When looking at ways to achieve a strategic outcome, Business Analysts and Product Owners need to understand the current state of the organisation and its associated markets. By interviewing internal and external stakeholders (including customers) and analysing internal and external data, opportunities to achieve the outcome will be discovered

On 27 October 2024, at a Trump Campaign Rally at Madison Square Garden in New York, Elon Musk, leader of the US Department for Government Efficiency (DOGE), stated that he believed DOGE could reduce federal government spending by $2 trillion by reducing waste, abolishing redundant agencies, and downsizing the federal workforce. (This figure keeps changing as I type this article).

On the day Donald Trump became President of the United States, he signed an Executive Order Establishing DOGE to allow Elon and his team to start analysing the federal government structure and finances.

This led to a press conference in the Oval Office on 11th February 2025 where a happy Elon Musk announced he had uncovered the ‘biggest fraud in history’ as he announced that he had found 20 million people in the Social Security database who were over 100 years old and had the death field set to ‘False’.

I immediately thought there was something not quite right with the statement. It is unbelievable that the United States Social Security Department would not have financial controls to stop such a fraud.

I then smiled to myself, remembering when I had analysed Insurance Policy databases for insurance companies in the United Kingdom and the United States and found some very puzzling dates.

For instance, I found live life insurance policies with birth dates in the early 19th Century. I also found live general insurance policies that commenced in the 19th Century, particularly Marine insurance.

Policy, Customer and Beneficiary databases tend to be ‘Master’ databases. Think of them as ledgers containing every customer, supplier or sales an organisation has ever made.

For institutions more than 75 years old, these master files were kept in paper ledgers. When computers were commercially introduced in the middle of the 20th Century, lots of data had to be transferred from paper records to create the first digital version of any master database. This presented the first opportunity for digital errors to be introduced by:

• Manual Processing. The initial migration was manual, so there was the potential for mis-keyed data to enter the database during the take-on process. – That is probably why a 369-year-old was uncovered by the DOGE team.
• Processing Principles. There were also parameters about what to include from the ledgers, which may have been all records, not just live cases. – When this migration was performed in the 50’s, 60’s and 70’s, having people with birthdates back to the mid 19th century would have been common, so the fact they are on the social security database is not that strange
• Extra Functionality. In all probability, to provide greater digital functionality, additional fields would need to be completed, and in some cases, these fields would be set to a reference point. – Elon was queried why so many people seem to be aged 150 in the social security database of 20th May 1875.

Since the first version of the master database was created, it is more than likely that the database has evolved, and several transformations of the data could have occurred, introducing further digital errors as follows:

• New functionality, New Software, New database structure. Between the first manual migration and today, master databases have evolved to take advantage of new functionality, new software, or improved database systems. Given the timescales between starting the analysis and Elons’s pronouncement, it is unlikely that DOGE would know the reasons behind some of the data values, as they would be lost in the mists of time.
• Regulatory change. At a minimum, such master databases need to comply with regulatory changes as they occur. However, the information required to meet the regulation often does not exist for historical records, so workarounds, such as default dates or codification of records to say they were pre-regulation, are developed. Again, it is unlikely that DOGE would know the reasons behind some of the codified data.
• Mergers and acquisitions. As time moves, organisations often merge systems and databases to create one overall system. The migration process to create a single database can involve changing data values or creation of additional fields to identify the original data source. The documentation of what was done and when can be forgotten over many years. – Now my American History is not that great, but it may be the case that each state had its own social security services before the federal one was implemented. If so, there may be a few differences between the data of each state, which could lead to a mix of data in the social security database.

Even in systems developed and used today, can cause date issues. There are now many different channels through which you can contact or register with an organisation (Web, App, Face-to-Face, Phone) and I am sure you have noticed the functionality available through each channel can differ.

• Different channels using different software. A mix of software can lead to data from different systems residing in a master database (such as payment codes). – Again, a lack of understanding within DOGE as to what systems are generating which data values could lead to misinterpretation of the data.
• Global Systems. Due to globalisation, it is necessary to understand the underlying data correctly. For instance, is 06/03/2025 March 6th or June 3rd? When calculations are performed on a date containing global customer or beneficiary data, are you sure systems are using the correct date formats to determine age? I don’t know if the Social Security database has always used the MM-DD-YYYY format, and I suspect DOGE analysts don’t either.
• Dates. Believe it or not, due to the high cost of technology, use of storage and memory was a prime consideration in the early days of computing systems. I was once in a meeting where there was a discussion about how many Megabytes of storage could be saved if the century portion of a date was not stored. Therefore, it is possible that the wrong century could be added programmatically during a database upgrade. – I am sure the DOGE team would not have even considered this, given the amount of analysis time they had before the announcement was made.

Given the life expectancy in the United States is so good according to DOGE. I briefly considered whether to live the rest of my life there. However, once I reviewed all the possible factors that could have undermined the statement ‘the biggest fraud in history’ I decided to remain in the United Kingdom.

I then performed further investigation on the subject and found out the Social Security Administration Inspector General highlighted this issue during audits in 2015 and 2023. When asked to fix the issue, the Social Security Agency estimated it would cost around $9 million to investigate and correctly update each record and decided not to pursue it because there was little or no added value to the service.

I also discovered that the 2023 Audit checked on payments made, which showed that only 44,000 people aged 100 or over received social security benefits and of those only 13 were over 112.

So the Elon and DOGE assertion that they had uncovered the ‘biggest fraud in history’ is ‘False’ as a full analysis of the data, its origin, evolution and its relationship with other systems was not undertaken.

This lack of context should now be fully explained to President Trump and the American people. I will not be holding my breath as the administration was able to use the incomplete data to undermine the Department of Social Security and get the headlines they wanted

According to Google today the ‘biggest fraud in history’ remains Bernie Madoff’s Ponzi scheme which cost investors an estimated $64.8 billion.

Although an interesting and topical story, I wrote this piece to provide learning points for Business Analysts and Product Owners

• Analyse data thoroughly and from as many different angles as you can. I expect the DOGE team calculated the age of each person on the database based on the date of birth field and then checked the value of the Death field before assuming that was all that was required to receive payments. No cross validation with other data sources, such as the payments database, was made.
• Check the Provenance and history of any changes to the database. If the source of the data cannot be verified or historical change logs investigated, confidence in the data should lessen.
• Compare your findings with other similar databases. If you can review your findings against similar databases, say the Tax or Census data master lists, there should be some consistency to increase confidence in the data.
• Try not to give too much away until you are ready. Giving early insights before you complete your analysis can harm the project in several ways. It can cause unhelpful reactions, and cause a lot of additional meetings, the findings can be challenged, diverting you from the analysis you need to complete to do a good job and steering committees can decide to develop something that will not provide the improvements that they want.

Finally, to ensure the business or technical change you are about to embark on is the right thing to be working on, take some time to get comfortable with the data and its context, look for the things I have suggested, before starting any development.

I would love to understand if other factors can lead to misleading data. I am sure my Data Scientist friends have lots of stories to share…..