Clarity.pk

Corruption Sniffing Bots

corruption-sniffing-bots

People in authority ought to make decisions for their country every now then, of which some are in the best interests of the nation while others are not. Only when there is a clear proof/indication of malaise (resulting in some financial benefit ultimately) should an anti-corruption inquiry be initiated. If the decision-maker (or his family) has not financially benefited from a bad decision then it may be treated as a bonafide mistake and hence, all such mistakes should be treated differently. For example, there might be a competence issue with that individual or he/she might not have complete or correct data/information.

Being one of the most commonly stated national dilemmas since 1947, corruption has forever stained the grounds of Pakistan. And from what we have heard about it, it certainly has had colossal implications, bringing us to our very first question; what really is ‘Corruption’?

Among several ways available, corruption is best defined by a simple formula: Family Wealth and Earnings (E) < Lifestyle (L)

As self-explanatory as the formula is, if a person’s standard of living and lifestyle exceeds the visible sources of his/her income (wealth and earnings), there is a very high probability of corruption involved. The logic is based on the fact that almost nothing in the world is free. Hence, for one to maintain a certain lifestyle, one requires money. If the evident finances of a person are not in line with his/her privileges, there is hardly any room for suspicion left.

WHY IS CORRUPTION A CRIME?

Before criticizing an act as illegal, it is imperative to understand the factors contributing to its illegality. As a basic question, if corruption assists people to improve their lifestyles, why is it bad at all?

While corruption appears to be an appealing, short -cut route to profound wells of wealth, it erodes public trust in authorities and the rule of law, by unfairly concentrating wealth in the hands of a few people. While the weak and the poor are mercilessly pressed under tax and inflationary pressures, the powerful are effortlessly able to enjoy the blood sucked out of the former. Ironically, these people are the ones influential enough to eliminate maximum traces of corruption yet blinded by personal gains. Every now and then, we witness new schemes and innovative ideas devised by these minds to clench as much money as possible, overriding all checks as enforcers look the other way. What’s even more tragic is the pervasive insensitivity towards corruption, worsening the situation as a whole.

SUSTAINABLE DEVELOPMENT & CORRUPTION

Jose Ugaz once said, “With corruption, there is no sustainable development.” It is, therefore, extremely important to use appropriate indicators for identifying its presence, two of which are as follows:

Indicator 16.5.1: The proportion of persons who had at least one contact with a public official and who paid a bribe to a public official, or were asked for a bribe by those public officials, during the previous 12 months.

Indicator 16.5.2: The proportion of businesses that had at least one contact with a public official and that paid a bribe to a public official or was asked for a bribe by those public officials during the previous 12 months.

Besides these indicators, in the midst of such a malicious and disturbing national state of Pakistan, what is that one valuable tool that can be used to diminish the prevailing adversity? Well, we believe technology.

TECHNOLOGY & CORRUPTION

Technology can be leveraged to complement the existing framework to seriously combat the menace. For starters, we can consider creating a central repository of individual financial profiles for all government officers and their dependents. This can be achieved through the development and/or integration of the following systems:

System 1: Payroll data of all government employees currently stored in SAP-based ‘Project to Improve Financial Reporting & Auditing’ (PIFRA) system should be comprehensive, correct, and CNIC-based as opposed to PIFRA account number. There shouldn’t be more than one PIFRA account number assigned against a unique CNIC. Additionally, any source of household income other than government salary should be explicitly mentioned. In this way, the government would keep the record of all known sources of income of each and every government employee.

System 2: CNIC and/or CRC (B-Form) based data of parents and all dependents of each and every government employee should be maintained, after cross verification from NADRA.

System 3: Complete service history of every individual government employee based on his verified CNIC, should be maintained. All orders related to transfers and postings should be generated from this central system. Data on take-home monthly salary should be maintained.

System 4: Filing of income tax returns should be mandatory for all government employees. Assets & Wealth Statement from the Federal Board of Revenue (FBR) should be pulled every year and automatically compared with the one submitted the previous year.

System 5: Data from State Bank’s Financial Monitoring Unit (FMU) should be pulled every month to check the account balance of all bank accounts of every government employee and his/her dependents.

System 6: Data from Land Revenue Management Information System (LRMIS) should be pulled every month to check the sale and/or purchase of property against the CNIC of a government employee or any of his/her dependents.

System 7: Data from the Tenants Registration System should be pulled every month to check if any government employee or his/her dependent is showing up as the owner of a rented property.

System 8: Data from Excise & Taxation department on ownership of vehicles or any other property should be pulled every month to check if any government employee or his/her dependent is showing up as the owner of any vehicle or property.

