Cemetery


I have decided to document some of the projects that I had been working on to some extent but gave up on at some point.
I'll explain what I had in mind and why I stopped. I might still want to get some of them done one day, drop me a word if you're interested in them in any way.
Depending on the context some of them are probably going to be described either in english or in french. I plan to add more of them over time.

I suggest Idlesoft's Cybercemetary as a soundtrack to read this page :



Bad manuals


Around 2006 I started to write tiny books about methods to produce bad pieces of art (bad music, bad drawing, etc)
I think one of the actual goals was to propose counter-intuitive methods that could lead to interesting results, but it was difficult to phrase them in a way that wouldn't read as "this is everything you should not do".
Of course "badness" is a subjective measure and a lot of the "bad" things that we love weren't made that way on purpose.
I own one of the existing books about bad photography (the one by Thomas Lélu). I like the fact that it only features lost-and-found photographs (classified and annotated), some of them are very nice.

Applied math book


Most math textbooks for children offer exercises with a very simplistic context (e.g. "Tom has three apples, Susie steals two apples from Tom, how many apples does Tom have now?" (the answer is "one")).
I wanted to expand on that idea in a transversal way, as the textbook goes on and the problems get increasingly difficult, the characters also grow up.
Tom never forgot those few apples that Susie stole from him when he was young and that led to complex stories of jealousy and revenge.
I wrote a few math problems for this book but it turned out to be more difficult than I thought, I'm probably not skilled enough at math and/or storytelling.
Also another related project would be to write math songs (like those children songs used to remember times tables and stuff) but this time for badass math theorems.

Photo-novel with adjustable happiness


This project was about an interactive five-boxes photo-novel strip.
The reader could use an interface to control if the mood was going up or down at each box.



Each box would have two possible children boxes (positive and negative) and so on.
That means 2^4=16 possible states for the entire strip and a total of 16+8+4+2+1=31 different panels.
In 2006, I drew a branching graph on paper with a basic scenario.



Then this paper was lost, when I found it again years later I couldn't remember exactly what those sentences mean (most of them implied a kind of joke).
I remember having developed a small web interface for it but I never took the pictures.

Of mice and oysters


This was a prototype for a board game that we made with my friend Léo Mathieu.
This idea was to split the game into two phases, one of it was about tile-building (think of Carcassone) and the other was an asymmetrical two teams race in the previously built maze to gather resources.
It takes place in the sewers, each tile is a piece of pipe that has water and concrete. The rat team could only walk on concrete while the oyster team could only move in water.





In order to balance the tile distribution I had to build a software that would play many games in a raw.



I stopped working on this at some point because I needed people to play the game in order to fix and fine-tune it.

Bit percentage


This is one of my most simplistic yet actually tricky project.
I wanted to make a software that would simply say how much of a computer file is or isn't (think of 1 as "being", and 0 as "not being").
I got it somehow working but it was taking too much time for large files.

Social play


I made several attempts at writing plays using unedited comments from social networks.
I started this before social networks were even really a thing, there was only chat clients and I had been using them to copy/paste sentences from one user to another and kept their answers as-is as a list.
The result wasn't that interesting, most of the time my interlocutor was just wondering why I would say such an out-of-nowhere thing.
Then I tried to make something else a few years later by picking specific unrelated posts from friends on Facebook and write a small play with them (each character was the actual person who wrote a given comment)
I managed to put together a few slightly consistent sketches about weather, work, chilling out, or growing a beard... But nothing that convincing.



Finishing works


It is possible to think that a poorly finished work is preferable to a brilliant work that is never going to be completed and "published".
For that reason I once wanted to offer my services as a "work finisher" (ideally within a team of people with diverse talents).
No matter what the work is as long as I can get it done (audio, films, drawing, writing...) I would finish it as fast as I can.
I will first estimate how much time it would require in order to set a price, then make sure we agree on some details and get to work.
Of course I offer no garantee on the quality of the work.
I'd still like to do this so consider me open for comissions, contact me.

