Sacré Bleu

Lunghu doesn’t often talk shop on this blog, but when he does … XXpect the unXXpected.

The occasion for this brief(?) departure from precedent is provided by two IBM acquisitions during the past week:  i2 (a British software house) and Algorithmics (a French-owned Canadian firm).  What does this mean?  Why should Lunghu care?

Let’s take these rhetorical questions in reverse order (sort of).  First and foremost, these acquisitions resonate for Lunghu because i2 dominates the market in law enforcement/ military intell analysis network visualization tools with its Analyst’s Notebook product.  This software generates graphical representations of criminal/ terrorist network structures, communications, and relationships.  It also has features for depicting multi-thread timeline event flows (such as those that occur during the course of a criminal conspiracy or terrorist plot).  In theory, such data visualizations can accelerate, improve or refine the process of intelligence analysis itself, but in the real world i2 charts are primarily used as a storytelling device, to explain a complex conspiracy to bonehead bosses or arrogant, self-important prosecuting attorneys.  Far too often, Analyst’s Notebook is used to produce a bogus corporate-style “table of organization” that purports to illustrate the network’s so-called (and sometimes imaginary) command-and-control hierarchy.

Lunghu has been using various versions of i2 software for about a dozen years, and he’s delighted that IBM has bought the company.  That’s because i2’s software –from an intell analyst’s viewpoint– pretty much sucks, and now there’s hope (however faint) for real improvement once some Armonk coders get their heads under the hood of this creaky old British jalopy.  Here’s a brief enumeration of the principal defects that a longtime user sees in i2’s flagship product:

  • Analyst’s Notebook’s market niche is in depiction/ visualization of network relationships –”link charts” in intell-speak.  However, the software has no social network analysis tools —nothing to assist in exploring network structure.  The software doesn’t have features to help the analyst identify cliques, cut points, k-cores … nada. Instead, users/analysts have to manually impose their own structure on the data visualization by dragging nodes to new locations on the chart.  i2 apologists might claim (with some jsutification) that the LE/mil intell industry doesn’t want all that high-falutin’ pointy-headed academic analysis capability  –it just wants to know who to go out and shoot.  So Lunghu poses the question: which is a better downrange use of your expensive munitions  –spray-and-pray or sniper fire?
  • Analyst’s Notebook generates only flat, 2D representations of mixed-mode network relationships.  It doesn’t have the capability to depict the interrelationships of multiple networks in a “layered” visualization –something that an i2 executive dubbed “2½ D” when Lunghu tried to explain to him why that would be useful.  It has a kludge that lets the user selectively filter out entire categories of nodes, but this feature is only marginally useful.  Analyst’s Notebook also cannot readily depict changes in a network over time – a 4D capability that is highly relevant to many aspects of intell analysis.  The current software requires the analyst to generate two (or more) 2D charts at different timeslices, and then (like a spectator at Flushing Meadows) scan back and forth between them hoping to notice all the changes.
  • Analyst’s Notebook link charts cannot be exported from the program to other applications: not as graphics, not as PostScript files, not as meaningful datasets, nothing.  This really annoys the military, which wants to see everything in a Powerpoint slide for briefings to command.  A grainy lo-res screenshot of your link chart doesn’t overly impress the 3-star, and makes your captain look bad.
  • Representation of network nodes is icon-based, which is okay as far as it goes, but the standard icon set looks dated, cheesy, and Brit-centric.  The software is picky about what types of graphic formats it will accept for substitute icon images, and does not allow image files to be substituted for node icons: they can only be superimposed on the icon, obscuring but not replacing it.  This means that when you want to represent a person, place or thing as a node on your chart, you can’t just use a filename or hyperlink for the visual representation of that node –you have to laboriously navigate a menu-driven wizard to select the file and place it on the chart.  Then you have to use another menu tool to tweak the formatting, size, etc.  For each such image …
  • Analyst’s Notebook is not well equipped to handle communications analysis.  Granted, that’s outside the scope of general intell industry practice at the national level in both the USA and UK, where NSA and GCHQ handle SigInt for the big boys.  However, within law enforcement, it is often the grassroots intell analyst who has to handle analysis of call records and intercept data.  The market leaders in this sub-niche are PenLink and JSI, both of whom focus most of their development efforts on administration of wiretap information: data collection, storage, and (generally) rudimentary reporting.  i2 has carved out its own space by specializing in the visualization of telecom data, basically as an extension of its link charting feature.  But as with social network analysis, i2 is weak on providing actual analysis tools to deal with telecom data.  Its principal product offering in the telecom analysis space is PatternTracer 2, which the company asserts …

Automates pattern analysis tasks within telephone call data
Identifies communication patterns happening around events of interest
Identifies repeating groups of calls (clusters) and patterns of calls within these clusters (the telecom equivalent of market basket analysis)

Lunghu’s principal problem with this tool is that it’s [1] outrageously expensive for what it does, [2] an opaque blackbox that uses algorithms of unknown precision and reliability, and [3] it doesn’t allow the user/analyst to specify, tweak  or weight any of the parameters used in analysis.  It’s not encouraging that i2’s sales and training staff has difficulty explaining what PatternTracer does and how it does it.  The good news is that these deficiencies inspired Lunghu to write his own telecom data cluster analysis subroutine … which actually works as intended!  And, since Lunghu knows what the algorithms are, and can specify the parameters before each analysis run, he has much more confidence in its output.  And guess what —the data can be imported into Analyst’s Notebook for visualization.  Fortunately for America’s civil liberties, no one cares and he never uses it.

