Yeah, this was one of main issues - it basically made bruteforcing the default by accident, because just moving the item onto anything immediately started giving you “answers” as to whether it could combine with anything.
At least with generic responses, they all start off as unknowns where you have to take time to go through and check one by one, so it encourages you to try to think of the correct combinations before trying something like that. With RtMI, just picking an item and then moving it anywhere else in the inventory accidentally bruteforced a bunch of combinations.
No. Generic responses discourages (at least) me, because I had already several ideas and none of them worked.
I should have used another word instead of “encourages”… I mean it becomes more time efficient to try to work out the correct answer.
With generic responses, bruteforcing will potentially take a lot more time than working out the answer, so bruteforcing becomes a last resort.
In RtMI we had a situation where bruteforcing was so fast that if you ever thought “what item can I combine this item with?” you could figure it out in about 2 seconds by skimming it over every other item.
Though I’m not advocating for generic responses, I think any logical combination should be rewarded with a unique response, I just mean in terms of “generic response” vs. “automatic answer reveal”.
Perhaps the right question is: why was the interface designed this way? What problem was this idea supposed to solve? Given that it has flaws (it encourages brute-forcing and increases the likelihood of accidental puzzle solving), it’s only natural to ask what the advantages are. And right now I can’t think of any. We discussed it back then, but today I don’t remember.
If I’m on a diet, it’s not a good idea to keep a dessert right next to me on the table. For the same reason, if we want to discourage brute-forcing, it’s not a good idea to make it so readily available.
Who said the market is on a diet? I mean, we hard core adventure gamers are, indeed.
But the majority of the people goes for super-size junk food (casual flash mobile in-app-purchases games).
So i guess that UI is a modern attempt to tame old style adventure games to make them digestible for casual, modern stomachs.
And, I totally agree with you, but I’m playing devils advocate, here
You can offer thoughts. From the story point of view you can see what Guybrush is thinking right now. I can’t remember an adventure game where this is possible.
Plus: You can suggest more sophisticated actions. In Scumm you have only the six verbs, so you can only “pull a plant” and not “bend the plant to a funny banana”.
But yes, that isn’t working well in RtMI. It’s still a one-click-interface with two options. 
Well, yes and no. The “junk food games” are played in the train, at a bus stop or in the evening before sleeping. They have to be “casual”, short and “catchy” to work.
RtMI isn’t that in any way. I have to take some time to play it. That’s a complete different experience. If casual games are a short episode of a TV series, RtMI is a full-feature movie that I watch in a cinema. So …
… this won’t and doesn’t work. At least the story is too long for a “casual play” at the evening. So why should I try to “casualize” that game? I wouldn’t play it in the train because it’s too long. And I wouldn’t play it at the weekend with enough time because it’s to “casual” and easy. If you would like to address casual gamers, you have to go the Telltale way and offer small bits.
So playing the other devils advocate here, I don’t get it why Ron is/was trying to “casualize” the UI.
I totally agree about that part (mouseover offers thoughts), but (to be clear, sorry) I was referring to the other part (combining inventory items does not require clicking to reveal if it is handled).
That’s precisely my point. I didn’t actually agree with the myself advocate. I believe the devil is not guilty.
After all, we all know that @Someone who speaks German can’t be evil 
Let’s go textual. A full textual interface as an IF game: users write whatever they want and an LLM checks if it leads anywhere against a “game file” document that serves as the ground truth.
(I actually tried this)
So? How did it turn out? Don’t leave us hanging 
Besides, it’s something that should be tried again periodically as the models improve.
I somehow envision chat-bots for NPCs, so you can have a more natural conversation instead of just picking from a fixed set of options. Though might be more suited to non-humorous games, or AI needs to get better at being funny.
As for a text interface, even if it would actually be able to correctly interpret the majority of inputs, I dread the variations of “I can’t do this” I’d be getting, especially if it’s not clear what’s a usable object and what’s just background (or foreground) scenery.
> LOOK AT CASTLE
You see a majestic stone fortress with towering spires and a glowing magical portal. It looks like it holds the secrets to your entire quest!
> EXAMINE PORTAL
You can’t do that. The portal is just a flat cluster of purple pixels on the screen. It is part of the background image.
> EXAMINE CASTLE WALLS
Those aren’t walls. They are JPEG artifacts shaped like a castle.
> GO TO CASTLE
You walk forward. You bump your nose directly into the monitor. You are still standing in the empty field.
> LOOK AT FIELD
It is a beautiful green meadow. There is a small, dusty rock sitting right by your left boot.
> KICK ROCK
You swing your foot. Your leg passes right through the rock because the rock is drawn into the scenery. You lose your balance and fall over.
