1. Inhibitors to Information Sharing
Anticipating How Information Sharing Efforts May Fail
2010
Walter Kitchenman
wkitchenman@hotmail.com
2. Purpose
• Identify challenges to Info Sharing among multiple depts. and agencies
• Show how Info Sharing adds value as users view data in new ways
• Direct the analysis to a general audience
• Focus on general functional requirements
1
5. OVERVIEW> DEFINITIONS
What is Info Sharing?
“Information sharing includes the cultural, managerial and technical
behaviors by which one participant leverages information held or created
by another participant.’’ - US Department of Defense
• Content and Knowledge Management
• Mining and deduping data from distributed databases
• Modeled data and a display of intellectual assets that facilitates expert
analyses
4
6. OVERVIEW> DEFINITIONS
Information is Created by Individuals
Info Sharing isn‟t simply an IT solution for exchanging content on
distributed databases. People must contribute and use the system.
• The Lone Wolf Scenario
– Information product developed by individuals
– No central repository
– Individuals have proprietary methods and knowledge
• The “Old School” Scenario
– Key managers may be technology averse
– Main work product on paper and in personal files
5
7. OVERVIEW> SOLUTION DEGRADATION
Even Successful Info Sharing Solutions Fail
Under Several Scenarios
• Information Overload
– Systems do not „connect the dots‟ opening the door to human fallibility
• The False Positive Scenario
– Search criteria or alerts are too broadly based
• The Turnover Scenario
– New systems projects are complex and long to complete, org. change
outpaces development times leading to a disruptive need to rebuild consensus
• The Ideological Scenario
– Alpha errors based on ideology or politics compromise data shared
– Pushing the envelope in Info Sharing challenges societal or corporate values
• Solution Degradation
– A failure to adequately test and upgrade software, hardware and data models
after implementation
6
8. OVERVIEW> FUNCTIONALITY
Good Info Sharing Provides a Means to see
Info in New and Improved Ways
In addition to making conventional reports available across departments
and agencies; modeling techniques segment a population, neural
networks recognize patterns and suggest actions; and behavior, objects
or individuals are flagged for further review by domain experts.
• Dashboard/Portal
– Push of recent reports and content, flagged reports and alerts
• Auxiliary Profiles for Report Components
– Important sub-sets of info in reports (e.g., a Profile of George Washington to
which all books in which he is mentioned are linked)
• Modeling and Neural Networks
– Population divided into segments to predict statistically probable behavior
– May resemble profiling by race, gender or other factors prohibited by law
– Variables or inputs „fire‟ an algorithm to recognize patterns and recommend
certain outputs or actions
7
9. OVERVIEW> EXAMPLES
Modeling Data - Many Enhanced Info Products
Use Non-Linear Modeling and Neural Nets
Non-Linear neural nets and other modeling techniques are used to mine
data in a wide variety of fields.
• SAS Institute (Cary, NC), Norkom (Dublin, Ireland) and Mantas (Fairfax,
VA) solutions identify money laundering through pattern tracking and
suspicious individual watch lists
• Fair Isaac‟s Falcon Credit Card Fraud Management System spots
suspicious card purchases in real-time
• IBM‟s Non-Obvious Relationship Awareness (NORA) determines the
relationships between people (initially developed for the gaming industry)
• Neural Nets are used to predict short-term increases in the NYSE
Composite Index
8
10. OVERVIEW> EXAMPLES
Modeling Data – Some Types of Segmentation
Geographic Segmentation Demographic Segmentation
SINGLE
WOMAN
ELDERLY
COUNTRY REGIONS
YOUNG
FAMILY
Attitudinal Segmentation
POPULATION COUPLE
GRAND-
PARENTS
SINGLE
POPULATION ATTITUDES MAN
9
11. OVERVIEW> EXAMPLES
Modeling Data - A Neural Network
HIDDEN
INPUT
• Neural networks process non-
linear statistical data and model
OUTPUT
complex relationships between
inputs and outputs
• Basic network consists of input
layer (variables), a hidden layer
of high dimension, and an output
layer
10
12. OVERVIEW> INHIBITORS
Inhibitors to Info Sharing
Inhibitors to adopting successful Information Sharing Solutions are
interrelated and involve people at almost every stage.
1. Conceptual
2. Economic
3. Technological
4. Cultural
5. Personal
11
14. Conceptual Challenges to Info Sharing
Concepts at inception determine the accuracy, scalability and future
functionality of the Info Solution.
Failed Concepts Stem from Four Key IT-Related Issues
1. Scope
2. Architecture
3. Design
4. Functionality
13
15. CONCEPTUAL>SCOPE
Scope
A failure to identify the best solution within the context of likely budget
constraints and other limitations can undermine ultimate success.