But enough about i2 …

What’s the meaning of these IBM acquisitions in the larger picture?  Conventional wisdom might well have it that this investment in data analytics is a defensive move on IBM’s part, a reaction to HP’s purchase of Autonomy or Google’s recent buying spree (PostRank, PittPatt, etc.).  Lunghu thinks not.  Instead, these acquisitions suggest a keen awareness of the operating environment –and market opportunity– on IBM’s part.  Let’s review the bigger picture:

❖  The post-9/11 defense boom is just about over for the worldwide military/industrial complex.  Bucketloads of counter-terrorism money from DoD, DIA, and CIA gave new life to i2 just as its core law enforcement market was approaching saturation, but the bust cycle is about to begin in that corner of the ‘security’ industry.  i2’s technology is showing its age, but it’s still a recognized brand name in government and industry, and that’s what IBM bought.

Algorithmics, incorporated in Delaware as Fitch Risk Management, Inc, is a member of Fitch Group, majority owned by Fimalac, a holding company based in Paris, France. Fimalac is around 80 percent owned by its Founder, Chairman and CEO, Marc Ladreit de Lacharrière.  Algorithmics is the world’s leading provider of risk solutions for market, credit and operational risk.  Financial organizations around the world use Algorithmics software, analytics and advisory services to help them make risk-aware business decisions, maximize shareholder value, and meet regulatory requirements.  Twenty-five of the top thirty banks and more than two-thirds of the CRO Forum of leading insurers use Algorithmics analytics software and advisory services. Clients include The Allianz Group, BlueCrest, HSBC, Nedbank, Nomura, Societe Generale, and Scotia Capital.

Algorithmic’s focus on credit, market and liquidity risk, as well as key customers in operational risk, will strengthen and expand IBM’s risk consulting services.  Algorithmics risk analytics software and services, combined with IBM’s acquisition of OpenPages and recent investments in predictive analytics, will provide clients with the broadest range of business analytics solutions.

❖  To summarize, the salient features of this newly-acquired IBM capability are [1] Anglo-Franco-Canadian-American business presence and corporate relationships; [2] market leading position in law enforcement intell and investigative-support analytics and [3] market leading position in financial industry risk analytics.

❖  So, where do the vectors of law enforcement, finance and risk assessment intersect?  Market surveillance, oversight and regulation, that’s where.  Lunghu believes that these acquisitions are an indication that IBM has (or hopes to have) the inside track on a massive technology support contract for an in-depth SEC/ FINRA/ CFTC/ Fed/ Treasury market surveillance and enforcement program.

Of course, this could just be wishful thinking on Lunghu’s part, since he believes such measures are long overdue.  But there certainly are signs that the US regulatory authorities may finally be beginning to walk the walk behind their somewhat tougher talk.  Stated intent from SEC, FINRA, et al. indicate that securities fraud, insider trading, market manipulation, tax evasion and related criminality are once again high enforcement priorities.  In order to investigate far-flung transnational criminal networks operated by and for the speculative degrees, regulators need network analysis capability.  In order to detect market manipulation within the gushing torrent of global trading data, regulators need risk analysis and pattern matching capabilities.  Without in-house expertise to develop such capabilities themselves, they’ll have to buy or lease it from someone who already has the capability.  A trusted vendor with whom government has had a century of  experience.  Someone like Big Blue.

Further afield, the inclusion of British and French components within IBM’s analytics effort may indicate that the United States is undertaking this market surveillance/enforcement initiative in actual, concerted partnership with regulators in London and Paris.  By insuring that everyone’s interests are represented to some degree or another, you build a measure of trust and get additional buy-in.  Kinda like Libya, Part Deux.

Best of all, for IBM this is an opportunity to expand markets in both the public and private sectors.  Financial industry players who are operating in grey areas of the regulatory space (and that would be just about all of ‘em) may well see the benefit of hiring the services of IBM’s risk consulting group … just on the off chance of catching hints about what capabilities the feds might be developing.  That’s what counter-intelligence is all about.  And on the public sector side, there’s no need for IBM to confine itself to federal government.  Each of the 50 states operates an employee pension fund and a bond financing department.  Plenty of risk to be assessed there.  Plenty to be investigated, too.  If individual states can’t afford IBM’s consulting fees (and most states are deeply in the red), perhaps they can pool their resources and take a regional approach.  At least, that could be IBM’s sales pitch.

Will any of this occur as Lunghu has foretold?  Who knows?  On verra, on verra ça. … Mais pas çe soir.  Demain, demain, toujours demain.


Tags: , , , , , , , , , , , , , ,

%d bloggers like this: