Proof Beats Promises For thirty years, tax software has run one calculation and asked us to trust it. The firms that win the next decade won't be the fastest — they'll be the ones that can prove a number instead of just asserting it.

There's a habit at the center of our profession so old that we've stopped seeing it. For decades, tax software has run a single calculation, printed a number, and asked us to trust it. One engine. One answer. No second opinion. And we've signed our names underneath, season after season, because that was simply how the work was done.

Step outside tax for a moment and notice how strange that is. In every other field where a number really matters, the important one gets checked by a second, independent method. Engineers don't ship a load calculation on one model's say-so. Accountants reconcile. Pilots cross-check instruments. The whole idea of a high-stakes number is that you verify it a second way before you rely on it. In tax — where the number determines what a family owes the government, under penalty of the preparer's signature — we've mostly run it once and moved on.

Not because we're careless. Because checking by hand was never worth the hours. Re-deriving a return a second, independent way used to mean doubling the work, and no practice can afford to prepare every return twice. So "run it once and trust the software" wasn't a lapse in judgment. It was the only economically sane option available. The tooling made verification a luxury, so verification quietly disappeared from the standard of care.

That constraint is the thing that's finally lifting.

What changes when a second number is cheap

The reason verification was rare is that it was expensive. When a second, independent calculation becomes something a machine can run in the background — automatically, on the same inputs, every time — the economics flip. Suddenly the second opinion isn't a luxury you buy for the complicated returns. It's a property of every return, running whether you asked for it or not.

And here's the part that surprised me about how it feels in practice: the most reassuring thing on the return stops being the refund. It becomes a difference of zero — the gap between two calculations that were run separately and landed in the same place. When that difference is zero, you haven't trusted the software. You've watched two independent paths arrive at the same answer. That's a fundamentally different kind of confidence than "the program said so," and once you've worked with it, going back to a single unchecked engine feels like flying with one instrument.

This is what I've started thinking of as the verification economy: the moment when proving a number becomes cheap enough to do on everything, and "I checked it two ways" stops being something you say about the hard returns and starts being how the whole practice runs.

Proof compounds; speed doesn't

Almost every tool in our world right now is racing to be faster. Faster extraction, faster data entry, a faster path from document to draft. That's fine — speed is genuinely useful. But speed is also the first thing that gets copied. Within a season or two, everyone's tool is fast. Fast becomes table stakes, and table stakes don't differentiate anybody.

What doesn't commoditize is proof. A firm that can say exactly how a number was checked, hand an examiner a record instead of a reassurance, and show its work on demand has something a faster engine can't replicate by shipping a quicker model. Promises are what you offer when you can't show your work. Proof is what you offer when you can. And in a profession where your signature is on the line and professional-responsibility scrutiny is increasing rather than easing, the ability to prove is not a nicety. It's the asset.

There's a client-facing version of this too. "Trust me, the software's good" is a weak thing to say across a desk. "Every number on your return was checked two independent ways before I signed it" is a strong one. The second sentence builds a kind of confidence that no amount of speed can buy, because it's about rigor, not convenience.

The practice worth building

None of this is about removing the preparer. It's the opposite. Verification is what protects the judgment and the signature that only a licensed human can give. The machine's job is to make the second check cheap and constant; your job is still to exercise the judgment, catch what the numbers can't tell you, and stand behind the return. What changes is that you get to do that on a foundation you can prove, instead of one you're asked to trust.

So the question I'd leave you with is the one I've been sitting with myself. When the next few years sort our profession, they won't sort it by whose software was fastest. They'll sort it by which firms can prove their work and which ones can only promise it. The good news is that the tools to be in the first group are finally becoming affordable enough to use on every return, not just the hard ones.

Build the practice that can prove it. That's the one that compounds.