Kai Puolamäki of Electronic Frontier Finland studied the relation between the corruption levels and voting behaviours of the countries that voted on OOXML. The findings were that the more corrupt a country is, the more likely it was to vote for the unreserved acceptance of the OOXML standard proposal. It should be noted that the statistical test does not prove anything about the reasons for this correlation. Regardless, the results are rather interesting, if not particularly surprising, and cast a shadow over the "strong international support" of OOXML.
I have expanded on his graph, and made it a stacked %. So it becomes clearer to see the correlation.
If the statistical test does not prove anything, why do you say the results cast a shadow?
And what ideas underlie this slur? That all people from countries with a high CPI are corrupt? That most people are corrupt? That a country with a high CPI index should be presumed to be acting corruptly? That countries with high CPI indexes should not be allowed to participate in international organizations? Or just that anyone who disagrees with you is corrupt?
Rick Jelliffe
Rick, this is a discussion topic. It is not proof of anything, it is just another of those remarkable coincidences that can be observed in this matter. Statistical tests, as you surely know, never actually prove anything, they are used to show more or less likely cases of correlation, and the results should be used to point to possible underlying causalities which need further investigation.
This result raises suspicion and casts a shadow of doubt over the fairness of the process, although not to a high degree of certainty. It shows that there would be some reason to investigate the underlying process for approval closer in more corrupt countries and look for irregularities. There is also a very clear correlation (probably more statistically significant, although I haven't checked) between unconditional approval and latecomers among ISO members and members of JTC1, as well as between approval and latecomers to the national committees.
Statistical tests are tools for finding things which might be rotten, ways to decide where to go from here. There are many things which appear to be not quite right in this process, that should be quite clear to anybody by now. This is just another indication to that. It smells and looks rotten, but we don't have actual proof in the legal sense of the word. For that, we need more than numbers and statistics, or at least some more of numbers and statistics pointing to the same possible conclusion.
This is not proof, but that does not mean that any discussions in the matter is futile and should be ridiculed. You should know better. Can you honestly say that you are convinced that all countries had a fair and just handling of the DIS 29500 ballot? Otherwise, you have no reason to criticize the very discussion of the possibility that corruption was involved. If you want to be constructive, find some reasons why corruption would not be a problem here. It is indeed a problem in many other circumstances, admittedly even in some previous standards work, but it might not be applicable here, and that would make the statistical analysis useless. Correlation without a probable link of causality behind it is just a game of numbers. However, I think there is a lot more to this.
Strange coincidences are unlikely to occur in groups unless there is an underlying reason. The more they are, the less probable it is that they are in fact all coincidences, even if any single one of them is not all that unlikely. Given enough irregularities and strange occurrences all at once, one simple and likely explanation is that they were in fact planned and coordinated, particularly if they are also similar. You cannot attack each individual occurrence and say that "this event by itself was fairly likely to have happened anyway". Statistics does not work like that. Throw a dice and you get a six, no big deal, it happens all the time. Throw it 10 times and you get a six 9 times out of 10, you start suspecting you can't be that lucky, and that there is probably something wrong with it.
In the bigger picture, there is most likely something deeply rotten at the core of all this. If you want to dispute this, you have a very high burden of proof, and I do not envy you if you want to try that.
So it is a discussion topic with no proof. But you use it as evidence anyway.
Why not say "This is just idle speculation, but all we care about is throwing mud so it doesn't matter?"
You say "Strange coincidences are unlikely to occur in groups unless there is an underlying reason." But the underlying reason is you guys: you blow up every little hiccup is blown up to be a scandal. You print unsubstantiated rumours. You don't retract information found to be incorrect. And then, knowing how each item is, you claim that it is the pattern in this shonky information that is important.
Rick Jelliffe
To Anonymous (121.210.129.72) who posted as Rick Jelliffe
NO ONE CAN DENY THE FINDINGS, even Microsoft can't (provided the inputs are correct and the methods are correct)
That is, "The findings were that the more corrupt a country is, the more likely it was to vote for the unreserved acceptance of the OOXML standard proposal"
If you are a statistician, refute the findings with your more correct statistics… if you are not and want to dispute the findings…. there are i think two options u can do here….