Factors Defining the Scope of an Info Sharing Solution
• Budget and Time-Frame
• Constituents
– Who are the participants and what confidentiality is mandated?
– Geography
• End Products
– Dashboard of available data? Real-time alerts?
• Legal and Cultural Restraints
• Roll-Out and Projected Growth
– How will solutions be rolled-out? By agency or dept.? By geography?
– Anticipated growth of content and participants (critical for IT decisions)
14
16. CONCEPTUAL>ARCHITECTURE
Architecture
A failure to identify core elements that information shared across
departments have in common can result in poor information architecture.
• Core Organizing Principle
– One element every piece of content has in common regardless of dept. or
agency (e.g., publishers link books to an author)
• Taxonomy and Components
– Find a minimum number of categories or key words (e.g., books as fiction, non-
fiction, biographies, mysteries, etc.)
• Deduping and Parsing
– Names, places, objects have variable spellings, misspellings or name changes
• Database Management
– Which type of dbase management best meets the functionality desired?
– Understand the benefits and limitations of any choices proposed
15
17. CONCEPTUAL>ARCHITECTURE>DATA EXCHANGE
Architecture – Integrating Different Formats
from Distributed Dbases
The product of different agencies and IT environments is made more
consistent and exchangeable by developing common standards.
• Data Exchange Standards - A Common Shell
– Content in almost any format and from many different dbases (distributed) can
be linked to a shell with common tags
• Newly Shared Info May Still Require Restrictions
– Permissioning protects confidentiality and personalizes disclosures by user
– Should info disclosed be personalized by additional factors such as taxonomy,
content type, dept. or agency, geography or other?
• Accessibility and Usability
– How will users access the information? Web-based? Closed network?
– In general there is a conflict between strong Info Security and user friendliness
16
18. CONCEPTUAL>DESIGN
Design – Pages, Templates and Navigation
that Display Information Best
The effectiveness of a solution may hinge on how well it is organized into
discrete pages, such as a Dashboard and Search Results, with well
designed content which may include graphs, reports, profiles and alerts.
• Info Architecture
– Site Map showing Pages of the Info Sharing System and the likely navigation
• Templates Map
– Number of distinct templates for pages and content displayed, e.g., Dashboard,
Profile Pages of auxiliary components (e.g., George Washington Bio) if
applicable, Alerts, search results
17
21. CONCEPTUAL>FUNCTIONALITY
Functionality
Functionality varies greatly by enterprise, but in general a system should
highlight the latest content and incorporate Decision Tools.
• Automating Actions and Alerts
– Activity (or threat) profiled, e.g., stolen cards for electronic goods
– Variables or attributes that identify a profiled activity, e.g., buying flat screen TV
– Automating alerts and routing actions, e.g., requiring call to customer service
• Personalization
– Disclosure of info tailored to user with automated requests for more confidential
info routed to the originator when necessary
– Means of protecting data may be deemed insufficient by key contributors
• Admin Functions and Speed-to-Market
– Amount of control of contributor over content, tagging and permissions vs. IT
depts. will be controversial and impact on speed-to-market
– Audit trails and reporting are metrics to measure success and build confidence
20
23. Economic Inhibitors to Info Sharing
The internal budget process and policies on contracting agencies, vendors
and headcount often involve timeframes that defeat urgent projects.
Accuracy of Financial Projections Hinges on Four Key Factors
1. Budget and Timeline
2. Documentation and Planning
3. Communication and Training
4. Maintenance and Enhancements
22
24. ECONOMIC>BUDGET AND TIMEFRAME
Are the Budget and Timeframe Realistic?
The lower the budget and more urgent the timeframe, the more important
the Inception Phase, workarounds and anticipation of obstacles.
• Introduce the Info sharing Solution in Logical Phases
– Hedge against budget overages by rolling out the solution in phases with
hooks for enhancements
• Timeframe
– Identify the internal processes for budgeting, submitting change-orders for a
project, that might need to be modified to meet deadline
• Using Contractors and Vendors
– The process for vendor selection, procurement, and the use of contractors
may need to be modified to meet deadlines
23
25. ECONOMIC>DOCUMENTATION AND PLANNING
A Proof of Concept and Detailed Description on
Paper Will Save Time Later
Requirements may change between initial requirements gathering and
implementation of new systems.
• Documentation
– Temptation is to spend money on development alone
– Inception Phase and proof of concept are strongly recommended
– Documenting a project as it progresses requires resources from inception
• Project Planning
– The Project Plan is built around milestones and deliverables and the types of
inhibitors identified here should be considered in terms of timeline and budget
– Mid-development work will stop as unanticipated issues are addressed
– Avoid delays by identifying Decision-Makers at each constituent agency/dept,
and for the overall project, who are empowered to respond within 24 hours
24
26. ECONOMIC>COMMUNICATION AND TRAINING
Constituents Using an Info Sharing Solution
May Value Perceptions as Much as Reality
Turnover within the enterprise may require a continual process of
consensus building and an on-going internal communications effort.