> LOOK AT SKY
The sky is bright blue with three fluffy white clouds. One cloud looks exactly like a key.
> GRAB CLOUD KEY
You reach up into the air. You grab nothing but empty oxygen. The cloud key is 2D and completely out of reach.
> SEARCH GRAB BAG FOR LADDER
You don’t have a ladder. You have a rubber chicken, a piece of old chewing gum, and a single copper coin.
> THROW COIN AT CLOUD KEY
You toss the coin into the sky. It flies straight up, blinks out of existence for a second, and lands back on your head. The cloud does not react.
> TALK TO CLOUD
“Oh great sky-key, reveal your secrets!” You yell at the weather. A passing bird poops on your shoulder. The cloud remains a stationary graphic.
> CRY
You sit in the un-interactive grass and weep. Your tears are real. The grass is not.
Jason Schreier spoke to Monkey Island creator Ron Gilbert about why he ditched publishers and took an unorthodox approach for his next game, Thimbleweed Park 2 — finding a rich guy
Some very interesting quotes in there…
there will be a few changes to make things more approachable, such as streamlining the verb box that the first game used for interacting with the world (but not streamlining it quite as much as Return to Monkey Island)
A-ha, our first real info on the new UI… good to hear we’re not going as streamlined as RtMI.
someone who wanted to fund the game. This wasn’t someone who worked in the video-game industry — it was a wealthy outsider who really liked adventure games and was a big fan of Thimbleweed Park… who, for the time being, has requested anonymity.
Alright, own up, which of us is it? It’s gotta be someone from this message board… I can tell you it’s definitely not me.
Gilbert put together a budget… It would cost more than the first game, which came in at just over $1 million
That’s cool, I thought maybe this sequel would be reliant on art and music from the first game to save money, but it the budget is bigger, that means he can craft plenty of fresh stuff.
This is brilliant. I don’t like the idea of a game screenplay written by AI, but… what if the author writes the story as a very detailed canovaccio (a plot outline), the thorough description of every character, the storyboard of the scenes and of the game rooms… and then gives AI precise boundaries and lets the AI manage the NPCs? This could entirely change the way we conceive the idea of “writing a story”. This could lead to single story line a la MI, or could be engineered as intertwined subplots with multiple endings. And no gameplay would be the same as another. This is very interesting, and I guess it could be even more applicable to large open world sandbox-like games. But maybe I’m late to the party, somebody is doing this and I’m just babbling about something’s been invented yet.
I developed a Claude “skill” some time ago; its goal was to create a “whodunit” mystery game. The player chooses the setting, the style, and what kind of character to play. The LLM creates the plot, wires everything up, writes a “game file” on disk, and then acts like a game master, conducting the game and reacting to the player’s instructions.
The results were good, and the games enjoyable, especially the feeling of slowly uncovering all the clues and reading unexpected funny lines that matched a character’s personality (snarky, sarcastic characters were particularly spot on).
The skill was developed a little poorly, though, and sometimes this led the LLM to confuse what the player knew with what they didn’t know yet. Today I would design the game in a way that would prevent this from happening.
One advantage of designing a game in this way is that you get real discussions with the NPCs and descriptions of everything. The LLM is simply instructed not to make up information that would conflict with the “game file”.
I’d include an example of game file here, but they are quite long and I’m terribly off topic, so if someone is curious, feel free to write to me a personal message.
Thanks for the link!
not a sequel and it’s not a prequel
I’ve thought more about this description and I think the story could be a reboot.
I just spoke with Ron! I am always reminded what a nice guy he is and fun to joke with. I tried to ask a bunch of questions but not make it too long either. Looking to hopefully post the video on Sunday in the afternoon UK time, might have it go live so may post the link earlier first.
I think my favourite part of the interaction was when he said the whole game wouldn’t be set in a mansion because that would be “weird” to which I said oh yeah - who’d do that 
Probably this famous (and wealthy) person who only heard he was mentioned in the credits, and assumed that was a positive thing as he didn’t listen to the rest of the story.
"It’s gonna be a beautiful adventure game, people are saying it’s going to be the best of all time… I’ve made a tremendous deal with Ron, one of the best deals he’s ever had. Publishers, they’re not involved this time, those were terrible deals, terrible deals Ron made in the past. Giving too much money, too much attention to the publishers, and they’re not doing anything for the game. Sad. They’re not making the puzzles, they’re not making jokes, and it’s a very unhelpful thing, they were doing a terrible job frankly.
And remember the UI, by the way, the classic nine verb UI, we love the nine verbs, remember when we used to have good UIs? And now we don’t even have verbs anymore, and it’s a shame, what an awful shame. The new game we’ll have verbs, I know a lot about verbs, we’ll have all the verbs, whatever it takes, we’re going to have all the greatest verbs…"