ResidualPlaybeta

Source of Madness

Carry Castle · 2022 · $19.99 · Action · deal grade B

  • Estimated net residual revenue: $1.4k to $2.1k per month
  • Opportunity: $2.9k per month at x1.65 dormancy
  • Estimated lifetime owners: 44.8k
  • 69%% positive across 1568 reviews · 14.2 reviews/month

Why it's flagged

  • Developer silent for 17 months
  • Studio quiet across its whole catalog
  • Last build shipped 27 months ago
  • Last discounted 1 months ago, 12 sales in the last 12 months
  • Proven demand elasticity: x1.6 review velocity during past deep sales

Analyst notes (AI-assisted)

A 2022 roguelike action game by Carry Castle that uses procedural and AI-assisted visuals to create intentionally unsettling eldritch horror environments.

Source of Madness has earned $223k lifetime on modest sales (44.8k units) and currently generates $1.8k/mo in residual revenue, making it a quiet earner for Thunderful Publishing. The core appeal, procedural nightmare aesthetics married to roguelike combat, is conceptually sound, but player feedback reveals a persistent disconnect between artistic intent (managed cognitive dissonance via generative visuals) and playability (unclear enemy shapes, sluggish controls, weak early balance). A revival hinges on whether the studio can rebalance combat and enemy readability without diluting the aesthetic signature; otherwise, the IP remains a niche archive piece with modest licensing upside. Most realistic play: watch.

  • Risk: The procedural horror aesthetic, once novel for early 2022, now competes against both established roguelikes and the saturation of generative-art-as-novelty titles; repeat players are thin.
  • Risk: Core complaint across reviews is visual/mechanical coherence: players struggle to distinguish enemy shapes and attack patterns, suggesting the art-first design may have compromised readability below a playability threshold.
  • Risk: Developer Carry Castle is a single-title studio with no public engagement in 17+ months; revival would require external production investment, not in-house iteration.

This case as Markdown · Estimates carry ±30-50%% error per title.