Abstract
Problem: Music streaming services on smart assistants have fundamental design flaws that frustrate power users — from broken shuffle algorithms to a complete disregard for the album as a unit of music.
Approach: Tim Cain, unable to turn off his designer brain, applies game design thinking to critique music streaming players (Google, Alexa, Spotify, Pandora) and drafts a specification for what a better one would look like.
Findings: Streaming services fail at true randomness, ignore album track ordering, lack context-aware playback, and offer almost no user preference controls for things like live vs. studio versions, censored vs. uncensored, or remastered vs. original recordings.
Key insight: Once you're a designer, you're always a designer — the analytical part of your brain never turns off, and that same skill set applies far beyond games to any product or system you encounter.
The Designer Brain Never Turns Off
Tim opens by reflecting on a familiar problem for designers: the inability to stop analyzing. When playing games, he can't help noticing bad UI, mismatched sound effects, thematically inconsistent art, or the Wilhelm scream popping up yet again. The same happens with TV shows and movies — questioning dialogue choices, character behavior, and special effects.
His conclusion: once you're a designer and you've done it long enough, there's a part of your brain you simply cannot turn off. This leads him into his real topic — he owns Google Home assistants throughout his house (office, bedroom, kitchen, living room, even the bathroom) and loves the convenience of voice-controlled music. But their music players drive him crazy.
True Randomness
Tim's first complaint is shuffle. When he asks the assistant to play music or shuffle a playlist, he gets the same songs in the same order repeatedly. The "random" function doesn't seem to understand what randomize means — connecting this to his broader video about randomness in games and how poorly people (and algorithms) understand it.
He describes a workaround: manually requesting a specific song or artist to "knock it out of the rut," after which generic play requests start varying again. His point — he shouldn't have to do that. The system also over-relies on thumbed-up songs, playing them disproportionately when he just asks for "some music."
The Album Problem
This is Tim's biggest frustration and what he calls perhaps "one of the most stupidly idiotic oversights I've ever seen in a piece of software in my entire career." Albums — the fundamental unit of music packaging for decades — are not treated as coherent entities by any streaming service he's tried.
What he wants
- Play albums in order. Many albums, from 70s stadium rock to classical symphonies, were designed to be heard sequentially. Some vinyl records didn't even have clear track delineations.
- Random album selection with sequential playback. He wants to say "play some Led Zeppelin" and have it pick a random Led Zeppelin album, then play that album's tracks in order.
- Album playlists. Instead of dragging individual songs into a playlist, drag entire albums. When shuffled, the album order randomizes but the song order within each album stays intact.
He suspects there's a monetary reason this hasn't been implemented, because otherwise it's inexplicable after over a decade of streaming services.
Context-Aware Playback
Tim wants the assistant to consider context when choosing music:
- Time of day — different music for morning vs. evening
- Day of the week — weekday vs. weekend vibes
- Location — the assistant knows if it's in the living room, kitchen, bathroom, or car (via phone). He doesn't want ambient music in the shower, and he doesn't want hard rock at his desk
These assistants already have all the location data. They just don't use it for music selection.
Playback Preferences and Qualifiers
Tim wants a rich set of options that streaming players currently lack:
- Temporal filters — "play new music" (definable as last 2 or 5 years), "play 80s music" (actually filter by release date, don't just find someone else's "80s playlist")
- Genre refinement per artist — "play Brian Eno ambient music" without getting his rock catalogue mixed in
- Skip live versions — option to prefer studio recordings over live/stadium versions
- Remastered preferences — choose whether to hear the most recent remaster or the original recording
- Censored/uncensored toggle — almost always prefer the uncensored version (some services have this, but it's not universal)
- Uploaded music as thumbed-up — Tim has 10,000+ songs ripped from hundreds of CDs. Some aren't available on any streaming service. He wants uploaded songs treated as automatically "liked" — reasoning that if he went through the effort of uploading a song, he clearly wants to hear it
Must-Haves
Tim ends with two hard requirements — dealbreakers without which he won't use a music player:
- Voice assistant integration — it must work on smart assistants (preferably Google Home, since he's invested in that ecosystem). Once you're used to yelling requests into the air and getting a response, you don't want to walk to a device to change settings.
- Music upload support — he must be able to upload his own library. His CDs are aging and rotting, boxed up in a closet, and many of those songs still aren't available on any streaming platform.
The Takeaway
Tim's parting message ties it all together: once you're a designer, you will always design things — and it won't necessarily be games. Sometimes it's just fun to think about things that aren't games, applying that same analytical framework to everyday frustrations.
References
- Tim Cain. YouTube video. https://www.youtube.com/watch?v=zX1Htqt9_iY