When the Database Becomes god

A look into the ethics and economics of databases and algorithms. How technology has outpaced the social conscience and society needs to catch-up.
Electronic databases are necessary, and there is no way that we can revert to the days of a pen and paper society, but do we trust databases too much?
Everybody uses a database every day. The question is really how much. The answer is many times more than you think. They control how our lives are structured.

  • Need to Google the closest restaurant in your area for lunch? You enter a query and Google recalls it from their database called BigTable. Google remembers your search by placing your activity in their database and uses this data for advertisers.
  • Went to the Doctor’s and the person pulled up your digital medical file? It is likely in one of the private medical programs using the inexpensive database program called MariaDB, or if they are linked with a Provincial program in Canada, it is likely through SAP, PeopleSoft, or Oracle.
  • You go to work and you key in your ID number or swipe with a card. This will funnel its way into a database program. Who pays you? Probably no one. It is automatically set-up in a database program. Supervisors and payroll officers just check to make sure all data was entered correctly.
  • You pay by debit at a grocery store? The transaction is saved in three or more databases. The first one is in the grocer’s bank database which links to you personal bank’s database and they make a handshake. Then the records are somewhat shared by the two, and the grocer also has its own personal database for sales, inventory, transactions, taxes, and more.
  • Went to Facebook to check your status? You are seeing the results built from Facebook’s proprietary TAO database. Why is it every time you click on Facebook there are advertisements that are tailored to your likings? That is because your viewing history has been saved to the database. Facebook has developed programming to target ads based on this data. This is their source for advertising revenue.
  • If you are sourcing or entering personal information in an iPhone or Android smartphone, then you are using a SQLite database.
  • Watching a movie posted on Netflix? Netflix has moved into a newer territory of database structures called NoSQL, where they use a database program called Cassandra. You may notice that whenever you browse Netflix, it remembers every movie you watched and what your preferences are. This is because your preferences are saved in their database.
  • Travelling to the United States from Canada? Your personal information is accessed from other Canadian security sources and pooled on file with the Canadian Border Services Agency (who probably use the SAP application for this but CBSA does not publicly comment about their database structure). The relevant data is automatically transferred to the US Department of Homeland Security’s Oracle database. 1 The CBSA also keeps a diary of your travel history. Canadian privacy laws require that any history older than 3.5 years must be destroyed. As to the information transferred to the U.S. database, it is not known if it is ever destroyed.
  • Need a loan? Organizations such as Equifax have harvested your information from the banking and credit sectors. Most financial institutions have agreed to share information with this organization for credit purposes. Equifax’s data is stored in an Oracle database2. Your bank or lending agency will likely check this source, and add an entry that you have applied before approving or rejecting any request.

We all take Database architecture for granted. If it wasn’t for Relational Databases or it’s evolving children, there would be no internet, no advanced computing, smartphones, or many other conveniences.
If a Government or institution had the power to bring together all this different data, it could build a complete portrait of almost every individual. This is theoretically already happening. The United States National Security Agency has a program that can theoretically do this called Prism.3 However, I am skeptical about the effectiveness because this would require massive daily manpower to filter through the results and assess any risks. It doesn’t seem feasible to track and monitor so many people.
There are many ways to view the collection of this data and its problems. This article intends to limit its scope to the morality and economics of databases.
Canadian and most Western societies view database information as morally neutral. However, this is not true. The collection of data has been a moral problem for thousands of years. For example, the 900 BC or so story of King David in the Bible who called for a tabulated count of every man available for the military. This was considered by God a serious national sin and led to a severe judgement.4 Military statistics along with statistics on almost anything are routine today and it would be hard to argue that any database or statistics are inherently evil. However, David’s judgement provides evidence that the creation of certain types of databases creates ethical problems. A contemporary example is a concern over presidential candidate Donald Trump allegedly endorsing a plan to register all American Muslims in a special database. Whether true or not, it demonstrates a tension that society has with the ethics of databases.
Databases are an expression of the human experience and are important. They naturally occur and have to exist for life to work properly. However, what happens if the database or algorithm becomes a god?
History shows us that this can happen. The Nazis, in collaboration with IBM, had a database infrastructure for the purpose of racial targeting. If they overtook a town or city, they would hire clerks to enter local citizen information. If a person or family was identified with Jewish blood, they were then considered non-human and were exterminated.5 The clerks who entered the data or the management, to my knowledge, have never been accused of abetting the holocaust but played an important part. They simply felt that they were entering data and that this was a neutral task. Nor did they have the power or the access with those who created the database to ask why, or who the data was for. They were simply to do the work, and not to question.
The database in the situation had the power to kill. This is an extreme example that rarely occurs but demonstrates there has to be checks and balances entrenched in our social psyche to ensure this never happens again.
Paul Sperry, author of the New York Times article, Obama collecting personal data for a secret race database claims that the U.S. Government is making a powerful new database that tracks ““inequalities” between minorities and whites.” It is intended to harvest information from credit cards, home loans, workplaces, neighbourhoods and tie it into a database that calculates the level of discrimination that exists. Those municipal governments that are identified “must find ways to close the gap or forfeit federal grant money and face possible lawsuits for housing discrimination.”
No one can disagree with the aims of the database or the attempt to correct a serious imbalance. However, the idea that technology, specifically an algorithm, can solve the problems of discrimination and hatred is a social engineering project that likely will fail. An algorithm or database cannot solve the problems of lack of respect or concern for others. If it could, these problems would have been erased eons ago.
Databases and algorithms are now central to trading in most stock-exchanges. Technology has introduced a new genre of trading called high frequency trading. These are algorithms created by physicists and mathematicians for trading companies accessing the electronic trade centres around the world. As Nick Baumann explains in Mother Earth, HFTs do not involve people just triggers in the stock market – usually small fractions of a cent and selling for a slightly larger sum. For example buying at $1.00 and selling at $1.0001. Thousands of transactions are done in milliseconds and this all quickly adds up. 6 The value of company stock is irrelevant. It is the variation that the algorithm is looking for. Baumann surmises:

