In 2011, Mike Lynch stood at the center of the biggest tech deal Britain had ever produced. HP (Hewlett-Packard) paid him $11 billion for Autonomy, the software company he had built from a Cambridge research lab into a global player. The press called him the Bill Gates of Britain. He had done what almost no European founder had done at that scale.

Within a year, HP wrote down $8.8 billion of that deal and accused Lynch of engineering the numbers that made Autonomy worth buying. Lynch spent the next thirteen years insisting the collapse in value had nothing to do with fraud and everything to do with HP mismanaging what it bought. A UK civil court eventually ruled against him but a US criminal jury, hearing the same core allegations, acquitted him on all fifteen charges. Two courts, two standards of proof, two different answers to the same question: what did each side actually expect from the other, and did either of them test it before signing?

Somewhere between the euphoria of closing the deal and the write-down a year later, an expectation broke. Maybe it was Autonomy's numbers promising more than the business could sustain or it was HP's confidence that it could integrate and grow what it bought or possibly both. What's not in dispute is that neither side had rigorously interrogated the expectations underneath an $11 billion bet, and the fallout consumed a decade of court battles, reputations, and money that dwarfed the original deal.

Most of us will never sign a contract that size but we all carry expectations into deals & situations far smaller and just as unexamined.

​
​Three ways an expectation goes wrong

Every expectation that causes you trouble tends to fail in one of three ways. It's unrealistic, built on a version of events that was never going to happen. It's unknown, sitting quietly in your own head where you haven't even fully named it to yourself. Or it's unexpressed, sitting in your head where the other person definitely hasn't heard it.

Autonomy and HP managed to hit at least two of the three. Whatever the true state of Autonomy's numbers, HP's expectation of what a fast, clean integration would look like was never stress-tested against the reality of running a British software company under Silicon Valley ownership. And whatever Lynch believed about the company's trajectory, he never had the conversation that would have highlighted the gap before the deal closed rather than after the shareholders' lawyers got involved.

You don't need a billion-dollar acquisition to feel this. You need a manager who assumes you know what "done" means, or a report who assumes you'll notice they're burning out without them telling you.

​
​The expectations you set for yourself

The harder version of this problem isn't the one you have of other people. It's the one you have of yourself. Somewhere, an engineer is running the equivalent of raising three kids, having dinner on the table every night, working ten-hour days, and looking put together when the day ends. Theirs sounds like this: answer every Slack message within the hour, ship without bugs, stay two steps ahead of the roadmap, and still make it home in time to be present with their family.

Nobody assigned that standard. It accumulated, one unexamined assumption at a time, the same way Autonomy's growth targets and HP's integration timeline accumulated before anyone stopped to ask whether they were still grounded in reality. Self-imposed expectations are the most dangerous kind precisely because there's no counterparty to push back. You are the one who has to catch it.

​
​What it costs when nobody does this work

Lynch's story is the extreme version of what happens when expectations go unexamined for too long. Thirteen years of litigation extradition, a criminal trial and a civil judgment that outlived him. None of that started as a legal problem. It started as a gap between what one side expected and what actually happened, left to fester instead of getting confronted early.

Most of us won't generate thirteen years of court filings. We'll generate something smaller and just as corrosive: the resentment of a manager who feels let down by a report who never knew the bar existed, or the quiet burnout of someone who has been failing a standard nobody but themselves ever set. Failed expectations create stress whether or not you acknowledge them. Silence doesn't neutralize the debt. It just defers the interest.

How to handle an expectation before it handles you

The way out starts with the origin. Before you can do anything productive with an expectation, you have to trace it back to where it came from. Is it based on what the role actually requires, or on a fear of what will happen if you fall short? Is it inherited from a past job, a difficult boss, a standard you absorbed without ever agreeing to it? Autonomy's growth numbers had an origin. So did HP's confidence that the acquisition would be smooth. Neither side seems to have interrogated theirs closely enough before betting billions on it.

Once you know where an expectation came from, test it for realism. Not for whether it feels important, but for whether it is actually achievable given the resources, timeline, and people involved. This is the step that turns a feeling into a decision.

From there, you have three moves. Some expectations are worth negotiating, opening a conversation about what's actually deliverable and finding a version both sides can live with. Some are worth eliminating outright, because they were never realistic and no amount of negotiation will make them so. And some are worth readjusting, keeping the spirit of the expectation while changing its shape to fit reality. A roadmap commitment might get negotiated into a different date. A self-imposed standard of answering every message instantly might get eliminated entirely. A promotion timeline might get readjusted once the actual scope of the work becomes clear.

None of that works, though, without the last piece.

The expectation you never say out loud isn't managed

You can trace the origin, test it for realism, and choose the right move, and still fail if you never say it to the person who needs to hear it. The most important thing about an expectation is communicating it in a way the other party will actually understand, not just in a way that felt clear when it was still in your own head.

Compare two versions of the same situation. In the first, an engineer expects to be considered for a promotion this cycle and says nothing, quietly frustrated when it doesn't happen. In the second, the same engineer tells their manager directly: I want to be considered this cycle, here's what I think that requires, is that realistic. The expectation didn't change. Whether it had a chance of being met did.

Lynch and HP had a version of this conversation available to them the entire time the deal was being negotiated. Neither side seems to have had it with the rigor the stakes deserved. Thirteen years later, courts were still trying to reconstruct what should have been said in the room before anyone signed anything.

It's tempting to file this under emotional intelligence and move on. It belongs closer to risk management. How you handle expectations, your own and everyone else's, is one of the clearest predictors of how much of your career runs on unnecessary stress. Managed well, expectations become agreements you can build on. Left unmanaged, they become debts that come due at the worst possible time, the way Autonomy's did.

You will probably never sign an $11 billion deal. But you'll set expectations this week, of your team, your manager, and yourself, and how carefully you handle them will decide whether they become a source of alignment or a source of stress you didn't have to carry.

​
That's all for this week.
​