Constant's pations

If it's more than 30 minutes old, it's not news. It's a blog.

Wednesday, September 29, 2004

Iraqi Insurgency: 1M US forces needed by 2008-10 time frame

Nobody wages war to fight. They do so to win.

To prevail, the US needs a quantum leap in ground forces in Iraq. But the only way to meet those numbers is to have a draft.

This explains what numbres are needed, and why the insurgency is building faster than current planner are prepared to admit.

What can be done? This explain why the buildup will be coplex and what America needs to do. Update: 7 Jan 2005 Insurgency at 200,000, beating planners forecasts by three years, prompting grave concern in the Pentagon.

Executive Summary

Current combat-manning is insuffifient. The US requires an additional 1M troops to effectively reach max-rate by 2008, for a two year push to snuff the insurgency, expected to reach in the mid to high 200,000 by that point.

Such a change is a three-fold increase in combat forces. It will take three years using 20 CONUS training bases to ramp up and sustain a 1M man draft force.

Draft-numbers are expected to start in the low 37k per mos in year one, and reach a peak rate of over 100K/mos by 2007-8. These figures do not include the logistics-support requirement in the neighborhood of 3-5M.

The longer the US delays making the decision to reach the 1M target figure for conscripts, the more difficult the situation in Iraq will be. At worst, the US will simply fuel a civil war with inadequate troops and force protection.


Estimating the real number of insurgents based in incomplete, inaccurate, and politicized information.


A. Identify model that would estimate the number of insurgents

B. Develop reasonable assumptions that could be independently tested

C. Develop a number of insurgents; forecast US requirements to cover that problem.

Overview Discussion

If we assume the Iraqi insurgency is a training-system, we might make some assumptions about the number of training sites, the number of people who are in the training system, and the total number of insurgents.

One approach is to use waiting-line queue theory. Although the formals require certain assumptions, we can actually use the formulas to test the integrity of the DoD-data. In short, we start where DoD leaves us, and then use the formulas to derive their assumptions; in some cases we even see that the formulas create inconsistent assumptions indicating the DoD-data is problematic.

For starters, what we can assume is that the total number of potential insurgents arriving for training exceeds the kill-rate. Put another way this is saying that the slope of the y-arrivals exceeds the x-kill-rate.

M/M/1 models are exponential, have infinite populations, and their interarrival times are exponential. This is a very good fit with Iraq. Also, we have anecdotal reports that the insurgent-growth-rate/training-rats are both exponential as they have been characterized as "being able to fill their ranks faster" as is the "norm" in a guerrilla-type operation.

Thus, we can take the DoD-sourced data on kills, advance the slope by a lead-time of X1, representing the lead-time for training which ranges from 1 week to for weeks; or 100 hours to 700 hours.

Also, of interest is the number of training sites. We know this number is large, otherwise DoD would target them.

Also, we know that the insurgents move without communicating, and cannot be tracked, as evidenced by the large backlog of FBI-NSA intercepts that have not been translated.

The initial arrival rate of 4000 per month, and 12 training sites appears to be a reasonable assumption. Of concern is that at this rate, the insurgents will number from their present number in the orders of 100,000s. Also, this growth-rate is far higher than the US can sustain, implying that the US can only combat [on numbers alone] if the training rate increases for Iraqis [which does not seem probable] or institute a draft.

The logical question becomes: "How many US troops are required." Assuming we have two trend lines [Rate at which US will achieve peak-value; and the continuing exponential growth rate in insurgent training], we find that the 100,000 value in insurgents links with 1M US combat forces, assuming a ratio of 1:9, or a growth in US forces from the present 400K to 1.5M starting in 2005, and reaming 1M additional forces by 2008, representing 3 years of insurgent growth.

Further, assuming the US has a force-support index of 1:7 [meaning for each combat troop there are 7 support personnel], we quickly get into largest numbers in the 3.5 to 7-magnitude.

In general the real costs for the war will be munitions and fuel; personnel costs are assumed to be burdened [food, medical, equipment, pay] at $50K per year. The large numbers for personnel only amount to peanuts compared to the current spend rate of $1B per week.

At worst, the US will hit 1.5M troops in Iraq by 2008, and peak for 2 years.

If the US actively contributes to a civil war by fueling the insurgency, it remains to be seen whether a 2 year window will solve the problem.

At this juncture, it is not likely the US will easily rampdown; at worst the rate of insurgent growth would far exceed the US ability to ramp-up personnel. Even if there were 20 dedicated training centers brining in 34,000 per moth, the US would still require a three phase step-increase to reach 1M additional troops; this factors in the losses due to 1-year-duty-service.

