15th November 2017

Yeah, no worries, cool

Last month the FCA announced that it was working with the regulatory risk analytics firm, Corlytics, to develop an “intelligent regulatory handbook” for the FCA in what it says is a world first.

The impact, we are told, is that instead of being an online textbook split into several blocks and modules, Corlytics has “worked with the team at the FCA to turn the handbook into a fully searchable database”.

This sounds fantastic news, “yes, no, very good” in fact.

I get the impression that the fictional world of the BBC programme W1A lives in real time, if that is not a contradiction in terms, and it is based in Canary Wharf.

I imagine that the meeting to decide to commission Corlytics was littered with comments like:

• “Let’s nail this puppy to the floor!” .

Or in regard to the improved functionality:

• “A couple of great ideas, right out of the bat, but we’re looking at here, is an idea so good you don’t just want to marry it, you want to take it home and f*#k it.

• OK guys, if we don’t whack this raccoon first time we’re looking at a total crapfest. No question.
  

Well, when we heard, we just had to ask, “how much did this cost”?

The reply will sadly not surprise anyone, but, why is so much spent on what is essentially a vanity project when a lower cost and different methodology could have been obtained or applied?

We have seen through previous Panacea FOI requests around new logos and website changes for the FSA to FCA transition that money is no object, especially when it is your money.

The FCA reply read as follows:

We refer to your request under the Freedom of Information Act 2000 (“the Act”) for “… the total cost incurred in producing the ‘intelligent regulatory handbook’.”

Your request has now been considered and we can confirm that the total cost of this work is £367,138.80.

Yours sincerely

 Information Disclosure Team

Financial Conduct Authority

We asked Jon Steer, the technical director and all round tech wizard at our IT agency, Durham Associates to give us some of his thoughts.

He noted that a slightly more technical slant on what has been done can be found here but it “still doesn’t make a whole lot of sense”.

Jon observed “Why would you use machine learning on a single publication unless you then wanted to use what was learnt for something else”? Are there other handbooks or archives that could then be indexed using an AI tool that has learnt all the relevant ‘jargon’ via the work with the FCA?

“Surely it would have been cheaper to have humans read the manual and tag each paragraph with the metadata rather than teach a computer to do it”.

He went on the observe that is was “difficult to know exactly what has been done and what the benefits are until it is launched – and even then so much will depend on the ease of use of the interface as much as the clever bits behind the scenes”.

It is interesting that Corlytics most recent blog entry is talking about AI and automation in financial software.

Jon points to this as being a possible answer:

Corlytics solutions use a dataset that ‘aggregates and standardises regulatory enforcement… Each case is reviewed by specialist lawyers who annotate the case providing metadata and insight’.

Perhaps their newly taught AI would be able to do some of that work autonomously.

So, the question is why did it cost so much if Corlytics could also be enjoying benefits from it? Possibly even with their offering to financial professionals with their Financial Dashboard.

And especially as the FCA website already has a substantial search function that seems to work very well.

To draw again on the W1A analogy, “If ever there was an opportunity for the FCA to stand tall and make a big, bold statement about how much it values the idea of valuing values, then surely this is it.”

And it only cost £367,138.

Panacea Comment, Regulation