Terms-Drift Table
An engineering economics framework for AI programming tool procurement decisions. By quantifying the expected cost of terms-change risk, it helps engineering teams make rational choices among free channels, paid managed services, and self-hosted deployments.
Timeline (last 90 days)
The monthly cost formula for the free tier is F = Rw * C + (p * E * C) / W.
The monthly cost formula for paid hosting is P = I + (p_paid * E_paid * C) / W.
The monthly cost formula for self-hosting is H = M.
Only when the free-tier cost F is lowest and all hard thresholds are met should one stay on the free tier; otherwise, buy or host.
The number of team engineers N is deliberately excluded from the cost formula and treated only as background information.
The formula deliberately omits productivity gains brought by AI tools because such gains are difficult to decouple from the tool versus engineer capability, and including them would introduce a confirmation bias entry point.
In the example (Rw=8, C=110, I=1800, M=4600, E=16, W=3, p=0.5), the calculated costs are free F≈1173, paid P≈1844, and self-hosted H=4600, with free winning on the cost line.
If rework Rw rises from 8 to 22, the free planning cost is approximately 2713, which is higher than paid, and the decision reverses in favor of the paid option.
If the terms weight p rises from 0.5 to 1.0, the free exit load doubles to approximately 587, giving F≈1467, which in this example is still lower than paid, indicating that terms-based fear alone is insufficient to reverse the decision.
If the team handles sensitive information and has not read the data terms within the current week, both hosted paths fail to pass the threshold and self-hosting wins, making cost no longer a valid argument.
2 more timeline events
All Facts (12)
The monthly cost formula for the free tier is F = Rw * C + (p * E * C) / W.
50%UnverifiedThe monthly cost formula for paid hosting is P = I + (p_paid * E_paid * C) / W.
50%UnverifiedThe monthly cost formula for self-hosting is H = M.
50%UnverifiedOnly when the free-tier cost F is lowest and all hard thresholds are met should one stay on the free tier; otherwise, buy or host.
50%UnverifiedThe number of team engineers N is deliberately excluded from the cost formula and treated only as background information.
50%UnverifiedThe formula deliberately omits productivity gains brought by AI tools because such gains are difficult to decouple from the tool versus engineer capability, and including them would introduce a confirmation bias entry point.
50%UnverifiedIn the example (Rw=8, C=110, I=1800, M=4600, E=16, W=3, p=0.5), the calculated costs are free F≈1173, paid P≈1844, and self-hosted H=4600, with free winning on the cost line.
50%UnverifiedIf rework Rw rises from 8 to 22, the free planning cost is approximately 2713, which is higher than paid, and the decision reverses in favor of the paid option.
50%UnverifiedIf the terms weight p rises from 0.5 to 1.0, the free exit load doubles to approximately 587, giving F≈1467, which in this example is still lower than paid, indicating that terms-based fear alone is insufficient to reverse the decision.
50%UnverifiedIf the team handles sensitive information and has not read the data terms within the current week, both hosted paths fail to pass the threshold and self-hosting wins, making cost no longer a valid argument.
50%UnverifiedThe framework specifies six hard thresholds (the data terms have been read by a named person within the last 7 days, sensitive data does not pass through channels whose retention policy cannot be described, the free server is not the sole remote option, there is a named owner who can suspend the channel, exceptions have expiry dates, and an exit drill has been executed) that cannot be overridden by a low F.
50%UnverifiedOwnership is defined as: the engineering manager owns the decision, the platform lead owns the exit drill, and finance owns the C and I input values.
50%