Irish Water – A Data Architecture thought noodle [Updated]

[preamble: This is a thought noodle. It’s not a solution. It just sets out possible options for an alternative approach. I fully expect issues and wrinkles to be pointed out. ]

There has been a lot of discussion about the legality of Irish Water’s use of PPS Numbers. It is correct to say that Irish Water has a legal basis f or requesting PPS Numbers under the Social Welfare & Pensions Act 2014. The Water Services Act 2013 also gives them the power to request data from the Revenue Commissioners and the Department of Social Protection (amongst others).

So, there is a legal basis for obtaining data. However, the Data Protection Acts require that the data being processed by a Data Controller be adequate, relevant and not excessive to the purpose for which it is being obtained.  Article 8 of the EU Charter of Human Rights also requires that processing be proportionate, a point that was stressed by the CJEU in the Digital Rights Ireland Data Retention case.

<update>Also, as Fred Logue points out:

@daraghobrien@dermotcasey Great article. Also note that when you see word “necessary” in DP act this is a proportionality test

— Fred Logue (@fplogue) September 25, 2014

</update>

So… is it proportionate for Irish Water to be processing PPSNs, notwithstanding the legal basis that might exist permitting it? When working with clients designing data processes, I try to encourage avoidance of excessive processing of data by looking at whether existing functions can be repurposed to minimise the number of hands data must flow through. Thinking “lean” is important. Looking at this from a Data Architecture perspective, we must first look at the purposes. There are two.

1.  To verify entitlement to a household water credit
2. To verify and validate child water allowances.

Next, we need to see if there are any similar functions currently operating in the State that might provide either a model to replicate or a function that can be extended to deliver these objectives. 

Household Water Credit

Prior to 2012, households were entitled to claim a tax credit for domestic waste services from Revenue. Each household applied and the credit was applied as an income tax benefit. PPSN information was not shared with local authorities or private bin collectors to implement the tax credit. Policing the credit was simply a matter of using existing Revenue powers to seek information into Revenue for audit purposes. While the system was retired in 2012, old code doesn’t die, it just gets commented out. Reintroducing this mechanism for the Household Water Credit likely have been simple and cost effective as the basic structures for implementing it had already been developed and worked. They were just mothballed. Therefore: in determining the proportionality of allowing a private company access to 4 million PPS numbers, did anyone examine the feasibility of reusing an existing system that would not require data to be shared outside of an organisation that already processes PPSNs? Did anyone consider reusing/recycling this processing?

The Children’s Water Allowance

Irish Water tell us that they need to have PPS numbers of children to confirm their eligibility for a water allowance.  There is an allowance. For children. A children’s allowance if you will. A benefit for children. That must only be given to children who are in receipt of Child Benefit…. So why not just either add the allowance for water to the existing Child Benefit payment, or clone the Child Benefit processing in the DSP to deliver the Child Water Allowance? This would have avoided the need to request PPS numbers of children, a sensitive matter for many. No data would be processed outside the existing state agency that deals with Child Benefit and the PPSN data of children. <update>Another tweeter raised the question of non-resident recipients of Child Benefit.

@dermotcasey @Tupp_Ed @IrishWater @daraghobrien nice try but many children living outside Ireland in receipt of child benefit from here. — orl (@cnocdua) September 25, 2014

This does not invalidate the approach outlined above. It simply adds a business rule to the data queries necessary to run the process. When working with clients on projects this kind of thing crops up a lot.  It’s one of the many reasons why, after half a life time doing this ‘data thing’ I advocate organisations invest in PLANNING and design for data before jumping into building databases.

Dermot Casey nailed the necessary business rule in “code speak”

@cnocdua @Tupp_Ed @IrishWater @daraghobrien update API query to “where child has PPS = true && resident in ireland == true”

— Dermot Casey (@dermotcasey) September 25, 2014

Translating that for humans: “IF a child has a PPSN AND is resident in Ireland THEN assign credit ELSE don’t assign credit”.

Of course, this assumes that the DSP has a data field that identifies if the country of residence is Ireland or not (and if they don’t then I would have to ask how any statistics about how many non-resident children are in receipt of Child Benefit are calculated).

</update>

Value For Money?

Given the set up costs of Irish Water, one must ask as well whether reusing/recycling or repurposing existing systems and processes to the objective of having credits and allowances might have resulted in a net saving to the exchequer, particularly in difficult economic conditions.

I cannot answer that and would suggest that is a question the C&AG should consider asking. However, from a Data Protection perspective, it would have resulted in a zero fuss outcome – “State Agencies process data the way they always have to ensure credits and benefits are applied appropriately – SHOCK!!” is not an attention grabbing headline. A private company that is processing PPSN and other personal data but is unable to give clear answers about the nature and scope of that processing IS a headline or dozen.

The Importance of the Information Asset Life Cycle

When I teach Data Governance or Information Quality or when I engage on consulting projects, I always introduce the POSMAD lifecycle of information. POSMAD is a standard model for any asset management consisting of six steps.

• Plan
• Obtain
• Store and Share
• Maintain
• Apply
• Dispose

Part of “Plan” from a Data Protection perspective is asking “Is there a less invasive/less privacy risky way of doing this?”, and from a ‘return on investment’ perspective it requires us to assess if the way we are proposing to do something is the best. Working through this life cycle allows organisations to apply “Privacy by Design” thinking earlier in the lifecycle of the data.

It appears Irish Water jumped straight to the “Obtain” phase because they had legislation that allowed them to do it, but nobody gave consideration to the PLAN stage. This is a function of effective Data Governance in an organisation and I would hope that the Government learns a valuable lesson from this as they formulate their Data Sharing and Governance Bill over the coming months.