A past time even less productive than tweeting
This past week I was invited to speak on text analytics at the MRMW 2012. The event was extremely well attended and I’m excited about the opportunity for text analytics in mobile.
Because of our early work in social media analysis with firms like Facebook and LinkedIn I was also asked to take part in a panel discussion on “Social Media Research Standards” which seemed to overlap a bit with text analytics standards. The MRMW conference seemed to be encouraging the idea of industry standards (with no less than three one hour panels dedicated to this topic).
The term “standards” sounds so good. Who wouldn’t be a supporter of standards? All great companies have internal standards. I’ve been lucky enough to always work at firms which have had the highest research standards, and at Anderson Analytics (OdinText) we continually strive to be better than any of our competitors.
But when we talk about “Industry Standards” that’s an entirely different thing. A standard is a norm or requirement. It’s easy to discuss the idea of standards in general forever without getting anywhere. When discussing standards in general, especially standards in new and emerging areas like text and social media analytics, there are many good arguments for not wasting time with standards (a few which were pointed out by me and other panelists):
- Text analytics and social media analytics are evolving quickly, it’s a moving target, and no one can predict what tomorrow will yield.
- Standards are OK in fairly irrelevant and established areas such as size of light bulb sockets. However, standards especially in technology, have a history of being supported by industry leaders when it correlates with their business strategy (here they are used to hinder innovation and competition by smaller or newer entrants). In those cases when they do not correlate with industry leader’s strategy they are ignored.
- Standards really only make sense when there is a clear benefit to everyone involved which is very rare indeed.
It’s impossible to be in support of standards in general. There seems to be only two reasons why the discussion of standards in general has been so popular among market industry trade orgs lately. It gives them a sense of much needed relevancy at a time when this is being questioned. In some cases like that of the failed ESOMAR and CASRO push of ISO research standards for the US, it was hoped that certification procedure would provide them with a new revenue stream.
Even though there were three panels on standards I did not hear a single suggestion of a standard that was specific enough to be discussed. Bandying about terms like “standard sentiment accuracy” in an area that is handled differently by different vendors and is viewed by some as proprietary IP is indicative of a naiveté in regard to the field. The idea that market research trade orgs, which have extremely limited knowledge of this field, should play any part in this is ridiculous. I’ve been very active in the text analytics community since 2005 attending events like the Text Analytics Summit and Sentiment analysis Symposium, and know the CEO’s of the companies in this industry would be as interested in this tautology as I am.
The other term bandied about by trade orgs pushing the idea of standards is “Privacy”. Again, a loaded word. The biggest argument here is that if somehow market research firms would be able to agree on some sort of self-imposed privacy standards related to social media or big data they would be exempt from possible future legislation in this area.
This too is ridiculous. Firms like Google, Apple and Amazon etc. have a greater reason and far more money to fight any eventual legislation in this area than the market research industry. We have already lost the battle of the minds among the general population, which see no difference between surveys and other spam. The days of exceptions are gone.
This said I’m happy to make a decision on a specific suggested standard. I suspect I would be as likely to accept such standard only if it supports our company goals, and would expect every other company to do the same. I imagine the chances of all of us agreeing on any standard would be small. Personally I’ve got better things to do than invent standards for the sake of standards. Keeping on top of the dynamic change in both text analytics and social media is a full time job.
[Disclosure: I haven't always been so against quality standards. Throughout my career I've had extensive training in several of these initiatives from Quality Circles/TQM back in the early/mid 90's to Six Sigma Master Black Belt training in early 2000. A few years ago Anderson Analytics was part of a special steering committee to plan and implement industry wide standards for data mining (at the time promoted heavily by SPSS). All of these approaches to better industry standards were once touted as a must for any successful business, yet all have been retired to the dust bin of business trends. Each required considerable investment of time with extremely limited if any ROI]