As technology has ushered in a brave new world on Wall Street, the nation’s watchdogs remain behind the curve, unable to effectively monitor, much less regulate, today’s markets. As in 2008, when regulators only seemed to realize after the fact the threat posed by the toxic stew of securitization, the financial whiz kids are again one step—or leap—ahead. 7

Things have changed and the Government has caught up. Matthew Philips reported in Bloomberg Business how the HFT algorithm started to plunge. Too much competition, increased charges by the stock markets themselves for speedier access, Government intervention, along with a sudden shift in market realities caused this genre to substantially decrease. The Security Exchange Commission introduced a program called Midas to identify and correct the abuses happening in the almost unregulated wild west of HFT trading.8.
The HFT problem shows that the problem of databases and algorithms is that they cannot exist independently without some external source of auditing or validation.
The power of a database caused serious questions during 2015 Canadian elections. Colin Bennett described in his article featured on iPolitics.ca website on how the three main Canadian political parties have been tracking individual Canadians political preferences. He thinks this process treads an ethical minefield.

By the people working in and reporting on Canadian federal politics, these systems are now seen as indispensible tools for the modern “data-driven” campaign. And the parties will argue that these systems help them engage and mobilize their supporters, enhancing the democratic process.
… The electorate is then profiled and scored. For instance, we know that CIMS ranks voters on a scale of plus 15 (right) to minus 15 (left). These profiles are then used to allow the party to allocate its resources more efficiently for its canvassing and get-out-the-vote operations.
The systems will also have information on a voter’s preferred contact methods. If someone does not want to be contacted, it should be recorded in the party databases.
…What they don’t talk about is how far out of mainstream democratic practice these databases really are. In most other democracies, they would be illegal.

Bennett strongly reinforces this thought with:

You have no legal right to learn what information a party database has collected about you, to remove yourself from a party database, or to restrict the collection, use and disclosure of your personal information. And for the most part, parties have no legal obligations to keep that information secure.9

The demise of Target stores in Canada can be traced to its database system which happened to be SAP. SAP shouldn’t be blamed here. This same scenario could have played out in any other competitor. SAP is a world leader in enterprise software with approximately “300,000 customers in 190 countries.”10 Target experimented by using SAP for the rollout in Canada. It was to be the starting point for a universal migration to SAP within all its US portals. However, an underlying breakdown happened that Target did not anticipate. Jose Castaldo, wrote a detailed report on its brutal death by design in Canada Business:

It didn’t take long for Target to figure out the underlying cause of the breakdown: The data contained within the company’s supply chain software, which governs the movement of inventory, was riddled with flaws.11

