Tuesday, 1 July 2014

It's all wrong!


Those recent posts on seminars reminded me of a bewildering seminar experience a friend of mine once had, when he gave a seminar at one of the top universities in the world. Of course, he won't let me tell you who he is, or which university. I already had to change this draft five times and argue for hours on Skype just to convince him to let me write it up completely anonymously. All I can say is that if one person so much as asks him, "Are you the guy in Mark's blog?" I'm going to be writing the next post from hospital, typing with my nose. I'm saving the ending until next time, although that probably won't be much insurance.

I think it's a good story, but he's not proud of it. When I told him that in this write-up I'd call him Richard, he said, "That sounds right. So long as you abbreviate it to Dick."

It was a university on the US East Coast. The first time he ever went there was one winter, as a postdoc, to give this seminar. It was a lucky break. He was invited several months earlier, but just a week before traveling he and his collaborator made a huge breakthrough in their computer simulations. So now he could unveil a major result to the top research group in his field.

It wasn't going to be easy, though. He had a horrible journey. His flight from Europe was delayed because of weather, and after it finally landed the roads from the airport were such a mess with snow and accidents and detours, that it took the cab three hours to reach the university. He was supposed to have arrived the day before, but in the end he arrived only 45 minutes before the seminar -- exhausted, jet-lagged, and still without the latest plots from John, his fellow postdoc and collaborator.

[Relax, man, no-one can identify you from this. A prominent East Coast university? Which is three hours from a major airport in poor weather conditions? That's probably all of them. And I made up that you came from Europe, just to throw them off.]

When he arrived the first thing he did was check his email.

There was bad news.

"There's a bug in our code," wrote John. "All of the results are wrong."

Richard closed his laptop after the first sentence.

What should he do?

John could be exaggerating. He was brilliant, he was enthusiastic, he was indefatigable, but it always took him ten attempts to get to the right answer, and if Richard wasn't there to set him straight, it took even more. Richard was the calm, thorough one. But how could he be calm fifteen minutes before the most important seminar of his life?

He tried to reassure himself.

This happened every time they got a result. Half of the time it was some temporary panic that resolved itself a day later. Unfortunately, the other times the problem remained, and it was their results that evaporated. Which was it this time? There was no way to find out.

"I was panicking," he told me later. "It wasn't like I could extend my visit for two weeks and give the seminar later -- although I seriously considered trying!"

No-one would blame him if later there turned out to be some problems that he could plausibly claim he didn't hear about until after the talk. After all, he was out of email contact until only minutes before he spoke. He had a few minutes to check his messages. He was only able to glance through a few of his many emails. Maybe he didn't even notice the one from John? Surely only spam was titled, "*** IT'S ALL WRONG ***".

So he gave the seminar.

He tried hard to be cautious about the results. "They're only preliminary," he warned. "There are still a few things we need to check."

He told me later, "I thought they were going to see right through it, and rip me apart."

But instead a professor in the audience, who was probably the world's leading expert in the topic, waved aside his caveats. "No no no! This has to be right. It makes complete sense. It's an incredible discovery."

He tried not to look so surprised, and to remain cautious. They asked for more details, and he stayed cagey, and that just made them more excited.

When the time was up, a small crowd gathered round him.

He pleaded serious exhaustion. Was it Ok if he checked in at his hotel and got some rest? He would join them for dinner later.

At his hotel he read John's full email. It was no exaggeration. The bug in the code was clear and unequivocal. It was obvious which physical effect had been neglected, leading to the astounding new results. With the code fixed, the result went away. End of story.

Richard was mortified. Everyone had been extremely excited.

"How could I go out for dinner with them now and tell them that it was all wrong?"

He begged out of dinner, claiming illness. The next morning he got an early taxi to the airport. The weather had cleared, and he escaped with his secrets intact.

As soon as he got home he met up with John and they tried to salvage the results.

The code was a mess. Richard calmed down, and methodically re-organised and checked every routine. In the mean time, John had more crazy ideas, and furiously wrote more messy code to test them. Some days their miraculous result returned, and others it sunk beneath the surface of bugs, analysis errors, unchecked assumptions and crude approximations.

It's not as bad as it looks. This was how they always worked. In the end their supervisor always set them straight. Up until now they met him once a week. Here was the routine: for fifteen minutes they tried to explain what they had been doing, before his eyes welled up with tears, and he quietly asked, "Is there a paper yet?" Then he spent 45 minutes dabbing his eyes and pointing out what they needed to do to get a clean result. The rule for the meetings was: either bring a paper, or a box of tissues. In the past year they made multiple bulk Amazon orders for kleenex, but they also wrote four excellent papers.

When he heard about the seminar he choked up immediately. As Richard put it, "He waved me straight out of his office, and the last thing I saw was his head in his hands." For the next month his sole contribution was to sit silently in his office, staring miserably out of the window. "It was just what we needed."

John loyally insisted that Richard had done the right thing. His understanding and support seemed to be endless. He worked tirelessly to make sense of their results. He worked late nights and weekends. He cancelled a long-planned concert trip with his girlfriend. His girlfriend dumped him, but now he had even more time to work on the code. He was the perfect friend and collaborator.

But only up to a point.

That point came three weeks later, when Richard got a call from that illustrious East Coast university.

"We would like to interview you for a faculty job."

Part 2 -- "Confusing Results"
Repercussions -- "Is this the end?"

No comments:

Post a Comment

[Note: comments do not seem to work from Facebook.]