BRENNEN

Industries / Government, education and community / Government / Military

Recruiting and training

Finding, testing and selecting people for the armed forces, by conscription or by volunteering, and setting the standards they train to.

7

Milestones achieved, from 1793 to 2023.

2

To watch for: announced, not yet confirmed outside.

3

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1793

    A whole nation requisitioned

    On 23 August 1793 France's National Convention decreed that "all Frenchmen are in permanent requisition for the service of the armies" and gave each group a task, beginning: "The young men shall go to battle; the married men shall forge arms". The decree, the levée en masse, put the whole population at the service of the war effort.

    sourcebooks.web.fordham.edu

    Credit

    the National Convention of France

  2. 1973

    Volunteers only

    A commission set up by President Richard Nixon in 1969 and headed by Thomas Gates recommended in February 1970 that the US end the draft. On 27 January 1973 Defense Secretary Melvin Laird announced that the armed forces "henceforth will depend exclusively on volunteer soldiers, sailors, airmen and Marines." CNA, a nonprofit research center, says the first years were rocky and pay fell behind until Congress raised it 11.7% for fiscal 1981 and 14.3% the year after; it puts the share of recruits with a high-school diploma at 71% under the draft and 97% later.

    LessonA volunteer force has to pay enough to compete with civilian employers.

    rand.orgcna.org

    Credit

    the Gates Commission; Melvin Laird

  3. 1980

    A test scored too easy

    From 1976 the Armed Services Vocational Aptitude Battery (ASVAB), the entrance test all US services use, converted raw scores to percentiles on a faulty scale that was too easy, so low scores were reported as higher percentiles than they were. Errors were first found in late 1976, studies by the Center for Naval Analyses in early 1979 showed the scale was still too easy, and corrected norms came in in October 1980. The National Research Council estimates that about 250,000 applicants who would not normally have been accepted enlisted from 1976 to 1980. Congress then capped the lowest-scoring group, Category IV, at 20% of recruits and non-graduates at 35%, and the crisis led to the Job Performance Measurement Project, which the National Research Council calls "the first DoD effort to attempt to formally validate enlistment quality standards against on-the-job performance criteria."

    LessonAn entrance test's scale has to be checked against how recruits actually perform, not only against other test-takers.

    nationalacademies.orgirp.wisc.edu

    Credit

    Center for Naval Analyses, for confirming the error

  4. 2016

    Women conscripted too

    In June 2013 Norway decided to make military service compulsory for women as well as men; the Barents Observer reported that this made it "the first European and first NATO member" to do so. In the 2016 intake, the first to include women, about 60,000 candidates competed for about 10,000 places and 32.7% of those chosen were women, Elisabeth Braw reported in War on the Rocks.

    thebarentsobserver.comwarontherocks.com

    Credit

    the Storting, Norway's parliament

  5. 2018

    Outsourced recruiting falls short

    In 2012 the British Army committed £1.36 billion over ten years to a recruiting partnership, including a £495 million contract with Capita, an outsourcing company. The National Audit Office (NAO), which audits UK government spending, found in December 2018 that the Army's yearly shortfall had ranged from 21% to 45% of its requirement, and that Capita missed the target for regular soldiers by an average of 30%, against 4% in the two years before the contract. The online recruiting system arrived 52 months late at £113 million, triple its budget; half of applicants for the regular ranks took up to 321 days to join, and 47% dropped out of their own accord.

    LessonHanding recruiting to a contractor still needs a buyer who understands the work; the NAO found both sides "underestimated the complexity of the Project.".

    nao.org.uknao.org.uk

    Credit

    National Audit Office

  6. 2022

    Fitness test rescored

    The US Army planned its Army Combat Fitness Test (ACFT) as one standard for every soldier, whatever their age or sex. An independent review by RAND, a research organization, ordered by Congress in the fiscal 2021 defense bill and published in March 2022, found that "The evidence base to support the ACFT is incomplete"; Army Times reported that 48% of active-duty enlisted women and 28% of active-duty female officers could not pass. The Army dropped the single standard and scored the test by age and sex from April 2022, and on 1 June 2025 replaced it with the five-event Army Fitness Test.

    LessonA standard that decides careers has to be shown to predict the job before it is applied to everyone.

    rand.orgarmytimes.comarmy.mil

    Credit

    RAND, for the review

  7. 2023

    Health records slow enlistment

    From 2022 US processing stations screened applicants through MHS GENESIS, the military's new electronic health record, which brought in each applicant's full medical history. The DoD Inspector General found in May 2023 delays of 9 to 40 days in processing medical waivers, permission to enlist despite a disqualifying condition, for the Army, Air Force and Marine Corps, and that the Navy's time from interview to contract rose from 33.8 to 63.4 days; the services had handled 54,206 medical waiver requests in fiscal 2021 and 2022 and approved 77%. In 2024 the Army put 68 nurse practitioner and physician assistant posts into 65 processing stations and said review time for medical prescreens fell from 4.3 to 1.1 days.

    LessonGiving screeners more information slows selection unless there are more people to read it.

    warren.senate.govairforcetimes.comarmy.mil

    Credit

    DoD Office of Inspector General

Watch for

  1. Germany's new military service

    Germany's military service law, in force from 1 January 2026, sends a questionnaire to young people that men must answer and women may, and makes medical screening compulsory for men born in 2008 or later from 1 July 2027. The government's target is more than 260,000 active soldiers and 200,000 reservists by 2035. In May 2026 the defense ministry said it had sent more than 206,000 letters and that about 86% of the men replied within four weeks, against about 3% of the women.