• Communication Plan
– Budgeting resources for a Communication Plan to promote the effort internally
is often overlooked but recommended in cases where Cultural and Personal
challenges loom large
– Constituents often value perceptions as much as substance
• Training
– New hires and existing personnel require education efforts about the use of
the Info Sharing Solution
25
27. ECONOMIC>MAINTENANCE AND ENHANCEMENTS
The Info Sharing Solution Will Require
Significant Annual Support
Info Sharing Solutions introduced will require significant annual support,
equal to 15% of development cost on the IT side alone. A failure to
account for IT and model degradations can defeat the effort.
• IT Enhancements
– Any solution will consist of hardware and software that will be upgraded over
the life-time of the system on an annual basis
• Solutions Degrade Overtime
– All underlying assumptions behind any advanced modeling techniques
require constant testing and refinement and this is particularly true in cases
where the macro-climate changes rapidly (e.g., a credit crisis)
26
29. Technological Inhibitors to Info Sharing
Get the fundamental concepts or approach right at inception in order to live
within budgets and provide scalability and future functionality.
Technology Decisions are Made Early Affecting Three Key Areas
1. Design and Usability
2. Legacy Environments
3. Project Plan
28
30. TECHNOLOGICAL>DESIGN AND USABILITY
Design and Usability Should be Considered
From Inception
• Design
– Incorporate creative design from inception (often the „look‟ is slapped on last)
– Technical choices made at the beginning will limit Design choices later
• Info Security
– Will data be extracted from existing secure, token based environments?
– Will security protocols for sign-in and verification discourage use of the system?
• System Architecture
– Do you need a Database Management System (DBMS) that facilitates sharing
distributed dbases and conceptual modeling of data?
– Choose solutions that fit the projected scale of the production system identified
when considering the Scope of the project
– What are the performance requirements: Consider speed and timeouts?
– What are the benefits and limitations of various tech choices?
29
31. TECHNOLOGICAL>DESIGN AND USABILITY>SEMANTIC DATA MODEL
A Semantic Data Model is Probably Best
Suited for Advanced Info Sharing Solutions
The logical data structure of a database management system (DBMS)
cannot totally satisfy the requirements for a conceptual definition of data.
• Traditional Database Management Systems (DBMS)
– Hierarchical, network or relational
• Semantic Data Model (aka Conceptual Data Model)
– Techniques to define the meaning of data within the context of its
interrelationships with other data
– First recognized by the U.S. Air Force in the mid-1970s as a result of the
Integrated Computer-Aided Manufacturing Program (ICAM)
– Conceptual schema control transaction processing in a distributed dbase
environment (e.g., U.S. Air Force Integrated Information Support System I2S2)
30
32. TECHNOLOGICAL>LEGACY ENVIRONMENTS
Legacy Environments Create Migration Issues
IT departments at different agencies deploy and have resources
dedicated to support solutions provided by specific vendors (e.g., Oracle).
• Diversity of Technology Environments
– What are the IT (or non-technology) environments that contribute data?
– How much data does each IT environment have to share and what is the
format, e.g., paper, .doc files, emails, other electronic, etc.?
– Are changes already scheduled that will inhibit or facilitate Info Sharing?
• IT Evangelism
– Is a particular contributing agency or individual tied to certain technologies and
vendors? Will this have an impact on the best Info Sharing solution?
• Window for Change
– It generally takes a lot of time to introduce big IT changes (Typically 18 mos)
– There is generally a small window to push IT changes without putting ongoing
operations at risk (e.g., avoid close of business each year)
31
33. TECHNOLOGICAL>LEGACY SYSTEMS>DATA EXCHANGE
Data Exchange Formats Must Accommodate
the Enhanced Functionality Required
Ideally shells can be created and items tagged with a common taxonomy
automatically with human intervention limited to a relatively few exceptions.
• Migration of Legacy Data
– How will existing info in many formats and from different environments be
migrated to, or exchanged with, any joint Info Sharing System?
– Think through the best use of XML or other shells as first outlined in the section
on Conceptual inhibitors
• Blind Spots
– If information from a number of distributed databases is being shared
successfully what blind spots, if any, exist?
32
35. Cultural Inhibitors to Info Sharing
Cultural inhibitors are generally due to strong individuals within an
organization that set the tone. Many turf fights are well justified.
Three Basic Environments Impact Content Ownership
1. Legally Dominated Environments
2. Info Security
3. Data Ownership
34
36. CULTURAL>LEGALLY DOMINATED ENVIRONMENTS
Legally Dominated Environments Tend to
Discourage Info Sharing
Collective experiences of companies, departments and individuals create
cultures that impede Info Sharing.
