4 x 5DII in Freezing cold, snow and wet

There was a recent report of a photo-journey to Antarctica, where several Canon 5Dmk2 dSLRs failed, while essentially "none" of the other cameras onboard did.

The above link is to DPreview.com, where a gentleman posted a "mine didn't fail" report. A long conversation resulted, with one poster pointing out (correctly) that there were a lot of issues and that the report of a non-failure wasn't particularly insightful, particularly when it was in attempted response to a field report of failures.

Unfortunately, while some of the exchange did get a bit heated, the moderators at DPreview have decided to slash-n-burn their way through the thread, and in doing so, resulted in collateral damage of posts that did not have any possible violations or controversies. Which included both of mine. As such, I see that I can no longer trust dpreview to retain professionally based objective works.

I'm not going to ask DPreview to consider undoing their moderation - that's their prerogative, and their actions reflect on their reputation only. Instead, I'll reiterate where it can't be removed:


Philip Harle wrote:
> Spent last week on the Light & Land photo trip to Glencoe. I was
> amongst 4 5DII users who managed to get their camera soaking wet and
> cold shooting for a whole day whilst it was constantly snowing. None
> of the cameras had the slightest problem.

I'm not about to re-write my long objective statistical analysis that has apparently been removed by the Moderators for whatever reason. My editorial comment on this matter is that I have noted that it has been removed.

To reiterate in much shorter form - - my apologies, but I'm not about to go into the same level of detail:

Part I:

- The above 'zero failures' report is a sample size of (n=4 x 1 day)
- The controversial Antarctica trip was a sample size of (n=26 x YY days)

That's at least a 60:1 ratio in the "power" of the respective statistical samples. As such, even if the suggested 20-25% failure rate is true, this report's sample size lacks sufficient sampling"power" to have a reliably high confidence to be able to detect the failure(s) in the first place.


Part II:

- most people don't really understand Statistics.
- most people don't really understand Test & Evaluation (methods & standards)
- most people don't really understand "self selected" sampling bias

Nor do most people understand how these interact and make the analysis of a complicated device used in uncontrolled settings and then subject to anecdotal reporting, variable judging and self-selection bias ... simply results in a mess to try to professionally analyze.

As such, all that can really be concluded is that the LL trip reported an 'alarmingly high' failure rate in 50% of their sample, which under a null hypothesis of 'All dSLRs are about the same' then may have been coupled with an 'alarmingly low' failure rate on the other 50%.

Part III:

It should be noted that even if the failures are eventually determined to have been caused by 'human error', there remains the niggling issue that said human errors were not randomly distributed, but clustered. To cut to the chase, something that significantly alters the probability of human-contributed errors ... infers a system design flaw.

Part IV:

How to rig a 'waterproof' test so that even an exposed Kleenex can pass?
I previously only said that it could be done. Here's some concrete suggestions as to how it can be done:


Method A: low flow rate + atomized to mist + extremely dry chamber + high temperature + air make-up + good separation distance = weak humidifier

Method B: medium flow rate + spread + very dry chamber + extreme cold + good separation distance = dry snow machine, or possibly even just verga

Method C: high flow rate + no spread + aimed horizontally at target + distance + gravity + splash control = water misses the test coupon

Method D: "before/after" weigh scale not sensitive enough to measure weight change from water, or use of a method that doesn't measure relevant change (eg, dimensions).

Method E: handling of sample after test (eg, time delay, allowing it to dry).



-hh

Iranian photoshop manipulation? An illustration

There's been media reports (eg, NY Times) this week about Iran's missile test ... and how the images distributed very well likely might have been faked - - a Photoshop manipulation.

I'm sure others have already done this, but I didn't notice any examples, so I've put one together quickly for anyone else looking for the same: here's a very simple Photoshop overlay of the two ("4 missiles" and "3 missiles") photographs, so that people can decide for themselves how suspicious they are of if this is (or isn't) a manipulation.




In other words, 'decide for yourself'.
(you can click on the image for a larger version)


-hh

Customer Service in the Internet Age

The above links to an MSNBC article of the title:
"Complaining Couple Banned from Cruise Line".


Briefly, the article discusses how a cruise line (Royal Caribbean) chose to permanently decline the business from a particular customer (that sounds nicer than "ban"). Apparently, they had found reason to complain ... and apparently ask for meaningful financial compensation ... over 80% of the time (5 out of 6 cruises).


There's a few interesting points in this topic.

First, there are most definitely consumers (such as the Morans) who will frequently exploit any excuse to forcibly leverage a cost concession. In this regards, Royal Caribbean is better off without them and does have the right to decline their business.

However, there is the issue with 'freedom of opinion' on discussion groups. It is safe to say that literally no discussion group has been free of pressure from special interests to delete or amend existing comments, which frequently leads to censorship issues.