    Achieved whencompulsory screening begins in July 2027 and an outside body, such as the Bundestag's Commissioner for the Armed Forces, reports intake meeting the targets.

    bundesregierung.debmvg.de

    Claimed by

    German Federal Government and Federal Ministry of Defence, December 2025 and May 2026

  2. A chatbot for recruiters

    The US Army says its recruiter chatbot, a successor to the public "Sgt. Star" assistant first put on GoArmy.com in 2006, answers from "a curated repository of recruiting regulations", was built in under three months, had about 12,000 users in July 2026 and could reach 28,000 by November. The Army's articles on it give no accuracy figures.

    Achieved whenan independent evaluation of its answers is published.

    army.mil

    Claimed by

    US Army, July 2026

Open

  1. 01

    Fitness-test reps from video

    Software that counts and judges hand-release push-ups and times the plank in the US Army Fitness Test from phone video, checked against a panel of trained graders scoring the same filmed volunteers.

    The brief
    Credit · unclaimed
  2. 02

    Enlistment rule answers, scored

    A public set of questions would-be recruits ask about published enlistment rules (age, education, tattoos, medical conditions and waivers), each with an answer and a citation to the rule, from DoD Instruction 6130.03 Volume 1 on medical standards and the services' recruiting regulations. Score public AI chatbots on accuracy and on whether they give the same answer each time.

    The brief
    Credit · unclaimed
  3. 03

    Time to join, counted

    A public table of how long it takes to join the armed forces, from application to the start of basic training, and how many applicants drop out, by service and year, for as many countries as publish figures, each number linked to its source.

    The brief
    Credit · unclaimed

Open 01 · the brief

Fitness-test reps from video

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

Software that counts and judges hand-release push-ups and times the plank in the US Army Fitness Test from phone video, checked against a panel of trained graders scoring the same filmed volunteers.

Why it is still open

In a 2026 study in the Journal of Strength and Conditioning Research (Van Dam and colleagues), 17 West Point physical-education faculty trained to grade the test scored the same 10 videos three times; the most lenient counted 1,073 repetitions (85.0%) and the strictest 513 (40.6%), and the researchers rated consistency between graders as poor, according to a review of the paper. Camera counters exist: Republic Polytechnic in Singapore says its system counts push-ups and sit-ups "using standard camera devices" but publishes no accuracy, and a 2026 Brazilian paper tested pull-ups only. I found no Army tool and no published check of one against graders.

mtntactical.comrp.edu.sgscitepress.org

Built with what exists

MediaPipe Pose or MoveNet, open-source models that find a person's joints in video; OpenCV; rules written from the Army's published event standards; the intraclass correlation and Bland-Altman analysis to measure agreement with graders.

  1. First step

    Turn the push-up standard into a list of rules, film 10 sets from two phone angles, and score a MediaPipe baseline against three graders.

  2. Done looks like

    30 consenting volunteers filmed from two angles for both events; agreement with a three-grader majority on each repetition at least as high as the median human grader's; repetition counts within 2 and plank times within 2 seconds; videos, labels and code published.

Open 02 · the brief

Enlistment rule answers, scored

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public set of questions would-be recruits ask about published enlistment rules (age, education, tattoos, medical conditions and waivers), each with an answer and a citation to the rule, from DoD Instruction 6130.03 Volume 1 on medical standards and the services' recruiting regulations. Score public AI chatbots on accuracy and on whether they give the same answer each time.

Why it is still open

The Army's new recruiter chatbot has no published accuracy figures, and it is for recruiters, not applicants. Medical waivers are a bottleneck, with 54,206 requests in two years, and the DoD Inspector General found the services did not track whether permanently disqualified applicants asked for one. I found no published test of AI answers on enlistment rules.

army.milairforcetimes.com

Built with what exists

DoD Instruction 6130.03 Volume 1 and the services' published recruiting rules; Inspect, an open-source evaluation framework; AI models such as Claude, Gemini and GPT, each asked every question several times (see the Note "All twenty times").

  1. First step

    Download the current 6130.03 Volume 1 and draft 50 questions from its sections, with answers.

  2. Done looks like

    300 questions, each answer and citation agreed by two people working separately (agreement of 0.8 or better on Cohen's kappa); at least four models scored by rule category on correct, wrong and hedged answers, and on how often their answer changes across repeats. Rule lookup only: no medical advice and no secret testing of real recruiters.

Open 03 · the brief

Time to join, counted

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public table of how long it takes to join the armed forces, from application to the start of basic training, and how many applicants drop out, by service and year, for as many countries as publish figures, each number linked to its source.

Why it is still open

The NAO's 2018 report gave one snapshot for the British Army. In February 2025 the UK government set an ambition of "a conditional offer within 10 days" and "a training start date within 30 days", but its figures appear only in ministers' answers to parliamentary questions: in September 2026 a minister gave medians for July 2025 to June 2026 of 252 days in the Royal Navy, 247 in the Army and 356 in the Royal Air Force. I found no regular official series.

gov.ukukdefencejournal.org.uk

Built with what exists

The UK Parliament's written questions search; NAO, GAO and DoD Inspector General reports; German defense ministry releases; pdftotext; an AI model to pull each figure with its quoted sentence, checked by a person (see the Note "Seven more flagged"); a CSV file on GitHub.

  1. First step

    Collect every UK written answer since 2018 that gives a median or average time to join, and enter the figures with their dates.

  2. Done looks like

    At least three countries, and at least five years for the UK, with every figure carrying its source and quote; updated every quarter.