FutureJobPath logo
The career map for the AI era
This page explains how the Durability Score is built — the components, the evidence behind each one, and the named sources. For who this work fits and what a career path through it looks like, see the Deep Read. For your personalized match, take the free quiz.
Where the 28 comes from.

Three components - Automation Resistance, Structural Moat, and Demand - add up to 28.

FJP Durability Score
28/100
Automation Resistance
8/40

Direct AI exposure is severe for routine coding, tests, and documentation, while the small remaining protection comes from legacy systems, debugging, domain context, and knowing why an old codebase behaves the way it does in production.

Sub-components
Substitution Resistance
3/30

Observed AI exposure is about 74.5%, and modeled median job-loss pressure is about 55.2%. Those signals fit the work: routine code, tests, scripts, translations, documentation, and bug fixes can be drafted by AI. Legacy-system knowledge and domain debugging keep the value above the floor, but not by much.

Sources feeding this sub-component
Anthropic labor-market impacts → Shows very high observed exposure for computer programmers.
Tufts American AI Jobs Risk Index → Models high job-loss pressure for the occupation under the median scenario.
Augmentation Leverage
5/10

AI is useful for explaining code, drafting functions, writing tests, converting syntax, and summarizing errors. The gain is real, but much of it flows to employers who need fewer routine code hours. Because demand is weak, leverage does not translate into strong worker protection.

Sources feeding this sub-component
Anthropic Economic Index usage-primitives report → Shows coding and document-generation tasks among common AI uses.
Stack Overflow Developer Survey 2025 → Shows mainstream developer-tool adoption and AI workflow context.
Structural Moat
13/35

The barrier is codebase and domain depth, not a license or physical setting, so protection is modest outside hard-to-replace legacy environments where mistakes are expensive and institutional memory still matters to the employer over time.

Sub-components
Physical & Environmental
0/10

Programming is desk and screen work. The federal occupational profile and task descriptions point to indoor computer-centered work, so there is no physical setting that slows substitution. Any protection has to come from judgment, domain context, or codebase depth instead.

Sources feeding this sub-component
Bureau of Labor Statistics Occupational Requirements Survey → Provides the federal physical-requirements baseline used across occupations.
Bureau of Labor Statistics Computer Programmers profile → Describes the work as writing, modifying, and testing code and scripts.
Regulatory Moat
1/12

There is no broad occupational license for computer programming. Degrees, certificates, vendor familiarity, and portfolios can help a hiring case, but they are employer signals rather than legal gates. That keeps the formal protection low even when a specific employer needs a trusted maintainer.

Sources feeding this sub-component
CareerOneStop licensed occupations data → Lists licensed occupations and does not show a broad programming license.
Archbridge State Occupational Licensing Index → Provides the licensing-burden cross-check used across occupations.
Robotics Resistance
8/8

Physical robotics is not the substitute path for this occupation. The pressure comes from software that writes, explains, tests, and modifies code. Keeping the robotics lane separate prevents software automation from being counted as if it were a physical deployment problem.

Sources feeding this sub-component
IFR World Robotics papers → Provides the physical-robotics deployment context used across occupations.
Credential Depth
4/5

The federal profile and occupational-preparation data place the occupation in a higher-preparation zone, usually tied to a bachelor-level path or substantial portfolio proof. That preparation matters, but it does not protect the routine code-to-spec layer from AI the way a required license would.

Sources feeding this sub-component
O*NET Online 15-1251.00 → Shows Job Zone 4 preparation for the occupation.
Bureau of Labor Statistics Computer Programmers profile → Names bachelor-level education as the typical entry path.
Demand
7/25

The labor-market signal is weak because the dedicated occupation is smaller, declining, and mostly supported by replacement openings rather than expansion, even though legacy systems and business-specific code still create some paid maintenance seats for new entrants.

Sub-components
Volume
1/10

The dedicated federal row is small for Tech: about 121,200 jobs and about 5,500 annual openings. Employment is declining, and the openings are mainly replacement flow. That gives some labor-market scale but a weak base for a new entrant choosing a multi-year path.

Sources feeding this sub-component
Bureau of Labor Statistics Employment Projections → Shows 121.2K jobs, -6.0% growth, and 5.5K annual openings for computer programmers.
Source Quality
3/8

The source match is clean, but the demand source itself is not favorable. Public data and the occupational profile point to replacement hiring, task automation, and a shift of higher-skill work toward software developers. Legacy code maintenance keeps the source quality above churn-only.

Sources feeding this sub-component
Bureau of Labor Statistics Computer Programmers profile → Names programmer duties and the automation pressure on repetitive tasks.
GitHub Octoverse 2025 → Provides developer-tool and AI-adoption context for software work.
Resilience
3/7

Resilience is limited because the active shock is inside the task core: routine programming can be accelerated or consolidated. The work holds better when it involves old systems, business-specific rules, and debugging where mistakes are expensive, but those are narrower seats.

Sources feeding this sub-component
Tufts American AI Jobs Risk Index → Models high AI-related job-loss pressure for the occupation.
JetBrains State of Developer Ecosystem 2025 → Provides developer-ecosystem context for AI coding-tool adoption.
What would move the score
Scenario 1
Routine programming consolidates faster

The case weakens if employers use AI tools and senior developers to handle routine code, tests, documentation, and bug fixes with fewer dedicated programmers. The exposed jobs would be ticket execution with little design, domain, or production ownership inside software teams.

Direction
down
Components affected
Automation Resistance, Demand
Scenario 2
Legacy modernization creates steadier seats

The case strengthens if organizations keep needing specialists who understand older languages, mainframes, custom business rules, and risky integrations. The threshold is not old code alone; it is paid demand for people who can change old systems safely over several years.

Direction
up
Components affected
Demand
Scenario 3
Programmers move into developer work

A mixed outcome needs review if the title keeps shrinking but workers successfully shift into software development, systems analysis, security, or data engineering. The original job would still be weak, but the skill path would be less fragile during retraining.

Direction
neutral
Components affected
Automation Resistance, Demand
Personalized job matches →
Want to find the careers that fit your specific profile? Take the free FJP quiz — 3 personalized matches.
Last reviewed June 2026 · Next September 2026