David Greenwood

download David Greenwood

of 24

Transcript of David Greenwood

  • 8/8/2019 David Greenwood

    1/24

    Lessons from the Failure

    and Subsequent Success ofa Complex Healthcare

    Sector IT Project

    David Greenwood, Ali Khajeh-Hosseini,Ian Sommerville

    ST ANDREWS

    DEPENDABLE SYSTEMS ENGINEERING GROUP

  • 8/8/2019 David Greenwood

    2/24

    Disaster Strikes! An attempt to automate a pre-existing manual system to

    dispatch ambulances fails!

    Lives are Lost Estimate of20-30 people died as ambulances arrived toolate on the scene

    Money LASCAD cost 20 million to develop & deploy and was

    decommissioned within 10 days (26/10/92 - 4/11/92)

    Jobs LAS Chief Exec John Wilby resigns within 72 hours of go-live

    October 26th 1992

    2

  • 8/8/2019 David Greenwood

    3/24

    Success is in the air!!

    the project was turned around

    the system was delivered resulting a60% reduction in ambulance dispatchtime

    September 1996

    3

  • 8/8/2019 David Greenwood

    4/24

    What happened in those 3 years and11 months that turned a disaster intoa success story?

    The Mystery

    4

  • 8/8/2019 David Greenwood

    5/24

    IT failures diagnosed as errors at thetechnical or project management level areoften mistakenly pointing to symptoms of

    failure rather than the projects socio-complexity which is usually the actualsource of failure.

    Tools that enable practitioners tosystematically elicit socio-complexityderived project risks reduce a projectsrisk of failure

    Condensed Abstract

    5

  • 8/8/2019 David Greenwood

    6/24

    London AmbulanceService An NHS Trust providing an

    ambulance service to the wholeLondon area (620 sq miles)

    In 1990s involved in aprotracted pay dispute withworkers union

    In 1991, 268 seniormanagement posts in the LASwere cut to 53

    Restructuring caused a greatdeal of anxiety to employees

    1.1 The Case study - LASCAD

    6

  • 8/8/2019 David Greenwood

    7/24

    London AmbulanceService Computer AidedDespatch (LASCAD) Is an exemplar of an IT enabled

    work-transformation project

    Comprised the automation ofambulance dispatch from call takingto ambulance dispatch

    The need for automation was notedin the 1980s but a first attemptfailed in 1987

    1.2 The Case study - LASCAD

    7

  • 8/8/2019 David Greenwood

    8/24

    A second attempt was initiated inOctober 1990 and failed in October1992 (LASCAD92)

    The planned implementation date wasJan 92 but by March 1992 live trials ofthe system were suspended due to lack

    of confidence On 26th October 1992 the system

    crashed spectacularly

    1.3 The Case study LASCAD92

    8

  • 8/8/2019 David Greenwood

    9/24

    Following a public enquiry the project wasrevitalised under new management

    Lack of disciplined technical approach(Page, 1993) (Finklestein, Dowell, 1996)(Hougham, 1996) (Beynon-Davis, 1999)

    Poor user involvement and ownership(Page, 1993) (Beynon-Davies 1995)(Finklestein, Dowell, 1996) (Beynon-Davis, 1999)

    Irrational persistence to continue(Page, 1993)(Beynon-Davis, 1999)

    Lack of adequate testing(Page, 1993)(Beynon-Davies 1995) (Finklestein, Dowell, 1996)

    Limited training(Page, 1993)(Beynon-Davies 1995) (Finklestein, Dowell, 1996)

    1.4 The Case study The Fall out

    9

  • 8/8/2019 David Greenwood

    10/24

    A radically different approach was adopted COTS solution rejected -> in-house development

    Participative prototyping adopted

    Incremental development subject to user approval

    The consequence:

    In September 1996 the project was successfullyturned around resulting a 60% increase inproductivity

    1.5 Case study LASCAD96

    10

  • 8/8/2019 David Greenwood

    11/24

    2. The Problematical Situation

    During an IT development project trade-offs aremade due to constraints (such as technicallimitations, time and budget) resulting in somestakeholders benefiting/losing more than others

    As a consequence different people havedifferent views of how much resource a projectshould be given and what/how trade offs shouldbe made and whether or not they will supportthe project in its current form

    The complicatedness of a project can thereforebe understood in terms of

    technical complexity socio-complexity (human/organisational/social

    complexity)

  • 8/8/2019 David Greenwood

    12/24

    Socio-complexity is a construct thatrepresents the complicatedness ofinteractions between a projects

    stakeholders It is important as the socio-complexity of

    IT projects is known to adversely affect ITproject performance in terms ofscope,time, budget and user satisfaction farmore than technical complexity

    2. The Problematical Situation

    12

  • 8/8/2019 David Greenwood

    13/24

    A conflict perspective was adopted in order tomake the study of socio-complexitymanageable

    This means that to get a better understandingof the phenomena, research was focused onstakeholder interactions that comprise conflict