Recipe generator


I have made several protoypes for recipe generators.
Such as this one or this one. Overall I've never been convinced that a recipe is the best way to convey culinary knowledge.
I prefer to think about this knowledge more as many small building blocks of knowledge (such as "mix powders and liquids separately") that can be combined in many different ways.
Once I started a new "serious" prototype for a recipe generator that would start from "whatever you have on your shelf right now" and make a meal with it.
I know there are websites that include such a concept as a search engine but they usually lead to recipes with additional ingredients that you won't necessarily own, so that breaks it for me.
I also particularly like the idea of letting users vote after trying a meal and improve the results in a "genetic evolving" way.
I have been personally part of one of those "evolving to the perfect cookie" collective experiments.



Jean-Luc et la machine


This was an idea of scenario for an adventure game.
It is not that interesting in itself but I wrote one document for it that I think is worth being shared here.
It is a list of many systemic or "non-typical" puzzles I found (mostly) in classic point-and-click games.
- Having to transfer beer from one mug to another right in the inventory in Monkey Island.
- Travelling freely on a map and landing at random places while searching for coordinates in Indiana Jones 4.
- Solving procedurally generated puzzles in Yoda Stories.
- Meeting random people on the map in Monkey Island.
- Learning new insults to fight with in Monkey Island.
- Looking at the scarf pushed by the wind in the background to know the precise moment when to spit in Monkey island 2.
- Following a character across multiples screens and watch their actions in Monkey Island 2.
- Speaking to somebody to know the right choice to do between an number of choices too high to bruteforce : picking the right book out of the library in Space Quest VI.
- Doing the same unsucessful action many times until it finally works like knocking on some door in Space Quest VI.
- Using a special code/formula backwards to get the reverse effect possibly like in Loom.
- Losing on purpose like in the big fan puzzle of Machinarium (the fan gets angry and leaves if you give wrong answers).
- Using an unexped piece of interface (the keyboard instead of the mouse in Bathos).
- Going to some place at a specific time of the day (think Majora's Mask).
- Using basic knowledge to solve a puzzle (I think there was something using the periodic table in some Space Quest game, but was it a piracy protection ?)
- Being chased by somebody (the small scissors guy in Clock Tower or Edna in Maniac Mansion).
- Not taking an obvious item just lying there (probably many Sierra games like Lighthouse).
- Using the mouse without clicking (like in Up a lazy ocean). Relevant if the pointer temporarily becomes the items taken from the inventory (e.g. a blowing fan).
- Doing a given action with the mouse while the character executes another action (some puzzles in Botanicula).
- Choosing characters with different skills (Maniac Mansion).
- Having many possible endings (Maniac Mansion).
- Recruiting random charachers (point & click : Burntime, rpg :Suikoden, strategy : Heroes of might and magic).
- Using an item on a map rather than in usual scenes (like special items in Legend of Mana).
- Using strategy-oriented resources (like in the first Dune).
- Making choices that fit with a previously stated point of view (King of Dragon pass)
- Putting items in places that move (a bomb in an elevator like in Duke Nukem)
- Reusing the same insults in response to another set of challenges in Monkey Island
- Having some actions lead to both desirable and unwanted results (pushing several pistons at once in Botanicula)
- Clicking thinks blindly (it happened to me in Full Throttle but that was because of a defective screen)
- Putting items or characters somewhere and let them do their job while working on somthing else (Burntime, Suikoden, etc)
- A place where items disappear in realtime (nuclear zones in Burntime)
- Giving up items knowing that they can be found again elsewhere (character swallowing and vomiting random items in GW's Ultimate haunted house)
- Money to be used wisely (Zack Mc Cracken)
- Dying on purpore (Leisure suit Larry, Planescape torment)
- Stealing items where they should be sold (Zelda, Spelunky)
- Crafting (done in an awkward way in Hand of Fate)
- Not trusting what you see on the screen (a newer and better example would be mushrooms in 1 hour 1 like)
- Using the sound (e.g. listening to the railway to know if something comes closer)
- Driving vehicles that go to some places and some others, RPG-style
- Doing the same action in rooms that look similar but aren't
- Waiting (for somebody to show up, for a bus, etc).
- Something has changed in the scene since last time the character was there, you have to search for it
- Doing mouse moves and clicks very slowly (e.g. to not wake up somebody)
- Doing something a precise number of times, (i.e. clicking the right number of times minus one, the last click being the confirmation)
- Using the inventory as a combination, the player must have exactly the right objects in hand and nothing more for something to happen
- Using a recurrent piece of dialog as a trigger (newer example : "the next person you say 'yes' to dies" in Reigns)
- Having to understand a langage (Legend of mana, newer example : Sethian)
- Having to memorize a long sequence of actions

Films


There are several films I either made or helped make and some of them will never get edited and published.
Most of the time the reason for this is because someone holds the files on their hard drive and doesn't take the time to send those files for me to edit them.
Among those films are :
- A movie about how Kinder toys are made
- A movie where someone gets slapped in the face many times
- A remake of Pattern Island
- Me freeing a Vauban fence

Mood report


Like the usual weather report but for how people are feeling.
I wrote some texts for this but well...

