Raft Consensus Algorithms

Beggars can't be choosers and we're all beggars.

1. Listen to the chorus

I was about 13. At the end of the year, for the 8th-grade graduation ceremony, my entire 7th-grade class would sing a choral rendition of something—Psalm 119, maybe? My friend and I weren’t going to be there that day, so once a week, while everybody else practiced, we sat in another room reading books from the school library. Which, to be honest, was basically everything I wanted out of life.

But then an administrator walked past that room and saw two unsupervised children doing extracurricular activities.

From then on, we were required to attend chorus practice with all of our peers. The teacher couldn't give us parts since we weren’t actually going to perform, so instead she came up with a separate curriculum just for the two of us that could be done in parallel—lol, no, of course not. What teacher has that kind of time? We just sat quietly in a corner listening to seventh graders learn to sing.

2. (comprehend)?

I was a young programmer at a medium-size tech company. It was, for a while, a great place to work. Even though we were writing proprietary software, we were allowed to publish generic parts of it as open-source libraries. I found this thrilling and often looked for opportunities to do it.

Until one day, I got a new grandboss. At a meeting to demo what we'd done that week, I shared one of those generic libraries I’d made—it was inspired by Python’s “list comprehensions,” so I’d named it comprehend.

Grandboss wasn't there, but he watched the recording and emailed the team afterwards: “Why are we working on items that aren’t on the roadmap (comprehend)?”

I probably could’ve put comprehend on the roadmap if I'd realized a week or two ahead of time that I was going to write it. But I hadn't. I'd just been working on a roadmap item, noticed it’d be easier if a library like that existed, and when I had spare time, I wrote it up. It was useful to the company, but not something anybody had signed off on.

Soon after, grandboss instituted a new policy forbidding such unscheduled work. Once a year, you could schedule a week in advance where you’d be allowed to work on passion projects, but the rest of the time, you needed to follow the roadmap.

So the next time I had spare time at work, I spent it on prep for my next batch of scheduled tasks—lol, no, of course I didn’t. That would be boring. I spent it in the break room playing Fluxx. Beggars can’t be choosers, and we’re all beggars.

3. The same or similar

At my next employer, I was leading a small team. For an upcoming project, we needed to hire someone with a certain rare skillset. It took a bit, but we eventually found and made an offer to someone with the knowhow who was interested in switching jobs.

But it fell through. He was worried that the job description was too different from the one he had a work visa for. “Job porting,” in U.S. regulatory jargon, requires that jobs be “the same or similar” as judged subjectively by immigration officials “based upon a preponderance of the evidence.” We were offering more money and more interesting work, but his best guess was that joining us would increase his risk of being deported. After a few weeks of back and forth, we gave up on trying to reassure him.

So then we shrugged and hired our second choice, a native-born U.S. citizen who—lol, no, of course not. There was nobody else in our pipeline. We just canceled the project.

4. She lacks impact

A 2018 BBC series recreated the training conditions for the World War II Special Operations Executive, in which civilian volunteers were selected to be sent behind enemy lines. The hope was that they could succeed at various covert operations precisely because they didn’t look like stereotypical soldiers or spies. The first phase of the selection process included an exercise the showrunners took directly from the recently-declassified manual. The recruits were split into two groups, leaders were appointed for each, and each was given the same task: using the materials at hand, get the entire team to a floating platform in the center of a pond, retrieve a radio there, and return to shore without anyone—or the radio—getting wet. Nearby were some barrels, planks, and twine.

Team A was led by Dan Dewhirst, a handsome and flamboyant property developer/actor whom everybody agreed showed obvious leadership potential. He kept the team focused, and in a little over half an hour, they’d assembled their raft and all climbed aboard. They almost made it to the radio before capsizing.

Team B was led by Vicki Wright, a former police officer. Her team hadn’t picked her as the leader, but the instructors wanted to “put her under a microscope.” She did not keep her team focused—there were multiple conversations going on at once, with different people brainstorming, proposing raft designs, and looking around. One of them, Alastair Stanley, found that just out of sight, a small pre-made raft had been hidden in the bushes. The team also realized that while they had to get everybody to the platform and back, nobody had said they all needed to go at the same time. So, taking turns in groups of two, they each rowed to the platform and back on a well-made raft well under its max carrying capacity. They completed the task successfully in 22 minutes.

So when it came time to make the first round of cuts from the training program, the instructors recognized the success of Team B’s strategy and Wright’s leadership—lol, no, they kept Dewhirst and eliminated Wright. “She lacks impact,” they said of her performance in the pond task. She let the planning degenerate into “chaos,” and then was “lucky” that other members of her team were able to find a solution.

(I looked them up on LinkedIn, incidentally. Dewhirst's still acting. If I found the right Wright, she’s now a financial regulator. Stanley, who found the raft, became a programmer.)

5. Playing Fluxx

Here is an ancient and terrible truth about this world: if somebody is doing a good thing, and you try to force them to do a different, better thing, they usually end up doing neither.

But it’s often in your personal interest to do it anyway.

Students left unsupervised doing self-directed study are all risk and no reward for a school administrator. If they manage to get themselves hurt, you’ll be blamed for not monitoring them. If they learn a few things on their own, you can’t exactly take credit for it. So when you see two kids reading library books, it really is your best move to have them do nothing instead.

The grandboss who gently chided me for doing extra work was behaving in exactly the way that got him promoted over my boss. She’d been his boss once, but he Demonstrated Leadership Ability while she just went around talking with people all the time. She wasn’t Setting Priorities and Taking Ownership, just helping get things done. He was the man with a plan, even if the plan rarely worked. No matter my role, no matter the company, I’ve almost always reported to a Vicki Wright who reported to a Dan Dewhirst.

If you look at Team A and Team B in the pond challenge and judge them solely by results, you can easily see which team had better leadership that day. Team B was all but guaranteed to succeed, so long as at least one person on the team had the right idea. But if you look at the leaders and the results for them personally, clearly it was Team A’s leader who made the right choice. He dominated the discussion, kept the spotlight, and got to stay on the reality TV show for another episode. Team B's leader, whose only interventions were aimed at preventing one voice from dominating the discussion, left feeling like she had done the very best she could, which I suppose is its own kind of reward.

But the overall effect of the system, of course, is that you end up putting the property developer/actor in charge of everything, and pretty soon everybody’s sputtering in the water.

So let’s talk about how we can achieve a better system. One that doesn't incentivize people to become obstacles to the team’s success.

Any thoughts?