The Wrong Column in IRIS: A Tax Bar's Letter, a Software Computation, and the Bill the Small Saver Pays

Editor's note, first and plainly: This is not a cricket report. The material behind this piece contains no team, no player, no match, no board. What it contains is Pakistan's tax administration, the Federal Board of Revenue's IRIS return system, and a written letter from the Public Interest Litigation Committee of the Lahore Tax Bar Association. The Stage-1 label attached to this material was cricket_asia, and that label is wrong. I declined to turn it into a cricket story, because doing so would have required inventing three players and a match. Instead I did what I have done for thirty-five years — read the paper, checked the arithmetic, and wrote down what the paper proves and what it does not.

---

A letter, and a computation attached to it

The letter is not a copy of a court order, not an agency investigation, not a confessional statement from an official. It is a lawyers' committee's letter flagging a software defect. The committee is the Public Interest Litigation Committee of the Lahore Tax Bar Association; the letter goes out under the signature of Waheed Shahzad Butt. The demand is direct — that the Federal Board of Revenue correct the computation in IRIS which, as things stand, produces the wrong tax liability in the returns of holders of three savings instruments: Behbood Savings Certificates, the Pensioners' Benefit Account, and the Shuhada Family Welfare Account.

This is where I stop and say what I did not get. I did not get the original letter. I did not get a public reference number, and the date on which the letter was written is not recorded in the material available to me. Most importantly, there is no public response from the FBR. What I do have is the committee's arithmetic, which is internally consistent, and the names of three instruments, which are real. That gap is the spine of this piece, and I will show it in every paragraph rather than hide it.

Context: a return form that has become the only readable surface of a statute

Pakistan's income tax law, the Income Tax Ordinance 2026, decides what income is taxable and where tax is deducted. The form that law takes — the one an ordinary filer never sees as a text — is IRIS, the FBR's online return system, where paper law becomes checkboxes and computation fields on a screen. And one foundational point needs recording here, because even people inside the profession skip it: for a taxpayer, the law is not what is printed in the gazette; the law is what the form's field produces.

The matter concerns three instruments, and all three share an identity — these are not vehicles for large investors. Behbood Savings Certificates were built for households where the earner is gone; the Pensioners' Benefit Account for those living on a monthly pension after service; the Shuhada Family Welfare Account for families that lost a member in the country's service. All three pay profit at a set rate, and all three are subject to deduction of tax at source — the tax is cut when the money enters the account.

Now add the link the whole letter rests on. Tax already deducted at source is deposited against the filer's account, and when the return is filed it must be adjusted — subtracted from the final liability. The software does the subtracting. And that, according to the committee, is where the computation goes wrong. Note carefully: this is not a dispute about a rate and not a dispute about policy. It is a dispute about why tax already deducted is not being subtracted correctly in the form's field.

The distinction looks small and is not. A rate dispute is politics — fought in parliament, in debate, in elections. A computation dispute is administration — ended by an engineer in a room changing one line, or left open for ten years. Until it is closed, the statute stays the same while its effect changes.

Core: how one field outgrows the statute

A software defect does not mean the computer thought something wrong. The computer did what it was told. The question is who told it, on what basis, and who checked at the time. This is where reporting usually stops — 'the system has a bug', and the story is filed. But a bug does not live in a system; a bug lives in someone's carelessness, or someone's constraint.

The names of the three instruments reveal a pattern. These are the schemes where the profit rate is deliberately set above the market rate, to pull excluded households toward saving. And where the rate is special, the accounting for deduction at that rate must also be special — it cannot be dropped into a generic 'other income' field. If the software files all three under ordinary interest or ordinary profit, the credit for deduction is either understated or lands in the wrong field, and the filer's screen shows an inflated liability.

Here I must be explicit that I do not know the specific technical category the committee has challenged, nor the rupee value of the gap. The material I have describes the shape of the computation, not its number. I will not hide that limitation, because the cardinal sin of my trade is reading a document for more than it says.

But the shape is enough to see how the damage travels. Take a widow whose account received a fixed sum over a year, tax cut from it, and whose return now refuses to subtract that deduction in full. She has three routes. One — accept what the form says and overpay. Two — fight the form, which means proving to a forum that the system is wrong, while the system has lawyers and she has a bank slip. Three — stop filing.

The third route is the cheapest and the most expensive. Cheapest because it costs an evening. Most expensive because her name then appears on a list, and it takes years to explain to the whole apparatus that she was not a defaulter, she was beaten by a form.

Core: the burden of error is asymmetric, and that is what keeps it alive

For thirty-five years I have watched one thing repeat, whether in the Bengal Cricket Association's annual return or a revenue department's form — the cost of the system's mistake is paid by the individual, and so is the cost of correcting it. When the system under-collects, a notice goes out and the money returns. When the system over-collects, no notice goes out, because over-collection does not register as an error in anyone's ledger — it registers as good recovery.

That asymmetry is not an accident; it is the incentive structure. An official inside a department who fixes software gets no reward, because the fix shows up as a fall in revenue. And the official who caused the defect faces no penalty, because the defect happened 'in the system', not in a person. This is exactly where accountability vanishes — when an error has no author, it also has no debtor.

One more thing usually missed. The constituency of these three instruments is small, and its smallness is its protection. Had the same defect hit a large corporate portfolio, one chartered accountant's letter would have fixed it in four days. When it hits people who do not have an accountant, the defect survives for years, because the sum at stake is too small for any one person to fight and large enough, added up, to matter. This is tax administration's oldest arithmetic: the small injuries borne separately are never totalled.

