Sunday, April 28, 2019

Ode to Kasiski

00101110 00000000 00000110 00001101 00000011 00011000 00001101 01010100 00001000 00001101 00010010
00000001 00011000 00000111 00000001 00010110 01001101 00001110 00010010 01001001 00010111 00011011
00001000 01000001 00010111 00000001 00010101 00000101 00000100 00000110 01001001 00010111 00010110
00010101 00010101 01010100 00000001 00001011 01001101 00010101 00011100 00000000 00010000 01010011
00011101 00001110 00000111 00011100 01000101 00001110 00000000 00011010 01001001 00001111 00010110
00001100 00000101 01010100 00000111 00001011 00001000 01000001 00000000 00000110 01000011 00010101
00000100 00010011 00000111 00011100 01000101 00001001 00000100 00000000 00001100 00010001 00011110
00000100 00001111 00010001 01001000 00010001 00000101 00000100 01010100 00000101 00000110 00011101
00001010 00010101 00011100 01001000 00001010 00001011 01000001 00000000 00000001 00000110 01010011
00000110 00000100 00001101 01001000 00010001 00000101 00000000 00000000 01001001 00010100 00010010
00011110 01000001 00000001 00011011 00000000 00001001 01000001 00000000 00000110 01000011 01010001
00001000 00001111 00010111 00011010 00011100 00011101 00010101 01010110 01001001 00001010 00000111
01000011 01000001 00111011 00000110 00000110 00001000 01000001 00000000 00000001 00000110 01010011
00000110 00000100 00001101 01001000 00001001 00001000 00001111 00010011 00011101 00001011 01010011
00000100 00010010 01010100 00000011 00001011 00000010 00010110 00011010 01000101 01000011 00011010
00011001 01000001 00011101 00011011 01000101 00011111 00000100 00011000 00001000 00010111 00011010
00011011 00000100 00011000 00010001 01000101 00011110 00001000 00011001 00011001 00001111 00010110
01001101 00010101 00011011 01001000 00000111 00011111 00010100 00000000 00001100 00000101 00011100
00011111 00000010 00010001 01001000 00010001 00000101 00000100 01010100 00000010 00000110 00001010
01001101 00001000 00000000 00011011 00000000 00000001 00000111 01010100 00001000 00001101 00010111
01001101 00000101 00010001 00011010 00001100 00011011 00000100 01010100 00011101 00001011 00010110
01001101 00001110 00000110 00000001 00000010 00000100 00001111 00010101 00000101 01000011 00010000
00000001 00000100 00010101 00011010 01000101 00011001 00000100 00001100 00011101 01000011 00010101
00011111 00001110 00011001 01001000 00010001 00000101 00000100 01010100 00001010 00001010 00000011
00000101 00000100 00000110 01001000 00010001 00001000 00011001 00000000 01000111

Friday, April 19, 2019

We've always done it this way

I was recently reminded of Rear Admiral Grace Hopper remark:
The most damaging phrase in the language is “We’ve always done it this way!”
When I was in high school I was a lifeguard at a waterpark with a wave pool, water slides, a cave with a waterfall, a pair of monorails side-by-side and a toddler pool. There were plenty of hazards. It was a big bustling park.

While working there I had dozens of "saves." Usually at the wave pool, where someone would get in too deep without good swimming skills. The waves would turn on and poor swimmers would find themselves in real trouble. Saving someone in the wave pool meant hitting the big red button on the guard stand to stop the wave machine, jumping in the water and navigating the crowd to get to the victim with rescue buoy in tow -- when working the wave pool, wearing the buoy was required. Depending on the victim's condition, a cross-chest carry to shallow water was sometimes necessary, but more commonly extending the rescue buoy was sufficient to help them recover.

When I went away to college I applied for a lifeguard position at the university pool. At that point I had years of experience. I'd been a good competitive swimmer in high school and thought I had a good chance at getting the job.

There was a rescue test before the interview. We went two at a time. I was in the first group. The pool was split into two lanes. Candidates stood at one end and current lifeguards were waiting in the water at the other. On the instruction of the pool manager we were to save the victims at the other end.

I reacted to the situation with what I knew. I quickly put on the rescue buoy tether, dove in the water and swam with my head up, eyes on the victim, to the other end of the pool where I extended the buoy. The victim grabbed it, I towed him to the side of the pool, helped him from the water and made sure he was ok. I was so focused on my rescue that I didn't see how the other person's rescue went. I felt confident about my effort. I'd arrived at my victim quickly, and had performed a saving motion that I'd performed many times before.

I took a seat in the bleachers to watch the next pair perform the exercise. When the lifeguards started flailing again, one of the candidates dove in and started swimming to the other end of the pool, but the other candidate didn't. Instead, she ran to the rescue hook hanging on the wall off to one side of the pool, grabbed the pole and hook, then ran to her victim and extended the pole. Her victim grabbed the hook and she pulled him to the side of the pool and helped him out.

When we were finished with the exercise, the pool manager came over and said,

"Everyone who got in the water to save their victim can go home, you're not going on to the next round."

Some of us were shocked. We had demonstrated that we could quickly react, that we could swim the length of the pool, that we knew how to use the rescue buoys, etc. Before anyone could ask why we were being sent home, the manager added,

"When you got in the water to save your victims, you took unnecessary risks and increased the likelihood that two people would drown. Never get in the water to do what can be done safely from outside the water."

He wasn't wrong. It's one of the first lessons learned in lifeguard training, but my experience working in the waterpark conditioned me to operate differently.


The memory of that experience recently returned to me. I was working a problem purely for the sake of working it when I realized my approach was wrong and the only reason I was working it that way was because I'd done it that way many times before, but always in the heat of an incident, working under pressure.

Postmortems are a standard part of the post-incident process. We often review what went wrong and how it may be prevented in the future, but we don't always ask questions about the investigative and recovery processes and how they could be optimized.

I recently started reading Daniel Kahneman's book, Thinking Fast and Slow. It's a fascinating read. Kahneman proposes that we are of two minds, System 1 and System 2 as he calls them. System 1 is sort of our default mode of thinking, it's the type of thinking we use when we react to something -- a loud noise, interpreting the look on someone's face. System 2 is the type of thinking we use to solve problems that require focus and concentration -- 17 x 24 for example.

With enough practice and familiarity System 2 thinking can become System 1 thinking. A chess grand master can look at positions on a board and recognize what the next best move is simply based on years of experience and pattern recognition whereas a novice player would have to engage System 2 and carefully weigh the options.

My experience at the wave pool conditioned me to respond a certain way. There was no rescue hook that could reach the middle of the wave pool. Saving struggling swimmers required getting wet, but the pool manager at university was absolutely right that for the circumstances in his pool getting wet was the wrong approach except under specific circumstances (i.e. a body on the bottom of the pool).

When you're working through a task or have completed it, it may be worth taking time to exam your approach. Is it the optimum method or are you doing things in the way you've always done them?

Paperclip Maximizers, Artificial Intelligence and Natural Stupidity

Existential risk from AI Some believe an existential risk accompanies the development or emergence of artificial general intelligence (AGI)...