Why Union Benefits Still Run on Software From the 1990s

There is a category of software nobody writes about because nobody outside it knows it exists. Taft-Hartley benefits administration is one.

I found it the way most people find these things, by accident. Back in the early 2010s, third-party administrators started hiring me to build the pieces their benefits software could not do. Not to replace it. To patch around it. A report the system would not produce. An import the vendor did not support. A workflow that existed in a spreadsheet because the platform had no concept of it.

I did that work for years, and the whole time I was looking at the same thing from the outside: an industry running critical infrastructure on software that had stopped improving, where every gap I got paid to fill was a gap the vendor had no reason to close. It stayed with me. Simplorium is what happens when I finally stopped patching around it.

Here is the shape of it. A multi-employer benefit fund is jointly managed by a union and participating employers. It provides health and welfare coverage, pension, annuity, and COBRA to members who move between employers within the same trade. A third-party administrator runs the fund’s day-to-day operations: tracking who is eligible this month, processing employer contributions, paying claims, and producing an audit trail that survives a Department of Labor review.

It is real money, real healthcare, and real retirements. It is also, in a lot of cases, running on software written before the people using it had email.

What “1990s software” actually means day to day

Remittances arrive on paper or in spreadsheets. An employer reports the hours each member worked and the contributions owed. In a modern system that is an electronic submission validated on receipt. In practice it is frequently a spreadsheet emailed to an administrator, who keys it in. Every manual step is a place where a member’s eligibility can silently break.

Eligibility is computed by hand, or nearly. Whether a member has coverage next month depends on hours worked across possibly several employers, against rules specific to that fund, with reciprocity agreements when they work under another local’s jurisdiction. This is exactly the kind of deterministic, rule-driven computation software is good at. It is often done in a spreadsheet by someone who has been doing it long enough to know where the edge cases are. That knowledge leaves with them when they retire.

There is no self-service. A member who wants to know if they are covered calls the fund office. So does the employer wanting to know what they owe. Every one of those calls is a person answering a question a system should have answered.

The audit trail is assembled after the fact. When the DOL comes, someone reconstructs what happened from records that were never designed to explain themselves.

Why nobody fixed it

Not because it is hard. Because of who has to say yes.

A Taft-Hartley fund is governed by a joint board of trustees, equal seats for the union and the participating employers. That structure exists on purpose, so neither side can direct fund assets to its own advantage. The Taft-Hartley Act itself is silent on plan administration and vendor selection, which means choosing an administrator is not a procurement process with rules. It is a governance decision, made in a room where both sides have to agree and neither side gains from being the one who pushed.

Now add ERISA. Every trustee on that board, appointed by labor or by management, carries the same fiduciary duty, and that duty is personal. Selecting and monitoring service providers sits squarely inside it, and litigation over exactly that has risen sharply in recent years.

Sit in that seat for a moment. You are a trustee, often a working person with another full-time job, personally exposed if a decision is later judged imprudent. The incumbent system is bad. Everyone in the room knows it is bad. But it has processed eligibility for fifteen years, the auditors are used to it, and nobody has ever been sued for keeping it.

Switching means you personally underwrite the risk that a migration of decades of eligibility history and contribution records goes wrong. If it goes well, the fund saves money. If it goes badly, members lose coverage and your name is on the motion.

That asymmetry is the reason this software has not improved. It has almost nothing to do with technology.

The vendors understand it perfectly. Pricing stays unpublished and quoted per module and per fund, so nothing is comparable without entering a sales process. Every capability is a line item. The switching cost stays deliberately enormous, because the harder it is to leave, the less the product has to improve.

So the industry sits in a stable, unpleasant equilibrium: an incumbent with no reason to build and a customer with no credible way to leave. Breaking it takes a trustee willing to spend political capital inside a joint board, on a vendor nobody has heard of, for a benefit that shows up a year later.

That is the real barrier, and you do not clear it by writing better code.

The general lesson

The industries running on the oldest software are rarely the ones where technology is hardest. They are the ones where whoever has to approve the change is personally exposed if it goes wrong and personally safe if they do nothing.

Find that asymmetry and you have found why the software is bad. You have also found what you actually have to build, which is usually not the feature list. It is whatever makes the decision defensible for the person who has to sign it.

Healthcare and retirement for union members will never be a glamorous problem to work on. It is a good one, and the people carrying it deserve better tooling than they have.