Doghouse Science #1: Hypotheses and Alternate Explanations

"Common sense is genius dressed in its working clothes." - Ralph Waldo Emerson

With all due respect to Emerson, I prefer to say something different: "Common sense is science in its working clothes." All too many woos think that science is something that people do in sterile white rooms with funny-shaped glasses full of bubbling, colorful liquids while wearing a long white coat and pocket protector. You know, just like they do in Hollywood movies.

Those may be common trappings of science, but that is the shadow, not the substance. Wearing a long white coat with an ID tag ending in "PhD." is not a sign of authority in itself. The authority is in the diligence in conducting experiments and recording observations. Today, I'll be writing a post about how I used science to solve a mundane problem.

This story takes place about a year and a half ago when I moved into my apartment and started turning it into a swinging bachelor pad. Which is to say, after getting all the matching furniture I bought in place, I needed to get all my electronics fitted into a rather large entertainment stand. I'm a geek, after all. I decided to get my local cable company's TV and Internet bundle.

There were a few problems with the installation, but that wasn't entirely unexpected. The new guy on the installation team messed up some connection out there in the apartment complex, so I was without television for a few hours, but that got sorted out.

Over time, I started getting internet outages where I would either nothing would come through or the connection would be as slow as molasses. They gradually grew worse. They mentioned going through maintenance and expansion here and there so I accepted the first of it as transitory. Eventually I had an outage that lasted over 24 hours and showed no sign of being fixed. I was going to call tech support and give them a piece of my mind.

But first, I had to get down to work. One common joke among tech support groups is "Problem exists between keyboard and chair." The vast majority of technical difficulties are the result of user error, and if I didn't do everything I could to prove it wasn't my fault, they would have no reason to suspect the problem was on their end. If I didn't go through all the common mistakes I could think of, I would look like a lazy fool to the customer service representative. This is what I thought up:

Hypothesis₀ (Null hypothesis): It is not the cable company's fault.
---Hypothesis_0A: My laptop has a problem.
---Hypothesis_0B: My wireless network has a problem.
---Hypothesis_0C: My router has a problem.
---Hypothesis_0D: My router is not hooked into the cable modem.
---Hypothesis_0E: The cable modem is unplugged.
---Hypothesis_0F: The cable modem is not connected to the splitter.
---Hypothesis_0G: The splitter's input is not hooked up to the cable output.

Hypothesis₁: It's the cable company's fault.
---Hypothesis_1A: The apartment connection was set up improperly.
---Hypothesis_1B: The company's internet service was down for my region.
---Hypothesis_1C: They provided me with faulty equipment.

Notice all the parts of the null hypothesis: If it was my fault for any of those reasons, they would be justified in forcing me to go through a script, and I'd have to go through the tangle of wires AND be on the phone, wasting the support person's time.

So I got to work. I checked my laptop's connection. It reported being connected to the router, but with no internet connection. I turned on my Playstation Portable and attempted to connect to the wireless router. It succeeded, but still failed when I tried to browse the web on it. 0A down: My Windows Laptop and PSP work on very different systems. It is highly unlikely they would both share the error if something was wrong with my laptop.

This also provided evidence against 0B, the wireless connections being the problem, since they both recognized the connection. I went one step further, turning on my PS3 and attempting to sign onto the PSN. I also turned on my old desktop and attempted to browse the web. Both of them had a physical connection to the router instead of a wireless connection. Both failed. The problem was unlikely to be in the wireless connection. 0B down.

I reset my router and watch it when I send an internet request. The lights blink at the same time. The router appears to be working. 0C appears unlikely, so I dismiss it until given evidence to the contrary.

I follow the cables. The router is connected to the modem. 0D down. The modem is plugged in, and its LEDs are blinking. 0E down. The modem is hooked into the splitter's output. 0F down. The splitter is hooked up to the cable in the wall. If it weren't, I wouldn't be receiving my cable TV, and I was. 0G down.

Hypothesis₀ (Null hypothesis): It is not the cable company's fault.
---Hypothesis_0A: My laptop has a problem.
---Hypothesis_0B: My wireless network has a problem.
---Hypothesis_0C: My router has a problem.
---Hypothesis_0D: My router is not hooked into the cable modem.
---Hypothesis_0E: The cable modem is unplugged.
---Hypothesis_0F: The cable modem is not connected to the splitter.
---Hypothesis_0G: The splitter's input is not hooked up to the cable output.

Hypothesis₁: It's the cable company's fault.
---Hypothesis_1A: The apartment connection was set up improperly.
---Hypothesis_1B: The company's internet service was down for my region.
---Hypothesis_1C: They provided me with faulty equipment.

Now that I did a thorough check, it's now much more reasonable to entertain the hypothesis that the cable company is at fault. Only after I addressed those known concerns do I call customer support. Thankfully, I got one who didn't blindly follow a script and listened to my description of the problem, and my troubleshooting efforts. He noted that no one else in the area complained about outages, so hypotheses 1A and 1B are down. He asks me for more information about how I have my television and modem set up: Wall cable -> Splitter -> Modem on the left, TV on the right. He proposes an experiment that will test two hypotheses, one of which I had not considered:

Hypothesis_0H: The cable connection to the splitter's left output was loose, but not visibly so.
Hypothesis_1C: The cable company provided me with faulty equipment. (Specifically, a splitter with a failing left output).

The experiment is simple: Unscrew both the TV and modem cables and switch their places, making sure they're tightly screwed on. If hypothesis 1C was correct, it predicted I would regain my internet connection but lose my good television signal. If hypothesis 0H was correct, it predicts I would have both working television and internet signals.

The result: I turn the TV on and get a clear signal. I refresh my laptop's browser and get my pages to load properly at a good speed. Hypothesis_0H alone predicted this outcome. I was justified to suspect, but ultimately wrong to think the cable company was at fault. But I'm happy with the result. The problem was solved, and I learned that a loose cable connection can provide a poor internet signal instead of no signal at all. It's possible I ended up shifting or loosening the connection while sorting through all my game systems.

Conclusion: My internet outage was most likely due to a loose connection.

That's how science works. It doesn't matter if you started out right. You use science to become right. Because I went through all the trouble of eliminating the other problems, someone who knew more possibilities was able to arrive to the real answer much more quickly. I didn't waste his time, and thus I didn't waste any of mine.