    This proved to be a break through as it revealed

    some of the constituent parts of socio-complexity at an actionable level of abstraction

    3. A perspective to make sense of themess

    13

  • 8/8/2019 David Greenwood

    14/24

    3. A perspective to make sense of themess

    14

    Individual Intra-group Inter-group

    Perceived

    Injustice

    Values &Satisfaction,

    Status

  • 8/8/2019 David Greenwood

    15/24

    Stakeholder ImpactAnalysis

    an approach to

    identifying sources ofsocio-complexity (latentpotential conflicts)

    therefore identifyingsocio-complexityderived risks

    4. A tool for practitioners to managethe mess

    15

    Tasks

    Assess impact

    of

    Resources

    Capabilities

    Values

    Satisfaction

    Status

    Time

    Multiple Roles

    Justice

    Input:

    Information

    about

    proposed

    changes.

    E.g. Who,

    what, how

    Process:

    Assess

    changes to

    existing tasks

    and

    processes to

    predict/identif

    y risks ofresistance/co

    nflict.

    Specific Identified

    Risks

    Output:

    Specific

    identified risks

    of resistance

    &opportunities

  • 8/8/2019 David Greenwood

    16/24

    The LASCAD case study was used toevaluate the effectiveness of StakeholderImpact Analysis (SIA)

    Test 1: Do risks identified using SIA in LASCAD92 maponto the sources of failure identified in other accounts?

    If true this provides some support that the socio-complexity derived risks account for the symptomsidentified in other accounts

    Test 2: Were risks identified using SIA in LASCAD92 weremitigated in the successful LASCAD96 project?

    If true this provides some support that SIA identifiesrisks that adversely affect IT project performance

    5. Method

    16

  • 8/8/2019 David Greenwood

    17/24

    Test 1 All sources of failure identified in accounts of

    the LASCAD disaster map onto identifiedsocio-complexity derived risks

    Test 2 Almost all the risks identified in the failed

    LASCAD 92 project were mitigated in thesuccessful LASCAD 96 project

    6. Results

    17

  • 8/8/2019 David Greenwood

    18/24

    Source: LAS Managementwere fearful for their jobs ifthey did not deliver LASCAD.

    Resultant: reluctance toreport negative informationto executives

    Why LASCAD92 went so wrong

    18

    Consequent: an irrational persistence to continuing

    Source: LASManagement distrustfulof control room staff &ambulance crew due toprevious conflict.

    Resultant: Negativeinformation discountedas politicalmanoeuvring

  • 8/8/2019 David Greenwood

    19/24

    Why LASCAD92 went so wrong

    19

    Source: The system was not explicitly designed taking intoaccount Ambulance crews values of rapid response

    Resultant: The system did not take into account crew experience

    or local knowledge

    Consequent: Ambulance crew resisted the system once it was deployed

    as its directives and automated selections did not match crew values.

  • 8/8/2019 David Greenwood

    20/24

    Why LASCAD92 went so wrong

    20

    Source: The system was not explicitly designed taking intoaccount Control Room staff status & job satisfaction

    Resultant: Software was perceived by staff to reduce their status

    & satisfaction by automating their work and removing manyelements of skill/local knowledge that existed

    Consequent: Control room staff opposed the system in March 92

  • 8/8/2019 David Greenwood

    21/24

    Why LASCAD96 went so right

    21

    Source: LASManagement weregiven a flexibledeadline

    Resultant: no

    disincentives toreport negativeinformation toexecutives

    Consequent: a rational persistence to continuing

    Source:Executivesresolved pay &conditions disputewith ambulance

    crew & controlroom staff

    Resultant:Negativefeedback taken

    seriously

    Source:Developmentprocess required asign-off fromcontrol room staff

    prior to go live Resultant:

    Negativefeedback aboutfunctionality could

    not be ignored &testing needed tobe convincing

  • 8/8/2019 David Greenwood

    22/24

    IT failures diagnosed as errors at the technical orproject management level are often mistakenlypointing to symptoms of failure rather than theprojects socio-complexity which is usually theactual source of failure.

    Tools that enable practitioners to systematicallyelicit socio-complexity derived project risks

    reduce a projects risk of failure

    Summary

    22

  • 8/8/2019 David Greenwood

    23/24

    Questions

    23

  • 8/8/2019 David Greenwood

    24/24

    Data Collection:

    Data on the LASCAD was collected from 5 separatesources and verified that each account broadlycorroborated one another

    Data Analysis: 1. Key stakeholders were identified

    2. Changes to key stakeholders work activity were identified

    3. The consequences of these changes were hypothesised on the stakeholderstime, resources, capabilities, values, status and satisfaction

    4. These changes were interpreted within the wider context of relational factors(e.g. tense relationships between individual & groups)

    5. It is hypothesised whether stakeholders will perceive the change as unjust(either procedurally or distributively) based upon nature of change and relationalcontext

    5. Method

    24