Five Fatal Assumptions: Why T-Shirt Sizing Systematically Fails for AI Projects
arXiv:2602.17734v1 Announce Type: cross Abstract: Agile estimation techniques, particularly T-shirt sizing, are widely used in software development for their simplicity and utility in scoping work. However, when we apply these methods to artificial intelligence initiatives -- especially those involving large language models (LLMs) and multi-agent systems -- the results can be systematically misleading. This paper shares an evidence-backed analysis of five foundational assumptions we often make during T-shirt sizing. While these assumptions usually hold true for traditional software, they tend to fail in AI contexts: (1) linear effort scaling, (2) repeatability from prior experience, (3) effort-duration fungibility, (4) task decomposability, and (5) deterministic completion criteria. Drawing on recent research into multi-agent system failures, scaling principles, and the inherent unreliability of multi-turn conversations, we show how AI development breaks these rules. We see this throu
arXiv:2602.17734v1 Announce Type: cross Abstract: Agile estimation techniques, particularly T-shirt sizing, are widely used in software development for their simplicity and utility in scoping work. However, when we apply these methods to artificial intelligence initiatives -- especially those involving large language models (LLMs) and multi-agent systems -- the results can be systematically misleading. This paper shares an evidence-backed analysis of five foundational assumptions we often make during T-shirt sizing. While these assumptions usually hold true for traditional software, they tend to fail in AI contexts: (1) linear effort scaling, (2) repeatability from prior experience, (3) effort-duration fungibility, (4) task decomposability, and (5) deterministic completion criteria. Drawing on recent research into multi-agent system failures, scaling principles, and the inherent unreliability of multi-turn conversations, we show how AI development breaks these rules. We see this through non-linear performance jumps, complex interaction surfaces, and "tight coupling" where a small change in data cascades through the entire stack. To help teams navigate this, we propose Checkpoint Sizing: a more human-centric, iterative approach that uses explicit decision gates where scope and feasibility are reassessed based on what we learn during development, rather than what we assumed at the start. This paper is intended for engineering managers, technical leads, and product owners responsible for planning and delivering AI initiatives.
Executive Summary
The article critiques the application of T-shirt sizing in AI project estimation, highlighting five assumptions that systematically fail in AI contexts. It proposes Checkpoint Sizing as a more suitable approach, emphasizing iterative reassessment of scope and feasibility. The analysis draws on research into multi-agent systems, scaling principles, and conversation unreliability, underscoring the need for a human-centric approach in AI development.
Key Points
- ▸ T-shirt sizing assumptions often fail in AI projects
- ▸ Five fatal assumptions identified: linear effort scaling, repeatability, effort-duration fungibility, task decomposability, and deterministic completion criteria
- ▸ Checkpoint Sizing proposed as a more suitable estimation approach for AI initiatives
Merits
Evidence-backed analysis
The article provides a thorough examination of the limitations of T-shirt sizing in AI projects, grounded in recent research and evidence.
Demerits
Limited generalizability
The proposed Checkpoint Sizing approach may not be universally applicable, and its effectiveness in various AI project contexts requires further validation.
Expert Commentary
The article offers a timely and insightful critique of T-shirt sizing in AI project estimation. The proposed Checkpoint Sizing approach acknowledges the inherent complexities and uncertainties of AI development, emphasizing the need for iterative reassessment and human-centric decision-making. As AI projects continue to grow in scope and complexity, the article's findings and recommendations will be essential for project managers, technical leads, and product owners seeking to improve their estimation and delivery processes.
Recommendations
- ✓ Adopt a more nuanced and context-dependent approach to AI project estimation, recognizing the limitations of traditional agile techniques
- ✓ Implement Checkpoint Sizing or similar iterative estimation methods to improve the accuracy and adaptability of AI project planning