SDSG Integrated Analytics Workshop (14 Oct 2011)

Open PDF
Creating agency: GCHQ Government Communications Headquarters (UK)
Document date: 2011-10-14
Pages: 3
Genre: report
Banner:
Classification: SECRET
Dissemination: -

This document is a summary of a workshop held by the SIGINT Development Steering Group (SDSG) on October 14, 2011, evaluating and comparing the strengths, weaknesses, and potential use of advanced analytics tools, specifically Palantir and Mamba, for intelligence analysis. It discusses the tools' capabilities in data integration, collaboration, and visualization, along with concerns about security, usability, and the risks of relying on commercial software.

Long Description:
No extended description available.
Programs:
MAMBA, ECLIPSE, RUMOURMILL, VISION, JTRIG

People mentioned:
-

Countries:
-

Tags:
Palantir, analytic tools, collaboration, commercial software, corporate cooperation

Identifier:
SDSG-Integrated-Analytics-Workshop-WriteUp

PDF SHA-256:
c1a368bde285aa5329741cab6ecd3c6f1750557909c89c2b01f699442b87e651

SECRET
SDSG Integrated Analytics Workshop (14 Oct 2011)

Palantir

Opportunities & Strengths Risks & challenges
Easy to share / publish knowledge amongst the analytical community Collaboration capabilities less mature than Mamba offering
Timeline capabilities are visually appealing Analyst education about value and use of target knowledge versus target data
Effective navigation through target knowledge (right-click & explore around) Dealing with high volumes and big indexes
Sensible division between searching around entities for target knowledge and searching for events or 'raw' data Concern that the tool gives the analyst greater potential for going down too many analytical paths which could distract from the intelligence requirement
Collaborative app and 'step-back in analytical path' offers potential for sharing tradecraft Mindset change for analysts surrounding any platform for analysis and collaboration
Easy to do data integration – quick to deliver an 80% solution Restricted potential (due to being commercial product) – it develops how the owning company want it to develop
Could take any data Openness of commercial 'black box'
Flexibility Code could be better commented
Usability / intuitive interface appears to be a strength (at least relatively) How many 'versions' of Palantir would the company support?
International user community including SF and DSD Integration - with other platforms?
Tool responsiveness
Mature stage of development
User Interface and usability
Presentation capabilities
Accountability

Mamba

Opportunities & Strengths Risks & challenges
Talks to the vision (and thus is powerful) Are hard coded-questions restrictive?
Great data integration Usability, usability, usability
Has the potential to deliver more effective and efficient QC'ing capability than current systems provide Appears to be fairly complex with a steep learning curve

1 of 3

[REDACTED]

SECRET

SECRET

Opportunities & Strengths Risks & challenges
Allows the analyst to filter the data prior to importing it onto the chart - means analysts aren't wasting time deleting Complexity - what skill levels would be necessary or could different levels of access be matched to analyst skill levels
Potential for very intuitive exploration of data and relationships between QFDs Deliverability and complexity of eclipse user-interface
Strong on sharing of analysis Challenge of uniform data models
Potential for on-line coaching via screen sharing Relative immature progress on graph change
Integration agenda – flexibility and illustrates the art of the possible
Open architecture
Has the concept of a 'hypothesis'
Offers good potential for bringing more collaborative opportunities between Ops and JTRIG
Rich layered model allows unique navigation and viewing of details of the events, especially multi-participant c2c events
Mamba potentially offers greater potential for closer sharing and collaboration with NSA Though currently doesn't deliver this

Mamba and Palantir 'summing-up' feedback

"A completely commercial product, in this instance Palantir, is clearly more polished and delivers more in terms of usability, compared with the in-house built offering (i.e. Mamba-based products) which offers complete flexibility, arguably thus more potential but 'Eclipse' doesn't sell simplicity."

  • External collaboration is at least as important as internal collaboration
  • Interoperability important between whatever tool(s) are progressed
  • A desire where at all possible to avoid local specialism (if the applications can support that)
  • Concerns over what capabilities truly work at scale
  • Consider competition with tools already 'bedded in' in the business
  • Attendees acknowledge the best assessment is realised through using any capability rather than demonstrations alone
  • Visual representation of the data is a key component to explore
  • Usability / complexity / polish will have an impact on the users
  • Could whichever tool have an 'unlock' application features based on analysts skill level, hiding buttons and features until needed and capable of utilising
  • Experiment with diverse range of analysts, from Skill level 0 up to L3's
  • Fully understand and consider the time necessary for analysts to change / take on new tools and ways of working

Some concerns expressed that there is still a very real risk that as CDO / CDL roles continue to develop across the 4th floor that such analysts will need to become users of a vastly growing tool set and that decisions / agreements on tool choices and tool mix will be hard

2 of 3

[REDACTED]

SECRET

SECRET

A 'shared' decision going forward is needed

Concern over the data model implemented around the collated / fused data and how it supports and integrates itself with a data model around conducting actual analysis and the building of intelligence pictures

Performance of any application / tool / framework is always a key consideration
Can any tool cope with the diversity of data types?

Vision and Principals (feedback)
No broad concerns over vision or principals
Action for ADS: But simplify vision and key principles
Action for ADS: Can the principals be condensed from its current two pages?
Action for ADS: Can it be shared more broadly, perhaps on GCwiki and linked to from these workshop pages

IAC (feedback)
Requires a clear future and decisions
Generic statements that could be considered Ops requirements:

  • More independence between front-end user-interfaces and the back-end data, i.e. build service tiers (helps future proofing against new capabilities)
  • Easier linkages between social media (i.e. collaborative tools / applications) and the tools / applications that support analysis and data exploration
  • Don't necessarily deliver the capability in the engineer's current favourite technology
  • Usability for the 'whole' analyst community is key (avoid future kerfuffles)
  • There is a need to really understand the user base amongst the analyst community for any capability going forward
  • The CapDev community need to give quick assessments to ISD over particular new capabilities, especially to ensure the right tool choices are made and roadmaps are not stove-piped

Another concern
Concerns that the Olympic questions for Rumour Mill have not been broadly enough shared, at least some attendees had not had visibility of them
Action: ADS to circulate (included in email with these minutes) and available at:
[REDACTED]

Attendees
[REDACTED]
[REDACTED]
[REDACTED]
[REDACTED]
[REDACTED]
[REDACTED]
[REDACTED]
[REDACTED]
[REDACTED]

3 of 3

[REDACTED]

SECRET