# Phase 7.1 Playtest Findings: Hi Fi Snake build026.zip / build027.zip ## Executive Summary The playtest for Hi Fi Snake (build026.zip and build027.zip) conducted on July 22nd-23rd demonstrated that while the developer's control responsiveness adjustments in build027 successfully reduced direct control frustration, they did not translate into improved player session times. The game continues to suffer from high early churn, with most sessions lasting under 1.5 minutes. The primary bottlenecks have shifted from basic input detection to severe performance issues (low FPS on common mobile GPUs due to high device pixel ratios), spatial navigation failures (players missing doors and hitting adjacent walls), and punishing early hazard scaling (lava in Room 2 killing players instantly). ## Current Game State Based on the current game-state GDD, the tested build features: * **Grid-Based Movement**: A runner/collector loop where players navigate a grid to collect pieces, grow a tail, and unlock doors. * **Active-Input Player Movement**: The player character only moves while input is actively held (joystick or keyboard) and stops immediately upon release. * **Player Speed Progression**: Speed starts at 4 in Room 1 and increases by 0.4 per room up to a cap of 12. * **Runtime Control Hint Selection**: The game dynamically switches UI hints based on the resolved device form factor (mobile vs. desktop). ## Developer Context According to the Developer Notes, two builds were deployed for this cycle: * **build0026**: Players had a hard time controlling the avatar due to an unresponsive virtual joystick. * **build0027**: Joystick responsiveness was optimized. Developer notes indicate this was successful in reducing control frustration, but did not improve session times. * **Performance Concerns**: The developer identified that at least 5 of the 21 sessions suffered from low FPS, likely caused by high `window.devicePixelRatio` on Mali-G52 and Mali-G51 Android devices. ## Positive Signals * **Reduced Control Frustration in Build 027** * *Confidence*: Medium * *Evidence*: Developer notes indicate that players in the latest build did not express direct frustration with joystick responsiveness, showing that the control optimization was technically successful. * **Successful Room 1 Completion** * *Confidence*: Medium * *Evidence*: Multiple players successfully collected pellets, opened the first door, and transitioned to Room 2 (e.g., MX session at 03:01, IN session at 01:09, US session at 00:54, US session at 00:48, IL session at 00:36). ## Negative Signals * **Severe Performance Degradation (Low FPS) on Mobile GPUs** * *Confidence*: High * *Evidence*: Multiple playtest sessions explicitly reported low FPS and lag, particularly on Mali-G52, Mali-G57, and PowerVR GPUs (e.g., MX Mali-G57 "FPS was too low so it became frustrating", BR Mali-G52 "FPS was too low", BR PowerVR "FPS was really low"). This aligns with the developer's note regarding high device pixel ratios. * **Spatial Navigation Failures (Missing the Door)** * *Confidence*: High * *Evidence*: Players frequently collected the required pellets but failed to align with or navigate through the door, instead colliding with the wall immediately adjacent to it (e.g., MX "didn't go through the gate but the wall", TH "hit the wall just one grid tile away from the actual door to the right", BR "hit the wall just by the side of the door"). * **Punishing Early Hazard Scaling (Room 2 Lava)** * *Confidence*: Medium * *Evidence*: Players who successfully reached Room 2 were almost immediately killed by the approaching lava hazard because they stood still or couldn't turn fast enough (e.g., IN "After opening the first door, the user just stood still and the lava came", BR "Lava is too fast for the players at the beginning. Death in the 2nd room by lava"). * **Turning Avoidance due to Perceived Effort** * *Confidence*: Medium * *Evidence*: Some players found turning the character highly tedious or physically demanding, leading them to avoid turning entirely (e.g., US Desktop "Feels like turning character is really hard and requires a lot of effort so players just try not to", BR "wasn't able to make a left turn"). ## Confusion And Friction * **Joystick Dragging Outside Active Bounds** * *Confidence*: Medium * *Evidence*: Mobile players frequently dragged their fingers outside the active joystick area, causing the character to stop moving or fail to turn (e.g., BR "Dragged the screen several times out of the joystick area and couldn't turn"). * **Tablet Joystick Acquisition Friction** * *Confidence*: Medium * *Evidence*: Tablet players struggled to acquire or move the virtual joystick initially (e.g., BR Tablet "Had a hard time moving the movement stick. Tried a few times tapping around it and then finally understood"). * **Store and Customization Flow Friction** * *Confidence*: Low * *Evidence*: A BR player spent significant time trying to buy skins in the store without having enough currency, indicating a lack of clear feedback or purchase flow understanding. * **Black Screen on Boot** * *Confidence*: Low * *Evidence*: A PE player experienced a 1-minute black screen before the Unity logo appeared. ## Retention And Replayability Signals * **High Early Churn and Short Session Lengths** * *Confidence*: High * *Evidence*: The vast majority of playtest sessions lasted less than 1.5 minutes, with many exiting under 1 minute (e.g., TH 00:48, BR 00:42, MX 00:27, MY 00:05, AE 00:20). Players frequently quit after a single run due to rapid deaths or performance frustration. ## Device / Browser / WebGL Concerns * **High Device Pixel Ratio Performance Impact**: Devices with Mali-G52, Mali-G51, Mali-G57, and PowerVR GE8320/GE8322 GPUs suffered from severe frame rate drops, rendering the game frustrating to play. * **Tablet Input Usability**: Android tablets (Mali-G57, Mali-G52 MC2) showed high friction with joystick acquisition and movement control. ## Current Design Vs Observed Response * **Design Intent**: Active-input movement is designed to provide precise, intentional player-driven movement. * **Observed Response**: Players find turning requires too much physical effort, leading to turning avoidance, or they freeze entirely when pressured by hazards (like the lava in Room 2). * **Design Intent**: Floor cues are designed to guide players to the door. * **Observed Response**: Players frequently ignore or fail to align with floor cues, missing the door by a single tile and dying to wall collisions. ## Missing Or Incomplete Systems Affecting Results * **Desktop Hint Asset**: The lack of a clear, persistent desktop control hint led to immediate exits on desktop (e.g., MY "player saw the game and the WASD hint then quit. Never moved"). * **Store Feedback**: The lack of clear error feedback when attempting to purchase locked skins without currency caused pre-game menu confusion. ## Confidence By Finding | Finding | Confidence | Why | | --- | --- | --- | | Low FPS / Performance issues on mobile GPUs | High | Reported across multiple sessions (MX, BR) and explicitly confirmed by Developer Notes. | | Spatial navigation failures (missing the door) | High | Observed in multiple sessions where players hit the wall right next to the door. | | High early churn and short session lengths | High | Quantitative data shows most sessions ending in under 90 seconds. | | Punishing Room 2 lava hazard | Medium | Multiple players died immediately upon entering Room 2 due to lava speed or freezing. | | Turning avoidance / high effort | Medium | Observed on both desktop and mobile where players struggled to execute turns. | | Joystick dragging outside active bounds | Medium | Explicitly noted in mobile playtest sessions where players lost control. | ## Data Quality Caveats This playtest batch consists of 20 recorded sessions. While it provides a good distribution of mobile and desktop devices across multiple regions (MX, BR, TH, IN, PE, VN, US, MY, IL, AE), the high rate of early churn in Room 1 and Room 2 means there is virtually no data to evaluate late-game speed scaling, procedural room layouts, or advanced upgrade choices.