I, XBRL Robot

You see lots of talk about the consumers of XBRL being investors, analysts and of course the regulators. At this point our customers are really only focused on the SEC. But there's another audience that is potentially as important as the regulators: The Bots.

Over in Marketing, Google has transformed practices both for B2B and B2C companies. A whole industry has sprung up to help companies manage SEO (Search Engine Optimization aimed at natural search results) and SEM (Search Engine Marketing or ad buying) in order to take advantage of the way Google and other search engines index information and help people shop and look for information.

I was honestly stunned to learn the other week that in 2007 30% of all trades were automated and that figure is expected to hit 50% by 2010. Also, investment firms, hedge funds for instance, are using bots to scan market data, press releases, news stories and the blogosphere for intelligence that can help them trade faster and smarter. It's one way to gain an edge.

Lightening fast communications to a broad audience plus automation are changing the market dynamics in some interesting ways. You'd expect these technologies to feed back into the information creation process for external reporting/investor relations in the same way they have for marketing. But there's work to do even while that type of optimization develops. Like everything else, along with these new opportunities come new risks.

The poster child for what can go wrong in all of this is United Airlines. This past September a single person read an old article about UA's 2002 bankruptcy on a Florida news site early on a Sunday morning. Because traffic on the site was so low it pushed the story onto the most read section. Google spidered the site, found the story and the bot didn't recognize it was old (guess that news site doesn't use NEWSML) and pushed it into Google's news feed. Within minutes people from all over the place were reading the story which probably created a positive feedback loop.

Cut to Monday morning and an "analyst" sees the story on Google's news feed and quickly places it on his firm's Bloomberg service. From there a Bloomberg "analyst" put it on to to Bloomberg's main feed and then all hell broke loose. People started to sell the stock thinking the backruptcy news was current and again the bots took over and the stock crashed.

One of the big causes in all of this was the fact that the story was ambiguously dated. XBRL removes ambiguity in data so that's a good thing. The other causes were lightening fast communications, the need for speed demonstrated by those so-called analysts and traders and the use of automation. Put them all together and it greatly amplified the situation. XBRL and other interactive data types will certainly accelerate the speed of communications and the use of automation and I wonder if the greater clarity in this data will offset the potentinal risks inherent in speed and automation. The lack of ambiguity will increase the trust in the data, as it should. Will controls lag? I think the point is that small errors can have large impacts in this world.

So it all comes down to data quality. Any honest appraisal of the data submitted to the SEC so far will show that quality is all over the place even before we start talking about any strickly accounting related issues. Our role as intermediaries is to push ourselves hard for the best quality. We do and we explain at every opportunity that we've set up our process to get the best quality and help people understand what that means. I'm not using quality in some fuzzy marketing sense here. Even today there are pretty solid standards that set the minimum requirements for good data. (See the XBRL US Taxonomy Preparer's Guide PDF for instance.)

Automation is a great thing. Until . . . well, you've seen the movie.

-- Ed Hodder