Copy One That Works

When I first began to repair computers, I quickly discovered the value of having a working computer next to one that was broken. Even better if the machine that was working happened to be the exact same make and model as the broken one. For starters, a working computer could access the Internet so I could look at manuals and search forums for other people experiencing the same problem. However, the advantages went way beyond that.

Having “one that works,” a functioning copy of something broken, gives you the ability to swap parts between the two machines in an attempt to isolate a failing component. If you think a particular part is the culprit, you can easily test your hypothesis by taking it from the working machine and putting it in the broken one (and vice versa).

A working version is also useful as a model to copy. In our lives, we employ the principle of modeling all the time. If you meet someone who is successful at something you aspire to, you want to ask them: “What steps did you take to get where you are?” The idea being, if you took a similar path, you could recreate what they’ve done. If someone serves you a bacon-topped donut (what a friend!), you may feel compelled to inquire about the recipe, with the hopes you can pull it off in your own kitchen (a recipe is a model, too). Working machines can’t answer questions about why they’re “successful,” but you can observe them and infer things about how they function. A working model is the basis from which to take measurements, see where parts are located, watch normal operation, and copy configurations and settings. What’s different between the working one and the broken one?

Again, Just One Thing

Be sure to only change one thing at a time when you’re swapping anything between a machine that works and a machine that’s broken. Always retest for the failed condition after you make a change. If you replace two or more components simultaneously, and it suddenly works, how will you know which change actually solved the problem? That’s right, you won’t.

Proceed cautiously with your working copy, especially if it’s your only one on hand. Make sure you reverse any changes you make to your model before proceeding to test a new theory. Tearing apart your functional clone is like messing around with your life raft while adrift at sea. Having two broken machines is much worse than one, especially if you intended to use the working one for actual work while you fixed the broken one.

Warning: Corruption Can Occur

While swapping components is a great strategy for isolating and testing suspect parts, be aware that sometimes transferring a working component to a failed system will destroy the part. For instance, if you have a dodgy power supply that has burned out a motherboard, and you swap a fresh motherboard from a working computer…well, there goes another motherboard! As noted above, try to think a few steps ahead with regards to your working replica: if accidentally destroying “the one that works” will leave you in a desperate situation, you should avoid swapping parts. Since this decision is highly contextual, you’ll have to use your best judgement about the risks and rewards of trading components.

Weird Stuff

You would think that having a pair of identical, working machines would allow you to move parts between them with no change in the working status of either system. That’s what you’d think, but you’d often be wrong! When it comes to swapping parts among identical machines, a down-to-the-atom level of similarity is an ideal that you’ll never reach (although the nanotech industry is working on it). You never really have two “identical” systems to swap parts between. Sure, they may be the same make and model year. Heck, they might even have rolled off the assembly line one right after the other! But, they’re not really the same by the standard of an atom-by-atom comparison. On top of that, when you’re repairing machines that have actually been doing work, even machines that started out the same at the factory will have diverged over time.

Slight, but meaningful, differences are behind the similar-but-incompatible phenomenon:

• Model year: this year’s model and last year’s model may look the same on the outside, but manufacturers often make internal changes that may be difficult to see (or understand!).
• Manufacturer’s revisions: apart from the model year, manufacturers will frequently decide to change—pretty much anything—and still keep the same model number. Components can be added or eliminated, miniaturized, rearranged, etc. The manufacturer may do this because they’ve found a cheaper way to make the product or to eliminate defects found in earlier production runs. On top of that, different facilities may be manufacturing the “same” product, but doing so with slight variations.
• Firmware/software version: for any system that has digital components, there will be software. Those crafty programmers, having nothing better to do, can churn out revision after revision—as if that’s what they were paid to do. Depending on what version of the software was current at the time it rolled off the assembly line, or what was current at the time it was last serviced/updated, you can have multiple “identical” machines each running slightly different code.
• Usage: over time, machines will begin to reflect how they are used. Differences in workload, accidents, location of installation, and time in use will begin to impact tolerances between components and leave unique wear patterns. Eventually, these differences may prevent a part from being swapped between two supposedly identical machines.
• Maintenance history: regular maintenance (or lack thereof), rebuilds, use of different brands of replacement parts, and customizations can subtly change a machine. If two machines are not maintained exactly the same way, they will diverge over time.

Note that some of these differences might not be visible to the naked eye (especially minute differences in tolerances created by the wear and tear of moving parts). Keeping this subtlety in mind will improve your chances of success when employing the “copy one that works” strategy.

Don’t Just Copy Machines, People Can Be Modeled Too

I’m not talking about strutting your stuff on the catwalk. We’ve discussed copying a working machine, but don’t stop there. If you know someone who’s really good at fixing a particular type of system, consider imitating them as well. How do they make their diagnoses? Which tools do they use? What kind of training have they received? While writing The Art Of Troubleshooting, I interviewed many great troubleshooters to learn their methods. You can do the same: ask if you can observe them on the job, or offer to buy them lunch and pick their brains. What do they know that can help you become a better problem-solver? Ask and you shall receive.

References:

• Header image: Katie Montgomery, photographer. Retrieved from Unsplash, https://unsplash.com/photos/j5dD6jNLhHk.