System 9: Travel history data from the Federal Investigation Agency (FIA) should be pulled every month to check if any government employee and/or his/her dependent has been traveling in and out of the country.

System 10: Data of credit card payment history should be pulled every month to check if any government employee and/or his/her dependent has paid his/her credit card bills through cash or any other source of payment which is not matching with the financial transaction history of known bank accounts.
System 11: Place of residence of all government employees should be digitally maintained, along with information on who pays the utility bills there and his/her relationship with the government employee living there. For cross-comparison purposes, utility bills’ data of all such residences should be maintained after pulling the same every month from concerned service providers.

System 12: Introduction of a hugely incentivized and/or reward-based ‘whistle-blower’ program within government offices should be introduced to identify/uncover acts of collusive corruption or any illegal cash transactions. The program would be backed by a robust tracking and registry system to identify and tag possible culprits and associates.

ARTIFICIALLY INTELLIGENT (AI) BOTS

With a technology-oriented program designed to ensure the system to system interaction, reports clearly identifying individuals presumed to be living beyond declared sources of income, would be stress-free to generate. But fundamental to the success of this program is the safety wall built around it to prevent abusive intervention by people who have access to it. In this regard, it is proposed that aggregate data be stored in an encrypted format only readable through Artificially Intelligent (AI) Bots. To further ensure privacy, these AI Bots should be programmed to vet this integrated (yet encrypted) information, in order to spot and report anomalies every month, to respective government employees. At least, a couple of weeks should be given to every such employee to submit back his/her response to the AI Bot. The system should also be flexible to accommodate digital correction of previously submitted information and/or addition of new information, within the allotted time-frame.

After the expiry of the allotted time, cases still on the data-anomaly radar may be sent to a notified scrutiny committee. This committee may call all such officials to explain the reason(s) for the reported aberration and decide the future course of action.

STEP BY STEP GUIDE

1st Step: Pseudo-Anonymize each available dataset and push it into a central data repository such that family wealth + income can later be compared with expenses (that depicts lifestyle).

2nd Step: Use multiple clustering algorithms to group anonymous individual cases such that each cluster is internally homogeneous and externally heterogeneous.

3rd Step: Committee of honest and wise people marks each cluster as possibly corrupt or clean.

4th Step: 80% of the Human Intelligence based findings finalized in Step 3 are fed into multiple (AI based) decision-tree algorithms called Bots so that each Bot develops its own understanding of what is corruption and what is not.

5th Step: Bots are given the rest 20% data which was unused in Step 4 and asked to predict. Bots then get feedback on how accurate their prediction has been and based on this they upgrade their understanding.

6th Step: For all possibly corrupt cases, clarification is sought electronically.

7th Step: Case to Case basis replies (ensuring an individual’s identity is not revealed) get evaluated by a Committee of honest & wise individuals, as to whether an Inquiry may or may not be initiated.

8th Step: Formal Anti-corruption proceedings get initiated against each referred case and their conclusions are fed back to the Bots to further improve their understanding of corruption.

9th Step: As new data continues to pour in, Bots automatically sniff out the possibly corrupt case and Step 6 gets re-initiated henceforth.

AVAILABLE DATABASES: FAMILY WEALTH + EARNINGS (E)

  1. Payroll Data from the Project to Improve Financial Reporting and Auditing (PIFRA)
  2. Family Tree from the National Database and Registration Authority (NADRA)
  3. Assets and Wealth Statement from the Federal Board of Revenue (FBR)
  4. Bank Account Credit Info from the State Bank’s Financial Monitoring Unit (FMU)

AVAILABLE DATABASES: LIFESTYLE (L)

  1. Utility Bills from Electricity, phone, internet, gas companies, etc.
  2. Movable and Immovable the Property from Land Revenue Management Information System (LRMIS)
  3. Vehicles Owned from the Excise and Taxation Department
  4. Bank Account Debit Information from the State Bank’s Financial Monitoring Unit (FMU)
  5. Credit Card Spending Information from the State Bank’s Financial Monitoring Unit (FMU)
  6. Residential Information from the Tenants Registration System
  7. e-Stamp Paper Information from the Revenue Authority

EXPLOITATION OF PRIVACY

It is not unusual to find people ask questions about the extent to which this system can exploit an individual’s privacy. “Isn’t it too Orwellian”, they ask. “Wouldn’t the senior civil servants be subjugated with such detailed invasive information available about their personal spending?” Truth is, every civil servant, in every country, is bound by law to declare his/her income and wealth statement every year. As perplexing as it may sound, the reason lies at the root of it. Such a system is necessary to ensure a venality free functioning of government bodies. Without a solid base for checks and accountability, the government would become nothing but an accumulation of national funds drawn from the majority to allow for the superfluous privileges of a few.