|This article is within the scope of the Portal Project, a collaborative effort to improve articles related to Portal and Portal 2. See the project page for more details about the article status.|
|This article is non-canon. Its subject matter does not take place in the "real" Half-Life universe.|
The Directed Design Experiments was a concept coined by Gabe Newell in November 2007. It consisted of shutting down Valve's production pipeline for a few months and turn the company into one big creative playground. Some of the experiments, described in detail in The Final Hours of Portal 2, led to what eventually became Portal 2.
In November 2007, Gabe Newell got the idea of shutting down Valve's production pipeline for a few months and turning the company into one big creative playground. He proposed the idea of "Directed Design Experiments" to his employees, hoping it would lead to a creative renaissance. The Valve employees started brainstorming in small teams for about 3 months, with no news from Valve to the public during that period.
At the end of the brainstorming in February 2008, the Valve employees gathered at a local movie theater in Bellevue, Washington for an internal "science fair" of their new design experiments, where was to be showcased "more innovation in 60 minutes than most game companies will see in a lifetime." Most of the experiments showed to be inspiring, some were funny, few did not work at all.
First experiment: Blobulator
The first team, "Team Shirley Temple", led by Ken Birdwell, focused on the idea of introducing liquid simulation into the Source engine, through the Blobulator demo. The demo consisted of blobs of a mercury-like substance that could procedurally attack the player, run over and drown enemies, and re-assemble like the T-1000 in the film Terminator 2: Judgment Day. The "npc_surface" featured in the Source Particle Benchmark stems from it; some its properties were eventually reused for the Mobility Gels appearing in Portal 2.
Second experiment: Modular A.I.
The second team, made of John Guthrie, Tom Leonard and Steve Bond, demonstrated their idea of modular artificial intelligence. Their experiment consisted of Overwatch Soldiers having different chips on their uniforms that would activate abilities such as flying and super-speed. A player would shoot an enemy, who would be blown to pieces, but with a twist: Another enemy would then run over, find stray ability chips other soldiers would have dropped, and would upgrade their abilities on the spot.
Third experiment: Two Bots, One Wrench
The third team, "The Elders of Zion", named as such because the group was largely comprised of Jewish employees, was led by Erik Wolpaw. They showcased the Two Bots, One Wrench experiment, the name being a play on the infamous 2007 scatological video 2 Girls 1 Cup. The experiment involved the player as a wrench-wielding hero, and two AI-driven robots, one brown with a deep, masculine voice, rough and gruff (voiced by Richard Lord, who also provided the placeholder voice for Wheatley), the other with a proclivity for wearing hats and a high-pitched, feminine voice. Their colors are seen to vary from one video to another. It was set in modified maps of Nova Prospekt, among others.
The primary goal of the experiment was to push in-game storytelling into a new direction: the two AI-driven characters would behave according to the player's action and events around them in real time, not as pre-determined scripts, and play with each other, resulting in sequences that proved to be hilarious and inspiring. This AI technology was later used to implement character dialog in the Left 4 Dead series.
Some visual elements reappeared in Portal 2: the two robots reused as ATLAS and P-body (the masculine one as ATLAS, the feminine one as P-body); the feminine one looking at a crate then looking at it again, reused for Wheatley looking down twice before meeting GLaDOS for the first time; and the different hats worn by the feminine one reused as part of the co-op backpack.
Fourth experiment: Time
The fourth team, lead by Kim Swift, was inspired to run a more practical experiment and see where the very successful portal technology might go in the future, and added the use of time to solve portal-based puzzles, the player recording themselves creating a portal, then reusing it afterward with a second portal to solve the puzzle. While Swift and her team were thinking this concept would potentially become Portal 2, Newell realized, after being initially optimistic, that it was not going to work, as it was not fun and too complicated. Swift, who left Valve in 2009, still thinks the concept could work.
She has taken this concept to Square Enix which developed it as Quantum Conundrum. A fan mod titled "Thinking with Time Machine" applies Swift's concept to a follow-up for Portal 2.
Fifth experiment: F-STOP
This experiment completely enraptured the audience. Upon seeing it, Newell remembers thinking, "F**k yeah!". This one experiment is said to have been worth stalling everything else at Valve for the three months of brainstorming. Named F-STOP, this never-before-disclosed project was headed by a team including producer Joshua Weier. Using the cartoon visual style of Team Fortress 2, Weier and his team mocked up a completely new, nonviolent, puzzle-based mechanic for a game: using a handheld camera to take photos of objects and through playing with perspective scale them up or down in some way to solve puzzles. The other employees considered it fun, memorable, and most important fresh and completely unexpected.
A few days after the science fair, Newell summoned Weier and the team to his office to ask them if they would be willing to look into making F-STOP a prequel to Portal, as he thought it was probably the big, unexpected idea that Valve needed for a sequel. Weier was a little shocked: Portal was such a sensation, such an outright phenomenon, no one wanted to be responsible for trying to one-up it.
But as the team worked on F-STOP, it did not become Portal 2, and it would take Valve nearly a year of intense development before they figured that out: feedback from playtesters invariably questioned why the sequel to Portal did not contain the portal gun or GLaDOS, what they considered central elements of the Portal experience.
The photo-taking mechanic for F-STOP was kept on the backburner, with the hope that it could be used in a future game. According to Lunchhouse Software's Tristan Halcomb, further tests were conducted in 2018 to see if F-STOP could possibly be revived as Valve's flagship VR title, though nothing came of this and ultimately the company chose to pursue a new Half-Life game instead. In 2019, Pillow Castle's similar perspective-puzzle game Superliminal was released: its similarities to F-STOP likely scuttled any further plans from Valve.
Lunchhouse Software, developers of the Source game Punt, were given access to Portal 2's source code by Valve and noticed most of the leftover code for F-STOP was present but unplayable due to missing elements. Upon receiving the Left 4 Dead 2 code, they were able to assemble a complete F-STOP demo. With Valve's permission, Lunchhouse began publishing a short documentary video series called Exposure which demonstrated the F-STOP mechanic and attempts to recreate what F-STOP would have been.
|Companies||Gearbox Software • Junction Point Studios • Valve Corporation|
|Production staff||Artists • Character models • Composers • Voice actors • Writers • Other|
|Software||Achievements • Compilations • Demos • Games (Half-Life • Multiplayer • Non-canon • Portal • Third-party) • Game Engines • HL HD Pack • Mods|
|Other Media||ApertureScience.com • Books (Lab Rat • Raising the Bar) • Fan content (Films • Games • Mods • Series) • Films • Merchandise • Preliminary Findings • Soundtracks (Songs • Composers) • The Final Hours (Half-Life • Half-Life 2 • Half-Life: Alyx • Portal 2)|
|Cut content||Prospero • Half-Life Alpha • Half-Life 2 Beta • Half-Life 2: Episode Three • Half-Life 2: Episode Four|
|Misc.||Canon • Development (Commentary • Half-Life 2 • Portal • Next Half-Life game) • Mr. Valve • Retcons|