Value: neoCOCOMO

sloccount priced software with COCOMO: the effort to write it by hand. That figure is still printed for comparison, but it no longer measures value, because an agent can re-type the code for a fraction of it. What stays expensive is what an agent can't regenerate from the repository: behaviour that is written down nowhere except in the code, and the history of fixes that shaped it.

neoCOCOMO follows the accounting cost approach (replacement cost new, less obsolescence), with replacement priced for the agent era.

The model

classic     = 2.4 × KSLOC^1.05                        COCOMO 81 organic (person-months)
capture     = 0.5·tests + 0.25·contract + 0.25·docs   share of behaviour pinned down outside the code, 0..1
reproduce   = classic × (0.10 + 0.45 × (1 − capture))  agent-era rewrite effort
knowledge   = 0.025 PM × fix commits + 0.005 PM × other commits
rediscover  = knowledge × (1 − capture)               lessons that live only in code and history
replacement = reproduce + rediscover
value       = replacement × leverage × obsolescence

Term	Definition
KSLOC	Physical source lines of product code (non-blank, not comment-only), like sloccount.
tests	test/source token ratio ÷ 0.5, capped at 1.
contract	interface level ÷ 4.
docs	doc tokens ÷ source tokens ÷ 0.2, capped at 1.
floor (0.10)	Even perfectly specified code needs review, integration and verification.
rediscover share (0.5)	Re-deriving undocumented behaviour costs up to half the original effort.
fix commit (0.025 PM)	About half a day to find, understand and fix a defect. Fixes are recognised from commit subjects (fix, bug, hotfix, regression, crash, revert, …).
other commit (0.005 PM)	About 0.1 day of decision-making behind an ordinary change.
leverage	0.75 + 0.125 × integrability index (0.75–1.25): integrable systems can be put to use by others.
obsolescence	staleness (1.0 if the last commit is ≤ 12 months old, 0.85 if ≤ 36, else 0.7) × (0.85 + 0.05 × legibility).

Money uses sloccount's defaults ($56,286 salary, 2.4 overhead) so the two figures are comparable. Set your own with --salary and --overhead.

Reading it

Value (neoCOCOMO, cost approach; see neosloc/value.py)
  Classic COCOMO (sloccount):    4.6 KSLOC -> 11.8 person-months, 6.4 months, $133k
  Behaviour captured:            38% (tests 0.14, contract 0.25, docs 1.00)
  Reproduce with agents:         x0.38 of classic -> 4.5 PM
  Rediscover uncaptured history: 102 fix / 474 commits, 19 authors -> 4.4 PM knowledge, 2.7 PM uncaptured
  Replacement cost new:          7.2 PM
  x leverage 0.90 (integrability) x obsolescence 0.81 (staleness, legibility)
  Value:                         5.2 PM ~ $59k  (at $11k per PM)
  Knowledge at risk:             38% of replacement cost lives only in code and history

Two things the model makes visible:

• Tests, contracts and docs make code cheap to replace, but they are where much of its value lives. A well-captured project has a low agent factor: the artifacts carry the behaviour, so the code itself is the replaceable part.
• Knowledge at risk is the share of replacement cost found only in code and history. It is the bus-factor number SLOC never showed. A small frontend with 1 KSLOC but 421 commits from 50 authors can carry 70% of its replacement cost in knowledge nobody wrote down.

Caveats

This is a cost-approach estimate, not a market valuation: it ignores revenue, users and network effects. All coefficients are in neosloc/value.py and are uncalibrated.