Bilbo’s escape from the Elvenking’s halls is not elegant. He packs the dwarves into empty wine barrels and floats them down a cold underground river — bruised, half-drowned, thoroughly undignified. It is also the only plan that works. The guarded gate is impossible. The wild river is merely unpleasant.
I have a soft spot for this chapter, because the most instructive delivery lesson of my career looked, from the outside, exactly like a barrel ride.
The lab that outperformed my whole organisation
At a global technology company, I ran a formal 24/7 production organisation — hundreds of people, proper service management, SLAs, escalation chains, the whole guarded-gate apparatus. Next to it sat a lab of five or six people. Whiteboards, broad skills, their own tooling, direct lines into the business. Officially an experiment shop. Unofficially, a wine barrel.
I got curious and dug into the numbers. The lab’s workload was roughly equal to my entire formal production. Many of its “experimental” systems were in fact business-critical production — running with no formal SLAs at all, just expectations that were consistently kept. And it was doing all of this far more efficiently than the formal organisation.
I started calling it an official form of shadow IT, and I meant it as admiration. What made it work was exactly what the formal apparatus had been designed to replace: direct customer connection, and pragmatic, skilled people who owned the outcome end to end. The elves guarded the gate; the barrels floated the goods downriver anyway — undignified, unapproved, and effective.
The uncomfortable conclusion was not “abolish the SLAs”. At scale you need the gate. The conclusion was that when the informal channel consistently outperforms the formal one, that is not a compliance problem — it is information about where value is actually created. Most organisations suppress that information. Very few measure it.
What this means for your transformation
Unconventional is not the same as reckless. The difference is discipline:
- Judge routes by outcomes, not by respectability. The lab looked unserious — whiteboards, no SLAs, no steering committee — and it outperformed the formal machine. If you only trust methods that look like governance, you will systematically overpay for elegance.
- Keep the roughness honest. The lab worked because expectations were kept, not because rules were absent. Controlled roughness stays controlled only while the trade — closeness and speed in exchange for formality — is named out loud and its promises are met.
- Measure the informal channel instead of suppressing it. Somewhere in your organisation the barrels are already floating past the guarded gate. The only real choice is whether you learn from that or punish it.
The same tension is everywhere in AI adoption now: the exhaustive evaluation that finishes after the opportunity has passed, against the faster, rougher, transparent approach that fits the real clock. Rigour matters. But rigour that arrives too late is just a slower way to lose.
Tolkien’s joke in the chapter is that the dwarves complain bitterly about the indignity, having just been saved by it. Steering committees do the same. That is fine. They don’t have to enjoy the barrels. They have to get to Lake-town.
“There is nothing like looking, if you want to find something. You certainly usually find something, if you look, but it is not always quite the something you were after.”
— The Hobbit, Chapter 4
Part 5 of ten in There and Back Again — Transformation Lessons from an Unexpected Journey. Previous: Navigating Mirkwood. Next: Sting Glows Blue — on the early-warning signals most programmes ignore.
Leave a Reply