that information to the language model such that it could figure out it is looking at its reflection.

It looked to me like you hit the beast's back or shoulders every time (even the one that worked). Is it possible that this could be a hardware based problem? Maybe your system is giving you particularly high latency, making it impossible to do fine, last millisecond adjustments to your aim. That would fit with the evidence in the video, since in several attempts the target started to move before you were in position but you still swung the blackjack. It would also explain why the blackjack indicator would be less helpful to you.

I know what you mean. The things the algorithm can do once it's warmed up are astounding, and the endless list of applications to try out is mesmeric. I overdid it early on and actually gave myself a bit of tendonitis from spending every spare waking moment experimenting with it. I'm trying to pace myself better now. But I'm right there with you as regards the philosophical implications. GPT-4 has some legit weaknesses as a logic engine, but its abilities of inference and deduction are no joke, even when you strip away its overwhelming advantage of knowing everything humanity has ever uploaded to the internet pre-September 2021. It can see conceptual connections that most people would not pick up on, and it can act on them. That sounds to me like general intelligence; and it's already near or exceeding typical human level! Without trying to sound alarmist, this is not something this type of model should be able to do based on the training data available to it. There are no examples for these sorts of highly specific original deductions for it to regurgitate. The general intelligence is some sort of new emergent phenomenon, and it's got quite a lot of people in the machine learning research community equal parts excited and spooked. I don't see any new comments on either the public link or my private copy of the conversation. Maybe continuing just makes a new instance for that user?

Ooh! We should compare notes in a few weeks. I've been trying for a while now to find tricks for re-establishing continuity between conversations. I've had some success, but nothing yet I would call satisfactory. For instance with the Adventures of Thrumm RP game, I had to start a new session because the ChatGPT client was taking on the order of 20s per token to generate its responses at the end and was crashing every 2-3 minutes. I felt like I successfully got it back into the character and in story for the new session, but it took something like 2 pages of text and over 40 minutes of work on my part. Judge for yourself how well I did: https://chat.openai.com/share/f14f77f7-2b49-497a-990a-b8ee6f405fb1 I'm envisioning an ultimate solution in the form of AI "personas" with associated memories and biographical information in a searchable database, which the chatbot can interact with through an API based on some minimal leading-prompts. Unfortunately that is still a bit beyond my depth as a engineer and AI whisperer... but I am making slow progress. Thanks! You are correct that these were each one continuous conversation (minus a few false-start branches where I submitted incomplete prompts by mistake or tried things that didn't work). I probably would not recommend going that long again. I really only did it in those examples as an experiment to see what would happen. I'd say beyond about 8 rounds of lengthy prompt-response the model's amnesia problem completely erases any benefit it gets from the extra context of the longer conversation. Plus in long conversations it sometimes develops pathologies like linguistic ticks or personality quirks. Starting new conversations periodically is a pain, but probably still best practice. It's a new feature! This is actually the first time I've used it so I'm not 100% clear how it works when you send it to someone with their own account. The controls are on the left next to the chat session title in the chat list: the icons from left to right are to change the conversation title, share the conversation, and delete the conversation. If you'd like to try adding to another person's thread, here's a false start of mine you could try it on. I'll tell you if it works. (Turns out ChatGPT is chronically bad at anagrams, so vandalize away.) https://chat.openai.com/share/8d7227ab-3905-4bf1-82a3-12be4899d48f