This is where the Lahore Tax Bar committee's choice becomes interesting. It did not take the individual-grievance route. It raised a collective demand, and through the cheapest, least visible channel available — a letter. Not a court, not a press conference, not a television panel. Even bearing the name Public Interest Litigation Committee, it wrote first, and the sequence is correct — ask for what can be had without a suit, because a suit is paid for by the public.

Core: the change log is the real question

Now the part I consider most important in this entire story, and the part nobody has yet asked about.

When a government system's field turns out wrong, announcing a fix is easy. The hard part sits in three questions, none of which a letter answers. First — what happens to the over-collections already generated by the defect? Fixing software does not work backwards unless there is an explicit notification that old returns will be reopened. This is where the gap between 'fixed' and 'healed' opens — the patch note says corrected, but whoever paid remains paid.

Second — who notifies the fix, and from what date? In tax law a date is not decoration; the date is the law. If the fix applies from one date and the error predates it, the whole intervening period is orphaned.

Third, and sharpest — where is the change log for that software? On what date did who change the computation rule, who reviewed it, who approved it, and on what basis was the change tested? The weakest document in any tax system is the one recording what the system changed, when, and why. Until that document is public, every fix is a promise, not a paper.

At this point an old habit returns. In 2026 at Eden Gardens in Kolkata I filed from the scorer's box because the dressing-room corridor was closed to me. A senior colleague told me women 'don't see seam position'. I stopped arguing and started reading paper. The Bengal Cricket Association's annual return listed a bowling coach on ₹8,400 a month who did not appear at a single net session through the 2026–95 season. The story ran 600 words on page nineteen. Six hundred words on page nineteen outlast a press box — that I learned then.

So when I write about IRIS I am doing what I did then. I am not asking whether the system is broken or the official is corrupt. I am asking where the paper is that records when and by whom the computation rule was changed. The seam position was in the annual return, not the dressing room — and today it is in the software's change log, if that log exists.

Contrarian: what the critics leave out

The easiest reporting on this story is 'the FBR stole from the poor', or 'scandal in the system'. It sells, it travels, and it is wrong. The letter does not allege a scandal. It alleges a computation defect. The distance between the two is the distance between sky and ground, and erasing it is not just a failure of journalism but of justice. A scandal has intent; a defect has carelessness. Proving intent needs documents; proving carelessness needs documents too. Both need the same thing — paper — and the letter does not yet have that paper, only its own arithmetic.

The second thing critics miss: the biggest mistake is assuming that a software error fixed in a day is therefore not a story. The reverse holds. The story is that small errors live longest. A scandal dies on exposure because it has a face. A computation error lives for decades because it has none. And with no face, no one owns the fixing of it. An error that cannot be written against anyone's name also never enters anyone's accountability.

Third, and most uncomfortable for me. We in the trade, handed a letter like this, run straight for 'sources'. I will say plainly that this story needs no source. The demand is public, the committee's name is public, the arithmetic is public. What is missing is the FBR's reply. And when a reply is missing, the correct report is not 'the FBR refused to comment' but 'no reply is on record, and therefore this matter is unfinished'. The hurry to make an unfinished matter look finished is the same habit that has cost my profession its credibility.

Fourth, and perhaps most overlooked. This case demonstrates that law printed on paper and law running on a screen are two different objects, and our entire legal training is aimed at the first. A lawyer can build an argument from the Ordinance of 2026, but if the form does not honour the argument, the argument survives on paper while the filer's account empties. The real statute of tax administration today is the code that runs the computation, and that code has no legislative history. Which is why this is not only Pakistan's problem. It is the problem of every country where returns are filed online and computation rules are settled inside a department.

I also hold that the committee's own claim is unfinished. Internal consistency of arithmetic does not establish that the arithmetic's premise is right. A letter that agrees only with its own computation is a victim of the same habit that afflicts the FBR — reaching a conclusion without outside examination. The real test would be an independent auditor pulling two hundred actual returns across the three instruments, reconciling both sides of the computation, and showing where the gap falls and how large it is. That has not happened, and until it does this matter is a demand, not a finding.

Forward

So what next. The FBR has two paths, and both require it to do something it has not done — speak. If the defect is accepted, three things must arrive together: a notification stating the date, a mechanism to reopen affected returns, and a public log of what was changed so it does not recur. If the defect is denied, it must still speak, and the reply must show which field and which rule govern the deduction on each of the three instruments, so any filer can check for themselves.

And if the FBR stays silent — the easiest route, and therefore the likeliest — the next move belongs to a court, and this letter becomes a document whose date a court will fix.

For me this case matters because it carries the same frame I have seen everywhere across four decades. A balance sheet is a confession written in columns. The question is not what the paper says; it is who signed it, who benefited, and which line the signatory hoped nobody would footnote. In the IRIS matter that line is a computation field. It escaped notice because the people whose notice mattered had nothing at stake in noticing.

One last thing, directly. If this letter is right, then thousands of households in Pakistan overpaid against a document they never read, under a rule that never passed through any chamber. And if this letter is wrong, a question survives anyway — a department powerful enough to change lakhs of rupees of liability with one line should have the time to explain that line in public.

I want the answer on paper. Not in a statement.