Vinton G. Cerf

Imagine for a moment a company I will call ACRONYM that maintains warehouses around the world from which customer orders for a variety of products are made. Imagine further that some of the inventory is in digital form (think movies, music, ebooks) in addition to being physical in nature. ACRONYM’s access to information in digital form turns out to be enormously enabling. The data is readily processed by sophisticated algorithms. This column is pure speculation on my part but inventing the imaginary ACRONYM Company helped me imagine how powerful it can be to have data available that permits fact-based analysis of business performance, projections for future business operations, and understanding of the marketplace the business serves.

ACRONYM gets its orders online, so they are already in digital form, with information ACRONYM needs to fulfill the order. Customers provide all the information requested since, without it, ACRONYM can’t or won’t place the order. The fact that the data is already in machinable form means ACRONYM can (in no particular order or priority):

1. Track the geocentric demand for specific products, which will inform them of where to place inventory and in what quantity in its many warehouses to reduce delivery time;
2. Track seasonal variations in demand to create powerful forecasting tools so that long-lead items can be ordered for inventory in a timely way;
3. Associate demographics with product orders to further analyze and anticipate demand and inform potential advertising campaigns;
4. Use reinforcement and other machine learning tools to prioritize stock keeping levels in warehouses;
5. For perishables, keep track of expiration dates to know automatically when to pull stock from inventory;
6. For potential contamination, automatically identify inventory to be removed and also to make notifications since order information typically includes email and mobile telephone information associated with particular customer orders;
7. Track delivery success rates, timelines, to detect performance variations that might require attention;
8. Keep track of long-tail ordering behaviors that might lead to concentration of long-tail inventory in central locations to be transferred either directly to customers or to distribution warehouses at need, reminding me of Inter-Library Loan systems;
9. Keep track of suppliers and to cope with supply chain disruption by deliberately exercising multiple sources where possible. Brands and brand-substitution come to mind; and 10. Place digital inventory appropriately in distribution networks to reduce data transfer requirements and latency for customer demand.

I am sure readers have already thought of a much longer list of desirable consequences of maintaining current and historic data associated with products, demand variations, seasonal and regional preferences, and other metrics that drive customer interest. One can even imagine deriving a sense of cultural dynamics and change from this data, not only to recognize fads (Hula hoops!) but also more subtle shifts taking place over longer periods of time. How and what we consume are indicators of societal preferences. Advertisements found in 19th–century National Geographic magazine offer a remarkable sketch of life in that time period. One imagines sociologists and anthropologists digging into data in the same way they might dig into an archaeological tell. Just as there are disciplines for real “digs” one can imagine disciplines for “digital digs” that guide the quality of data collection and analysis as more data is uncovered.

 

This also leads me to think about the provenance of data used for these analytical purposes. Provenance factors into data quality as do measures of data integrity. Maintaining data quality from source to final analysis seems intimately connected to the utility of the results obtained. One can imagine testing for digital data and algorithm quality by making predictions and evaluating their success rates. Certificates and digital signatures over hashes of data are obvious potential tools to ensure a kind of end-to-end integrity of information. To the extent that ACRONYM is willing to share any of its accumulated information, one can imagine it could contribute to a wide range of emergency response preparedness planning exercises as might be conducted by a federal emergency response agency, for example.

I am conscious that I may have just spent my time stating the obvious and your time reading it, but the exercise has only reinforced my belief that quality information has become a highly valuable commodity in our increasingly information rich and dependent world.

Sumber: ACM Magazine