There is an old saying that the Internet interprets censorship as network damage and routes around it

In general, this potential problem rapidly becomes a slippery slope when the Message Board is being hosted by the company...a case of where Marketing trumps ethics and fear of "Brand Damage" is more important than hearing real consumer feedback (in order to improve the Brand): it is a manifestation where the corporation reveals a lack of confidence in the strength of their Brand.

A few years ago, I was a reader of just such a group that was hosted to benefit Divi Resorts (a Timeshare company) and there rapidly became a zero tolerance ... enforced through censorship ... for any and all possible criticism from their customers. As such, the group was a sham: nothing more than a marketing mouthpiece, not a source of honest, balanced information, or for frank consumer/supplier dialog... something that would be more expected when one has spent $10K+ for a timeshare. And what was the fate of property that Divi had rejected years worth of customer 'feedback' about? Its death spiral continued until it was shut down in 2006. Glad I'm not a big investor in Divi Resorts, as any company who clearly chooses to ignore repeated warnings about their product's shortcomings is not going to be particularly successful in the long run.

Thus, the general conclusion that I have come to is that the only discussion groups that can be functionally trusted are those upon which censorship is impossible, which is USENET, or a Message Board whose official published policy is that all discord will be discussed fully in the open for the public to witness firsthand (good luck finding one of these).

But there is good news for this for consumers in that the Internet - - despite its shortcomings - - has been a resource that has restored some power back to the consumer, for previously isolated individuals can now communicate, and it is common to compare the company's response. It used to be that a company could merely tell 50 customers, "Gosh, you're the first one that this has ever happened to!" to try to smooth things over and to minimize financial recompense, but today, that claim is far more likely to get caught and thus revealed as a lie.

Since its only a fairly small percentages of consumer who have caught on to this information sharing, and that information is still dispersed across the 'Net, a businesses' tactics to use 20th Century spin-doctoring will still work 80% of the time. However, the downside risk is that when it doesn't fly, that business gets hit hard because their claims get destroyed by these collaborative power of 21st Century communications, and then they get nailed a second time ...even harder... for having flat-out lied to multiple customers.

This is not the case here with Royal Caribbean, but the amazing part is that a surprising number of companies still "Don't Get It". Here, RC was in a pattern of responding to individual complaints - - they didn't see the pattern until it was pointed out to them by other consumers.

And for the Morans? The Internet revealed their pattern of behavior, which in this case is now helping a business to avoid future complaints from them. Yes, its a two-way street: they made their bed, so now they have to lie in it.


-hh

John Gilmore was right

I recently ran across what I thought would be an interesting blog from former Apple Executive  for Federal sales, David Sobotta.   After a fairly straightforward dialog exchange ... which by definition, must have included some elements of disagreement (else why bother to reply), Dave terminated the dialog.  The topic wasn't really all that important, or profound: we simply disagreed, based on our personal perspectives.  It didn't even bother me if Mr. Sobotta wanted to curtail the discussion because he tired of it, nor even using his Moderator responsibility to have done this...that would have merely been rude.  

Instead, what bothered me was that he cut things off so as to get the last word in, a childish way to "win" a disagreement.  FWIW, this isn't my interpretation of what happened, for in this blog posting  David made his motive and intentions explicitly clear that that was precisely what he chose to do. 

Were I willing to stoop to David's level, I could have listed each of my discussion points here, where David can't touch it, to childishly get in my own "last word".  But I loathe being a hypocrite, so that won't be happening here.

So why am I writing anything?  Because this is unfortunately yet another example of something that I'm finding disappointing ... no, make that downright disturbing ... which is yet another example of someone who proves himself incapable of recognizing that with the Information Genie out of the bottle, all liars invariably get caught.  

  

This is precisely what Gilmore was warning of a decade ago:  the nature of IT is that if one tries to squash something in one outlet, there's now hundreds of alternative venues, so the word still gets out.  

There's already been hundreds of corporations who have learned this lesson by getting burned over the past decade - they were applying their 20th Century belief that they can manipulate and selectively lie without getting caught - - but then got caught, courtesy of this unbottled IT Genie.  It is effectively the Pandora's Box of the 21st Century.  

Overall, it does seem that there's a segment of the population who are otherwise very very smart people who think that their intelligence lets them stay on top of the lies that they make and avoid self-contradiction.  Unfortunately for them (my online friends know that I'm probably thinking of a certain former Clemson University Professor too), archives have a 'perfect' memory that merely takes perseverance to search, which reveals the self-contradictory claims and other hyperbola that people try to employ to obscure reality to try to force it to conform to their biases.

Thus, as John Gilmore reportedly said:

"The Internet interprets censorship as damage and routes around it".

And the latest example is in the news this week, namely the drama of the website "Wikileaks", which posted documents that revealed criminal banking activities in the Cayman Islands.  It is the apparent crook who is trying to get the website shut down, so as to squelch the dissemination of evidence of his criminal activity.   But in the meantime, the news of the event has probably caused a few thousand more copies to get spawned.   To try to suppress them all is an exercise in futility, so the next thing we know, the criminal will be claiming that he's a victim...of getting caught.  

-hh