How Network Analysis Can Help Identify Money Laundering Schemes

bayesimpact.org

82 points·pyduan··6 comments

Interesting analysis, but better without the names.

It's dangerous to publicly suggest that people might be laundering money without evidence (even with a disclaimer), especially people with unusual names on a site indexed by search engines.

That's debatable. First, it's public information he used. Anyone could have analysed it themselves if they wanted to find suspicious people. Further, if someone discriminates against these people just because their names appear on a website alongside "we're not accusing them of anything", then that person may be doing something wrong. But that person probably discriminates against all sorts of people for trivial reasons anyway, and maybe we should think of ways to make them behave in a more socially tolerable way if we don't like it.

Especially since they are liable to be sued in the UK, who have a very strict libel law.

Finding the datasets for these kind of endeavours seems like a full time job. Hats off to them for doing the work and making it happen!

TLDR;

Connect all people, addresses and businesses in the business register in a big graph, sort each node by its amount of neighbours and work from there through the list of best connected people.

The underlying idea is that money laundering requires a good network (of people, addresses, companies) while staying "under the radar", so unknown people that are well connected are potential money launderers.

Too bad the analysis can't be reproduced. I can find data about the companies (http://download.companieshouse.gov.uk/en_output.html) but not the directors.

Anyone has had more luck?

Reply on news.ycombinator.com