In short, Target could not properly move merchandise into stores and lost buyer confidence by transportation, customs and packing problems, inconsistently stocking, not having stock, and improper product placement because the data-entry into the database was not properly done. Labour was also being diverted from sales to clerical entry and validation, and those that did the data entry were not qualified to do such. They made too many errors.
Databases are to reflect a set of realities. Sometimes real world realities are not the same as outcomes of the data. It depends on how the data is filtered. This was recently shown in the child care problems in Manitoba. The Province of Manitoba had a reputation of having the highest number of children under Provincial care among all the other Provinces in Canada. What did the Manitoba Government do? They changed the database semantics on who qualified for those statistics. The change disqualified a percentage from being counted and, therefore, reduced the amount needed to be stated from their database. Consequently, Manitoba no longer negatively stands out among its counterparts.12 The moral here is that databases can be manipulated to state realities that do not exist.
Another problem of database driven companies is accountability and flexibility. Employees or middle managers cannot question or recommend changes to the business model because it may require significant and costly changes to the database system and the corporate culture. Nor are supervisors or managers in the position to speak to the small group of stakeholders who have the power to approve the changes for software engineers to reinterpret and implement.
Most changes are rudimentary and small, but a few can come across that are clearly wrong. In these situations, the middle manager or supervisor must support and enforce the strictures upon employees. By refusing, they can lose out on promotions, get demoted, or even fired.
This can lead to an amoral workforce who do not care nor agree with the objectives of the company. Rather, they see their jobs as ones that are to fulfill the data entry and validating of the database and nothing more. Upper management sees employees not as people, but as figures that either continue to fulfill the database or detract from targets. Supervisors and managers themselves also are amoral. Their role is to enforce the rules not make them and use whatever means to get the workforce to comply.
Many firms have programmed their database programs to obscure questionable practices. Cellphone carriers are especially featured in this category. Today’s cellphones are nothing more than a wireless computer that transmits data for a short distance to a physical network. The data then is served through the network and connected to the appropriate end-point. Any service whether voice, text, message, chat, surfing etc., is pure data transmitted in the TCP/IP protocol. The TCP/IP protocol is absolutely no different from what is accomplished on a home computer – a device that freely connects with other devices throughout the world without any charge for each attempted connection. However, the cellphone companies have distinguished the TCP/IP protocol into different payment categories, phone, text, call display, downloads, usage, etc., by each attempted connection. This is all monitored and stored in a database. The customer then is charged for these different usages of TCP/IP. Why the distinction between TCP/IP on a home computer and a cellphone? There shouldn’t be. It is the same data being transmitted by a similar device. It is because consumers are historically accustomed to phones and internet being completely different entities. The reality is, there is no difference on the technical side. The two have merged and have been for years. There is no reason to continue charge a per connection fee or impose a long distance charge. These are legacy words of technologies and infrastructures that have long ceased.
Another problem of the current generation is that of mercy. Databases do not forgive. One inappropriate Facebook post could potentially cost you your job, future occupation, maybe a relationship, and worse yet, even if it is removed, the posting can perpetually linger somewhere else on the internet. Your error is permanently marked and may never be forgotten. It is hard to overcome.
What can Canadians do to ensure that databases do not play god? There has to be a counter database made by a third party or Government that monitors and applies ethical standards to any database that attempts to socially engineer, economically take advantage of a target group, or govern a workplace.

  1. http://www.intel.pl/content/dam/www/public/us/en/documents/white-papers/performance-oracle-dept-homeland-security-paper.pdf
  2. https://blogs.oracle.com/cx/entry/customer_success_equifaxs_orac
  3. https://en.wikipedia.org/wiki/PRISM_(surveillance_program)
  4. I Chronicles 21:1-16
  5. Edwin Black. IBM and the Holocaust: The Strategic Alliance Between Nazi Germany and America’s Most Powerful Corporation-Expanded Edition. US: Crown Publishers. 2001
  6. http://www.motherjones.com/politics/2013/02/high-frequency-trading-danger-risk-wall-street. Too Fast to Fail: How High-Speed Trading Fuels Wall Street Disasters, by Nick Baumann. Mother Jones. January/February 2013.
  7. http://www.motherjones.com/politics/2013/02/high-frequency-trading-danger-risk-wall-street. Too Fast to Fail: How High-Speed Trading Fuels Wall Street Disasters, by Nick Baumann. Mother Jones. January/February 2013.
  8. http://www.bloomberg.com/bw/articles/2013-06-06/how-the-robots-lost-high-frequency-tradings-rise-and-fall#p3 June 6, 2013
  9. http://ipolitics.ca/2015/09/01/theyre-spying-on-you-how-party-databases-put-your-privacy-at-risk/
  10. http://www.sap.com/corporate-en/about/our-company/index.html
  11. http://www.canadianbusiness.com/the-last-days-of-target-canada/
  12. http://www.cbc.ca/news/canada/manitoba/children-care-manitoba-numbers-1.3441611

Leave a Comment