I've demonstrated no such thing as my memory being "a poor servant", only that it isn't perfect all the time. So what? Nobody's is. Scientific experiments also have a magin of error and there are probably millions of technology malfunctions every day. Scientists are also fallible even with simple things ... just look at the Hubble telescope ... the main mirror was ground incorrectly. Does that make them completely incompetent? So be fair minded about this. It isn't reasonable to blow a minor glitch in the haste of a forum post into a complete memory meltdown.
j.r.
Your analogy with experimental errors is incomplete. Experimental error, at its simplest form, comes from the smallest value a too can measure. Take a rulerl, as example. Its error margin for any given measurement will be
+ half of the smallest measurement unit (ex.
+0.5mm). Now, if you take several data points, the error margin will not be as simple as that, because you must factor the statistics you used to study the data and also the method used to gauge the final error figure. The methods will also attempt to search for anomalous data points (usually the "tails" of the Gaussian curve, values which are too big or too small and deviate too much from the population distribution). Something probably went very wrong in these cases, take your pick- operator error, typos, the wrong tool was used, the measured object was not part of the study subject, you name it. So, you have the tools' error margin and methdological issues within the experiment. You must also add to this cases where there are errors in experiment design and also on the analysis of the data. As you can see its a complex subject; the more complex is the method, the bigger are the odds of errors seeping in.
Consider the case of chemical assays- The assay itself may have say, a detection margin of 5ppb. This means I can take for granted a result like, say, a 300
+5ppbAu? No. 5ppb is the detection limit of the machine. Lots of things can go wrong within the proccess. They range from sample contamination to mistaken or changed tags, passing through operator errors at several stages of the proccess. It doesn't matter how skilled are the operators, how sophisticated is the lab, how tight are its QA/QC norms - **** happens. And if these problems are not detected (and sometimes they are not), the next step which is to build conclusions over the assay results may be compromised. And again **** happens. Its an anecdote, so take it as what it is- I've seen my share of stuff built over chemical assays which were later found to be actually explainable by errors (contamination, switch of sample tags, etc). Now, this is not anecdotal- you know how these errors can be easilly detected in complex proceedures? They are usually irreproductible. Single, isolated data points or data sets.
Now, put this within the context of visual observations (actually the recollection of them). Lots of things can go wrong and generate errors. They will range from seeing something from an odd angle to flaws in our memories of the event. The things that may go wrong also include our conclusions built over these observations. Usually we detect these problems, as in the cases where we somehow recognize it was something seen from an odd unexpected angle. Sometimes, however, something goes wrong and we do not. An UFO sighting might then arise. And guess what? It may be irreproductible. The blimp became a flying saucer. The very same quality UFOlogists use while attempting to explain why there are no reliable data is one of the trademarks of experimental/methodological errors.
Even our memory can betray us. I, for example, can not remember any reports of problems in Hubble related to unit conversions. I remember, however, problems in its optics related to manufacturing issues and upgrades at its hardware, corrected after memorable shuttle missions. I also remember reports of issues related to units conversions being related to the crash of a probe in Mars. You claim to have seen reports of problems related to units conversions in Hubble. Someone's memory is malfunctioning. Maybe mine's maybe yours (that's my opinion, BTW). But this is not the real issue, the real issue is that memories may not be as accurate as they might seem, even if certain aspects are faithfull, some others may not be.