1. Find a statistician and prove the inputs from the statistics are incorrect making the findings incorrect.
2. Simply accept the findings… and ignore it (which microsoft does to mitigate the negative effect).
…. and you said "you blow up every little hiccup is blown up to be a scandal"
1. Sudden last minute surge in voting members of the standard body and "MIRACULOUSILLY" voting for OOXML (EFFI findings?).
2. Some members (not inclined to vote for OOXML) are not allowed to vote for a lame excuse (no more room for them?)
3. Some private letters to business partners from MS saying MS Office to ODF conversion are not perfect is contrary to
what the MS PR is saying about converters (Converters that can perfectly convert from MS OOXML
to ODF and vice-versa). So they are effective saying in public "don't worry about conversion, it's easy and has been done!"
and then, they say in private "worry about conversion, it will negatively impact your business options"… nice try :)
4. MS admitted that some letters were sent asking their business partner to join and vote for OOXML but said it was a
mistake from one of their employee (a manager making such as mistake seems very smelly).
5. MS Says "We are doing this because IBM first did this!"… childish, anyone? Did IBM really vote stuffed?
6. "We (MS) never objected to ODF being a standard so don't object to OOXML"… on what merit? last time i checked, there are
too many technical problems (big ones) embedded on OOXML and is currently only supported by MS!!!
… and some more not listed here but I hope these may be enough for you ;-)
It's not that these little hiccups are blown up to be a scandal… these little hiccups reveals the scam!!!!
"LET'S GIVE CREDIT TO WHERE THE CREDIT IS DUE…. " and the credit for this scam goes to Microsoft!
I too am a person benefiting indirectly from microsoft software… (i use XP and Office dude) but, i could use any other softwares too from many different sources not just microsoft… i am not blinded by MS or Open Source… i use whatever is required when there are specific requirements laid before me… on the other hand, when it is my prerogative to choose… i choose whatever suits me :-)
Jack M.
Jack: If you think that tests of statistical significance mean anything without a causal link. you need to study some more statistics. You could draw a graph with the average height of a country's population and the corruption index and probably find a statistical pattern, but that does not mean that one causes or reflect the other. Even the original article makes this clear. However, having said "this proves nothing" in the fine print it is utterly disingenuous to then say "it proves something" in the headlines.
1) There is no abuse of process if a national body takes an interest in a standard and decides to vote on it. That is how ISO works. You are only objecting because some of them didn't vote the way you like.
2) The problem was not with chairs, but with fixed size committees. Fixed membership, limited term task forces of stakeholders (like juries) are a common way to look at technical issues. There is no abuse of process there. You are only objecting because you don't like the people sitting in the chairs, not with their number.
3) What on earth does this have to do with corruption?
4) The mistake did not have any impact on the vote and was properly reported to the Swedish body and the public. Surely this is an example where potential corruption *didn't* occur. There was no abuse of process resulting.
5) What on earth does this have to do with corruption?
6) What on earth does this have to do with corruption?
Cheers
Rick Jelliffe
An image tells more than a thousand words…
1- I suppose that the higher the corruption level, the lower the investment in public wealth (schools, administrations, infrastructure, research…), because the decisions are taken for private interests, in opposition to a pater familias way to handle the public good.
2- a long term duration of high corruption level is probably well correlated with deep poverty for ordinary people.
3- to be forced to use OOXML format as a de facto MS-only full implementation implies :
- to upgrade most of the private and public PCs to support Office 2007
- to pay Vista licenses
- to pay MS-Office licenses
- to repeat those items every 3 years (the most expensive proposition of the list !)
At a country level, it will be a huge sum of money, perhaps impacting significantly the level of public services.
If there are specialists in this forum, It would be interesting that they collect data to build maps emphasizing this impact.
For ex. maps using a ratio like MS-'tax'/public_expenses_for_schools (or road infrastructure, hospitals, communications, social security…).
We need simple and effective ways to impress the news, and the answer must be sent when the subject is hot.