Architectural decay appears when a system stops absorbing change cleanly. Exceptions replace extensions, local decisions accumulate, and a once-stable structure begins behaving like legacy. No single symptom proves failure; the concern is a pattern of friction, workarounds, coordination, incidents, and growing infrastructure investment. section 4.3 section 1.2
Decay is driven by competing forces rather than by the absence of an architect title. Autonomous teams can preserve coherence when boundaries, interaction rules, and guardrails have explicit owners; freedom without accountability lets local choices erode system-wide assumptions. section 1.2
Treat recurring pain as architectural evidence when:
- Developers repeatedly avoid the same module, service, or integration, skip refactoring, or accumulate long-lived TODOs. section 4.3
- Small changes spread across unrelated modules or teams, producing broad pull requests, long regression cycles, or repeated coordination meetings. section 4.3
- Workarounds, recurring production incidents, or rewrite discussions indicate that existing assumptions may no longer hold. section 4.3
Table 4.1 Architectural rigidity assessment
|
Signal
|
Severity
|
Evidence to
look for
|
Owner
|
Next action
|
|
Developers
report repeated friction in the same part of the system
|
Low if
occasional, high if every change in that area becomes painful
|
Repeated
complaints, long-lived TODOs, skipped refactoring, recurring delays
|
Team owning
that module or service
|
Review the
boundary, dependencies, tests, and ownership around that area
|
|
Small
changes require increasing coordination or testing
|
Low if
limited to one release, high if unrelated teams are repeatedly involved
|
Pull
requests touching many modules, long regression cycles, release coordination
meetings
|
Architect
with affected teams
|
Identify
why the change spreads and whether a seam, contract, or module boundary is
missing
|
|
Developers
hesitate to touch certain modules, services, or integrations
|
Medium if
only one person avoids it, high if the whole team avoids it
|
“Only X
knows this,” delayed fixes, fear of breaking production, lack of tests
|
Engineering
lead or module owner
|
Create a
recovery plan: improve tests, document behavior, add observability, and plan
incremental cleanup
|
|
New
features are implemented through workarounds or exception
|
Medium if
temporary, high if workarounds become normal delivery practice
|
Feature
flags used permanently, special-case code, manual operational steps, bypassed
standards
|
Product and
technical owner together
|
Decide
whether the workaround should be removed, formalized, or treated as a signal
that the architecture no longer fits
|
|
Rewrite
discussions become more frequent
|
Medium if
frustration-driven, high if linked to repeated delivery or incident pain
|
Repeated
rewrite proposals, failed refactoring attempts, escalating maintenance cost
|
Architect
and leadership
|
Separate
emotion from evidence: identify the specific constraints, cost of change, and
possible incremental replacement path
|
Use the architectural rigidity assessment to connect each signal to evidence, an owner, and a proportionate next action. section 4.3
