Howard, Todd

game director and executive producer at Bethesda Game Studios; led development of Arena, Daggerfall, Morrowind, Oblivion, Skyrim, Fallout 3, Fallout 4, and Starfield

Todd Howard: Skyrim, Elder Scrolls 6, Fallout, and Starfield

Lex Fridman Podcast · Nov 29, 2022

• Todd Howard
• Bethesda Game Studios
• Skyrim
• Fallout
• Starfield
• Elder Scrolls
• open-world game design
• NPC AI and reactivity
• procedural generation
• video game modding culture

Core ideas

Howard describes Bethesda's open-world design philosophy as fundamentally about maximizing player agency — "saying yes as much as possible" to whatever a player tries, which inevitably produces both chaos and unintended exploits (players piling crates to scale walls, finding developer test cells never meant to be discovered) and requires sacrificing some graphical fidelity and frame rate in exchange for letting the simulation keep running. He explains that the games run the entire world simultaneously rather than pausing distant systems while the player does something specific, with NPCs further away updated at a much lower tick rate — a deliberate trade-off since much of that simulation is invisible to the player but produces emergent moments (an NPC noticing and picking up a better weapon dropped in combat) that make the world feel inhabited rather than staged. He distinguishes this from a pure "theme park" design (build only what the player will see, front-load the shops) versus building systems robust enough to produce believable behavior even when unobserved, and says future titles will push further toward simulating factions and world roles that exist independently of the player's quest involvement rather than existing solely as quest-dispensing storefronts.

On NPC writing and dialogue systems, he explains that the most beloved characters (citing Skyrim's guards and the companion Lydia) work not because of deep individual backstory but because they are reactive — small state machines that respond to specific player behavior (sneaking at night, unusual outfits) — and that voice actors, recorded continuously over years of a project's development rather than only at the end, substantially shape and refine characters through ad-libbing and iteration. Asked about AI/language models for NPC dialogue, he says the technology is clearly coming and has obvious application to Bethesda-style games, but flags open questions around real-time deployment, computational cost, and voice synthesis as current barriers rather than fundamental ones. He describes an intentional deepening of companion relationships in Starfield beyond simple affection meters — companions can be in a committed relationship, get genuinely angry at a specific player action, and return to affection afterward rather than the relationship status resetting to zero — because he considers these emotionally significant relationship and companionship moments (which he says he personally values deeply as a player) among the most impactful things a game can offer, alongside a broader defense of gaming as time well spent rather than wasted.

Howard traces his personal history from a self-taught childhood programmer on the Apple II (inspired by a TRS-80 Star Trek game shown in his fifth-grade class) making increasingly ambitious games as a hobbyist, through a business degree and a near-miss corporate finance job at Circuit City, to driving to Bethesda's offices unannounced on a college break and eventually joining the company. He recounts the commercial failure of Redguard (a game he loved, inspired by Prince of Persia, Raiders of the Lost Ark, and Ultima) as a formative professional crisis — Bethesda was in serious financial trouble at the time, and he felt personally responsible — and says the lesson he drew was to stop being conservative and "swing for the fences" on future projects, a philosophy that led directly to Morrowind's ambitious handcrafted open world after the company was restructured under ZeniMax. He also discusses Fallout 3 as one of his favorite projects, made under significant fan skepticism (from both existing Elder Scrolls fans and devotees of the original isometric Fallout games) but distinguished by intentionally different tone work per game — contrasting Fallout's post-apocalyptic satire of 1950s American optimism, laced with dark B-movie humor, against the more "blank slate" character approach of the Elder Scrolls series, where Fallout instead gives players a more defined pre-game backstory (born and raised in the Vault) to create emotional stakes from the opening moments.

On process, Howard walks through Bethesda's project lifecycle: early "lunchtime conversation" concepting of tone, setting, and one or two signature features (done well before the prior project wraps); a first-playable slice to prove out core feel; a larger "vertical slice" once more of the team rolls on; full production; and a final six months to a year of "glue" work reconciling disparate systems (skills, quests, economy, story) into a coherent "game flow," which he says doesn't meaningfully exist until that last stretch. He describes character-creation design as deliberately minimizing irreversible early choices so players don't feel forced to restart hours in after realizing a build mistake, and treasure/loot placement (including a dedicated "clutter" team) as a genuinely difficult, iteratively tuned art — too little feels empty, too much creates compulsive over-collection that paradoxically makes a space feel worse. On combat, he's candid that enemy AI is deliberately dumbed down from what's technically possible ("the illusion that the enemies are smart, but they're really there for you to kill them") because true intelligent resistance from a large group of enemies is not fun, and that games generally aim to keep players in the narrow band between boredom and frustration/confusion without tipping into either.

Discussing Starfield specifically, Howard describes the technical challenge of procedurally generating believable planetary landscapes (building large handcrafted tiles offline and wrapping/blending them around a sphere, rather than relying on fractal generation that looks unrealistic), a deliberate design choice to lean into the loneliness and desolation of largely empty planets (evoking Buzz Aldrin's "magnificent desolation") rather than treating that emptiness as a problem to eliminate, and dialing back originally more punishing systems (like running out of fuel and being stranded) once playtesting showed they were simply "fun killers" rather than compelling risk. He connects this to a broader reflection on science fiction's role in shaping real ambition — noting his own admiration for what he calls the "Avengers of NASA" at SpaceX — and expresses hope, without overclaiming direct causation, that games like Starfield might inspire some players toward actual careers in space exploration. He closes with reflections on the studio's now decades-long relationship with modding (deliberately designing games to be built with the same tools modders will use, and supporting modders professionally), the joint-venture nature of joining Xbox after Microsoft's acquisition, ongoing adaptations of Bethesda IP into other media (an Amazon Fallout TV series he was initially skeptical about until connecting with Jonathan Nolan, and an Indiana Jones game at Machine Games built around a known protagonist rather than an open world), and general life philosophy centered on curiosity, resilience through repeated setbacks (Redguard's failure, Fallout 76's rocky launch), and pursuing what he loves rather than what pays, as the throughline of his career.

Excerpts