On sheer numbers alone, the US has to make a decision quickly as to how many troops it wants in Iraq. The issue is not funding [which is already sky-high], but how quickly those troops are going to be inserted into Iraq, at what level, and how quickly the various units around the country will be deployed.

The longer the US waits to come to grips with what is happening in Iraq, the slower and more delayed the decision will be to ramp up to 1M troops, plus support. In the meantime, the insurgents are expected to increase their manning and popular support.

Estimating the Insurgent numbers with a reliable model

  • Two problems: Available support forces insufficient; and the actual insurgent numbers are not only low, but they have greater latitude than forecast

    1. Fewer Iraqis available. Only 5,000 Iraqi forces trained, not 95,000.

    Last February, Secretary Rumsfeld claimed that more than 210,000 Iraqis were in uniform. Two weeks ago, he admitted that claim was exaggerated by more than 50 percent. Iraq, he said, now has 95,000 trained security forces. But guess what? Neither number bears any relationship to the truth. For example, just 5,000 Iraqi soldiers have been fully trained, by the administration's own minimal standards. And of the 35,000 police now in uniform, not one has completed a 24-week field-training program. Ref

    2. Insurgent ability to work is broader. Firm analysis from Special Operations Consulting -- Security Management Group Inc.

    According to a new report on the attacks occurring daily in Iraq, in the past month alone more than 2,300 attacks have been directed against civilians and military targets. The violence sprawls over nearly every major population center outside the Kurdish north. The sweeping geographical reach of the attacks suggests a more widespread resistance than the isolated pockets of insurgency described by Iraqi government officials. The study was conducted by Special Operations Consulting.Ref

    NY Times Mentions the report: Iraq Study Sees Rebels' Attacks as Widespread
    By JAMES GLANZ and THOM SHANKER, Published: September 29, 2004 ref

    Confirmation of greater problems

    More countries, but unsustainable troop lvels. Ref; Ref

    Running the numbers

    Total = 24.55; Data source: 11 of 55

  • Model on service time


    Assuming an input of 4,000 per mos, 5.5 arrivals/hour.

    Servicetime: 4.166/hour

  • Blending

    720 hours per mos. Blending model. Fiddling with numbers until get some boundaries:


    Arrivals ____per hour 5.1
    Service time ____hours 100 [1/7 of a month, 720 hours]
    Number of agents 2000 training sites


    Productivity %
    Outbound calls per secondminute (throughput)
    Average waiting time secondsminutes
    Service level % waits less than secondsminutes

    Sensitivity Analysis

    If we cut down the service time to 100 hours and say 80% wait less than 80 minutes, we can get away with 530 locations; or if we have 2000 sites, we can increase the output by 4x.


    1. Arrivals ____per hour 5.1
    2. Service time ____hours 100 < ---- reduced training Blank: ---- > Output

    3. Number of agents ______ --- >530
    4. Threshold ______ --- > 530
    5. Productivity % ______ ----> 96.2
    6. Outbound calls per minute (throughput)
    7. Ave wait time ______ ----> 83.5

    8. Service level _____ 80% waits less than ___ 100 minutes


    Upper end: Where the number of arrivals starts to approach the required number of operators

    6 arriving/hour is equivalent to 4320/mos

    Using this model:


    1. Arrivals ____per hour 6.0
    2. Service time ____hours 700 < ---- 4 weeks training Blank: ---- > Output

    3. Number of agents ______ ---> 6301 [1:1 training: trainer]
    4. Threshold ______ ----> 6301 [1:1 training: trainer]
    5. Productivity % ______ N/A
    6. Outbound calls per minute (throughput)
    7. Ave wait time ______ N/A


    8. Service level _____ 80% waits less than ___ 100 minutes


    Finding critical values from the number of trainers


    5.1 arrivals/hour
    700 training hours

    Critical value [flips from 0 to infinity, and vice versa]
    -- around 3580 trainers


    Input: Arrivals ___ 5.1 hour
    Service time ___ 700 hours

    Forced: Number of agents 3580 < ----- adjust up and down to examine

    Output: Watch for a change from infinity to a discrete number

    Average waiting time _____ 3387.88 minutes
    Service level % waits less than ____ 19.34 minutes


    US needs to make a decision on max-troops required to combat an insurgency that is growing at a sustainable-exponential rate. If the number is 1M, then that will take three years with the following:

    2005 2006 2007
    Year1 Year2 Year3


    per mos 34 68 136

    bi-weekly 17 34 68

    If the peak number desired is higher than 1M, it will more draft notices per month than the initial 34k.

    If the number of troops desired to reach peak rate is desired to be before 2008, the number of training bases and personnel per/month needs to be increased from 34K/most and 20bases.