• Fear of Discovery
– Information destroyed on a schedule to hedge against liability actions or other
judicial proceedings
• Sharing Proscribed
– Sharing of information may be prohibited, e.g., anti-trust regulations and
criminal proceedings
• Lengthy Review
– To provide for one institutional voice, designated parties review content
released to external parties adding days or weeks to a process
35
37. CULTURAL>INFO SECURITY
Info Security Dominated Environments are
Built to Protect Data Not Share It
Any hacking of the environment or data destroys the culture and model so
all data is protected equally – complicating Info Sharing.
• Data Protection Paramount
– No tiered levels of data confidentiality since no breach is tolerated
– Not accessible by Internet and may require security tokens (e.g., SecurID or
biometrics)
• Hosting Environment Problematical
– Production environment not fully disclosed complicating development
– Applications of varying confidentiality levels housed on same servers (based on
resource management since all data is protected equally)
36
38. CULTURAL>DATA OWNERSHIP
Data Ownership Regimes are Generally
Meant to Restrict the Release of Information
Risk averse cultures make it unlikely that one individual can quickly sign off
on Info Sharing and the concurrence of several departments is required.
Identify the culture that you must engage and possibly change.
• The Bureaucracy
– Several individuals sign off on data to be released internally or externally
• The Data Owner
– Within a dept., a Data Owner, controls info and its release
• The Chief Data Officer or Chief Information Officer
– An Executive-Level Manager develops and oversees compliance with policies
generally designed to restrict or protect data
37
40. Personal Inhibitors to Info Sharing
Successfully implemented Info Sharing Solutions fail if people do not
contribute data or incorporate the system into the daily work-flow.
1. Organizations Lose Interest
2. Reluctance to Share
3. Commissars and Apparatchiks
39
41. PERSONAL>ORGANIZATIONS LOSE INTEREST
Organizations Lose Interest
Organizations can lose interest over time in projects that seem high
priority at any given moment.
• This Year‟s New Initiative
– New initiatives introduced by consultants every 12 – 18 months
– Info Sharing is one of many projects a jaded bureaucracy has seen before
• Turnover
– Managers move on to new roles or new organizations before completion
– Initial sense of urgency dissipates
40
42. PERSONAL>RELUCTANCE TO SHARE
Reluctance to Share or Use a New Solution
Info Sharing is neither the primary objective of the enterprise or the
individuals upon whom successful Info Sharing depends.
• Undermining the Core Mission
– Sharing intellectual assets, especially outside the environments with which
people are familiar, is viewed as undermining the core mission
• Confidentiality of Sources at Risk
– Fear of losing control (and suffering the consequences)
– People who change jobs tomorrow need those promised confidentiality today
• Changing Personal Work Flow
– Individuals don‟t change their way of doing work to use, or contribute to, the
Info Sharing Solution
41
43. PERSONAL>COMMISSARS AND APPARATCHIKS
Commissars and Apparatchiks Survive
Through Mastery of the Old Info Doctrine
An organization‟s trusted team may not facilitate Info Sharing projects.
Roadblocks may come from those who are very senior (commissars) or
very junior (apparatchiks) as survival techniques are challenged.
• Commissars
– Greatest saboteurs negotiated their way through org. changes (purges) long
before the latest Info Sharing interlopers
– Info Sharing efforts require IT skills outside the comfort zone and decades-long
successful processes and survival techniques are challenged
• Apparatchiks
– Automatically enforce restrictive core business model or culture they know
– Value to organization is navigating arcane systems and procedures
42
45. SOLUTIONS>SUMMARY
Get Initial Concepts Right and Motivate the
Right People
• Get Concepts and key taxonomy right upfront with a solid Proof of Concept
• Understand the benefits and limits of key tech decisions (e.g., DBMS)
• Identify how legacy info is migrated and data exchanged
• If sharing information from distributed databases what are the blind spots?
• Identify all business processes that must change
• Empower specific individuals to make project decisions within 24 hours
• Make the Info Sharing effort a critical part of job performance reviews
• Build consensus with a well executed Communications Plan
• Get the lead in an Info Sharing Project sufficient status (anticipate turnover)
• Use newly Shared Info to improve the product (e.g., Data Modeling)
44
46. For More Info –
http://www.linkedin.com/in/wkitchenman
Walter Kitchenman is an author and consultant on strategic issues in financial services. He
spent more than a decade as an international banker in Latin America and Europe and helped
launch the leading boutique advisory firm covering the strategic use of IT. Most recently he was
VP in charge of knowledge management at MasterCard Worldwide. He has a graduate degree
from Johns Hopkins School of Advanced International Studies (SAIS) and BA with special
honors from the Elliot School of George Washington University.
2010
Walter Kitchenman
wkitchenman@hotmail.com