Data is not a noun

I use this article to present a thesis that we can all be better served if we start treating information as a verb.

The phrase ‘data-driven enterprise” is widely abused. Is the phrase truly expressing a choice? Versus what? Part of the thesis here is to become aware that (the same) data could drive you in different directions. Very similar people (all seemingly rational) could receive/review the same ‘data,’ interpret in their context and come to wildly different conclusions on the topic at hand.

I am suggesting that this phenomenon is the result of treating data as a noun, where data are seen as immutable objects, or as sets of facts and statistics whose interpretations are obvious, self-evident and even binary.

Treating data as a verb is effectively about ‘receiver context’ and ‘relevance realization.’ Verbing data is about fostering dialog, thought and the co-development of stories that precipitate action (whether in business or elsewhere else in life).

I present this thesis in three parts:

- The current context and the roots for this thesis (David Casagrande wrote a paper about this in 1999 drawing on a lot of wisdom from Shannon and Weaver from their seminal work on information theory.)

- The 5-stages of data internalization when it is treated as a noun (apologies if this sounds like the 5-stages of grief . . . but like with grief, denial is one of the stages!)

- Building a bridge: The stages of verbing data.

The Business Context & Roots in Information Theory:

How often have you come out of a meeting wondering ‘how people got to the decision that was made based on (or regardless of) the data that was discussed?’ Does it upset you even more because ‘we are a data-driven organization’ is part of the overriding spiel all around?

Avinash Kaushik states this as: “Data, good data, smart analysis, is less impactful than we would all like because it brings with it two things almost all company cultures are geared against: Reality they don’t want to know.”

David Casagrande’s 1999 paper titled ‘Information is a verb’ triggered this thesis. David’s paper had its roots in Shannon & Weaver’s writings (1949) and this abstract gives you the gist: “Current notions of information are inadequate for cognitive models because (i) they only account for information gain that reduces uncertainty (ii) assume binary logic (iii) fail to account for semantics and pragmatics and (iv) cannot account for shared and externalized cognition . . . A different model . . . ‘information as a process of state change (i.e., the term is used as a verb).”

The 5-stages of Data Internalization:

I would posit that we could all make some incremental improvements if we consider the transmitter’s preparation (the business context, the receiver’s context, before getting to an accurate narrative/visualization of the data) and the receiver’s preparation (how does this data fit with my mental model of our business, and can I remain open-minded if it doesn’t fit with my priors). This could result in some non-trivial order-of-magnitude improvements in data transmission and its reception. But as Morgan Housel reminds us “when confronted with a pile of dull facts and a pile of compelling anecdotes, the anecdotes are always the path of least resistance for your brain to cling to.”

I have therefore found it instructive to codify the stages of data internalization in the human/business context as:

1. Do I (and should I) believe this?

2. Do others believe this? Will we be able to make several others also believe it?

3. (When confronted with conflicting information), question the validity of source(s) or the credibility of the messenger(s).

4. Consider/evaluate/investigate to look for conflicting or validating information. Can this data be ‘explained away?’

5. Is this questioning our past decisions or current actions?

Does this sound unduly cynical? ‘5-stages’ is arguably a misnomer though we’ve all seen it play out this way in a number of instances. The real issue is, are some of these behaviors apparent around you?

The current process (or jumble of reactions) doesn’t behave like this when data is validating what we believe. This really comes into play when data questions what we believe.

Building a bridge — Verbing data:

The 5-stages described above may sound like the 5-stages of grief, but are all facets of ‘Relevance Realization,’ a concept expounded in some detail by John Vervaeke. And ironically, I found the same lens of Relevance Realization to be the catalytic bridge that helps the verbing of data. The key is that the process of verbing applies equally to data transmitters (data-scientists and analysts) and data-receivers (the operations business partners and decision-makers). It’s a shared process for human information processing.

In that spirit, we focus the dialog on:

- BUILDING TRUST: Agree that the data are credible and acknowledge (known) limitations.

- RELEVANCE: Agree on the relevance of the data to the business or components of the business.

- LOCAL MAXIMA: What do the data directly say at a topical level or in specific terms? What are the possible tactical implications?

- GLOBAL MAXIMA: What (if anything) do the data imply more broadly? Possible implications for business in general?

- CELEBRATE:

o Celebrate discovery.

o Recognize the contradictions (or validations) with current/past knowledge.

o Acknowledge the (remaining) unknowns without causing stasis.

- ACT: What do we need to create the confidence to act?

o If we act, what is the scale and range of potential impact?

o Is there a role for experimentation?

o Who are the other counter-parties and stakeholders in these actions?

There’s no 6-letter acronym here to feign a flowing pattern of desirable behavior. That would be unduly, even inappropriately, formulaic. We are not engaging in a conversation where one side is ‘selling’, and the other side is ‘believing’. In contrast, data are triggering dialog and thought. Data are not treated as passive sets of artifacts. Data are engaged to verb and reverb.

This approach involves an attitudinal reset and a broad commitment to actively verbing data; and if we aren’t motivated to do that, well then, for some of us, data will remain a noun.

But if you buy the thesis, then I have to ask: ‘how do you data?’

Notes:
1. Is ‘data’ singular or plural? The grammar can be confusing, but for purposes of this article, the word is used in singular to denote the concept and in plural when referring to sets of numbers. Here’s an interesting article from The Economist.
2. Relevance Realization is one of John Vervaeke’s many concepts. I have borrowed and reapplied here in a narrower domain.
3. David Casagrande’s original paper can be found here.
4. Avinash has great and relevant content on www.kaushik.net
5. Morgan Housel’s article can be found here.
* Attribution required for these images.