Errata – or – That’s Just Plain Codswallop!
We all need trust. Trust that the sun will shine tomorrow (OK – not necessarily in Victoria in February); trust that gravity will keep our feet firmly on the floor. We also tend to trust what we read, believing that what someone has spent time to put to words is likely reliable. There are, however, times when being too trusting is dangerous.
Designing medical devices requires trustworthy data. Engineering decisions need a solid foundation of facts, or they will topple like a termite-riddled timber. Where do we find these facts? The internet is usually the first go-to source for information like manufacturer’s spec sheets, physical constants, formulas and regulations. Fast retrieval of facts is an aid to today’s rapid design cycles, but this benefit comes with a caveat: know your source.
Wikipedia is an excellent example of the power of crowd-sourced information. Generally it percolates to the top of search-engine rankings due to the breadth of topics and the speed with which one can get an overview. Its strength – the huge number of contributors – is also a weakness, since misinformation often gets posted as well. Errata arise in online sources due to contributors’ lack of time or knowledge, miscommunication, vandalism, or vested interests. Some examples can be humorous, others (such as mistakes in algorithms) can cause failures in devices relying on them.
We trust science because it is based on reproducibility of results, and the same method can be used with online sources. Check references for corroborating evidence, check multiple independent sources for consistency and you can place more trust in your results. Just because the website is reputable does not guarantee error-free content. The US National Institute of Standards and Technology (NIST) provides a list of fundamental physical constants at http://physics.nist.gov/cuu/Constants/index.html Take note of their disclaimer – there is no warranty of accuracy.
Some information has no easy corroborating source. For example, manufacturers of integrated circuits publish data sheets detailing the interfaces for their ICs – information that is essential to connecting and using them properly in a circuit. Such data sheets can become quite voluminous, like the 4110 pages in Texas Instruments’ Technical Reference Guide for the AM335x ARM Cortex-A8 Processor. It is certain that this Reference Guide (now at revision G) contains errors. It even happens that ICs contain errors: remember Intel’s Pentium floating point bug?
As designers working frequently on the bleeding edge (pun intended) of medical technology, we often uncover new errors in components we use while trying to do something unusual and pushing the design envelope. Reading the manufacturer’s spec sheets and errata is not sufficient to arrive at a problem-free design. This makes it all the more important for us check our information sources, be thorough in our verification and validation (see “Unveiling the Verification Plan in Medical Device Design”), and assess potential impacts due to component failure using our risk management processes.