Song lyrics


I wish I had a textual database of song lyrics either in english or french.
I'd especially like to use them with AI.
I spent a lot of time trying to rip lyric websites or use existing database in unconvenient formats.
I finally gave up but I was surprised by the difficulty of the task, I understand there are probably legal implications though.
Although I do have the sorted list of the 2000 most commons (roooted) words in english songs.
And based on that list, here is what Tropes thinks are the most comming topics in songs.



I'd be interested in anything about this I would not know about.

Reid


Once I became intrigued with this article to the point where I decided to make a mostly textual interrogation game out of it.
I wrote a short story about a ghost in a lonely house involving a few suspects and alternative scenarios but stopped there.

Youtube recommendations remover


I've always found it annoying that Youtube tries to recommend me personalized videos based on my previous seen videos even when turning the history off so I wrote a Firefox plugin to hide them.
There was a time where I could visit youtube as a generic viewer and the videos on the right bar would only depend on the current watched one but as it became increasingly difficult to maintain, I finally abdicated and as I'm writing this my right bar is now a mix of indistinguishable related and unrelated links.

Scenaristic structure generator


Around 2013, I began conducting numerous tests on generating scenarios, ultimately aiming to create a comic book generator.
My goal was to define a generable structure for a scenario, for instance, having a set of elements with identities and goals, and a general narrative arc composed of those elements.
The generator would then start with a general arc (e.g. "the dog wants to eat the cake") that could then be recursively split up into smaller elements and goals (e.g. "the cake is hidden in the drawer", "some human character wants the knife", "the knife is also in the drawer", "the dog is afraid of the knife", etc).

I dissected existing movies, tales, and games to understand how their narratives were structured, creating spreadsheets that referenced the reasons for including specific elements in a story, the causes and consequences of actions, and the crucial elements that needed to be highlighted for the audience to understand the story.
Most stories I analyzed this way turned out to have one or two general "goals" (such as "this character wants to obtain this item") that either conflicted or fortuitously combined at some point during a course of about five "cause-consequence" steps. Everything else was sub-stories or content that wouldn't fundamentally change the story if removed.
I also asked random people about what they thought could make a scenario good in general. This turned out to be a tricky question; the answers I received were mainly about specific details rather than the general structure.
This approach might now be rendered somewhat obsolete by modern AIs, but I still consider it an interesting challenge.

Most of the files I kept from this project don't make sense out of context, but this one could be somehow interesting.
That project was a good example of a recurring programming architecture problem that I often encounter. I wrote some kind of letter about it at some point. I still encounter that problem, but eventually learned to use callbacks and closures, which, I suppose, are an important part of the solution.

The comic-generator I was initially aiming for remained in a prototypical state, but I've used some of the work for textual generators, such as the proverb bot. It might look like it's just filling gaps in predefined sentences by picking from a fixed set of words, but it's actually searching an entire dictionary for words that meet certain criteria in the context (for instance, for a synonym, antonym, hyponym, hypernym, etc., of a given concept).
By the way, I don't like the common obsession with Propp's Morphology, which, in my opinion, might provide analysis tools for certain types of stories but is not that useful as a guide for generation, or even doesn't make much sense for analyzing modern scenarios.

Automated music mixing


I partially wrote an article in French around 2014 in which I shared some thoughts about music technology.
I don't want to finish and publish it, but I'll simply list some of the topics covered:
Specifics of human curation, audio analysis, and collaborative filtering. These are methods commonly used for music recommendation but could also be relevant in the process of mixing tunes together.
The fuzziness of authorship regarding covers, sampling, and generated music (the last point would be more relevant today regarding AIs).
The (underused, in my opinion) benefits of DJing on a computer instead of vinyl.
Ideas for dynamic video game music based on existing unrelated tracks.
A hypothetical prototype for automatic mashups (taking into account audio analysis and human inputs).