Customize

245 Protesters Killed to Date in Iran

Discussion in 'News And Current Events' started by FintanDunne, Jul 15, 2009.

Thread Status:
Not open for further replies.
  1. FintanDunne Member

    SeaofGreen-Blog.png

    iran-casualties.png

    245 Protesters Killed to Date in Iran

    The International Campaign for Human Rights in Iran has just stated
    that the number of demonstrators killed during protests in Iran far
    exceeds government reports.

    We present results of a methodical evaluation of news reports, showing
    a high probability that between 245 protesters and 285 have died to
    date. We also conclude there has been a systematic plan to conceal
    fatalities. A plan which was in place even before the killings began:

    [SIZE="2"][COLOR="DarkGreen"]Sea Of Green Radio NEWS[/COLOR][/SIZE]
    Wednesday 15th July, 2009 at Midnight Tehran time.

    Play Audio at: mp3a.gif
    http://seaofgreenradio.blogspot.com/2009/07/245-protesters-killed-to-date-in-iran.html

    Regime's Protester Fatalities Concealment Plan:

    • Allow around 35% of fatalities to enter hospitals.
    • Hide 65% of the bodies in IRGC morgues / cold-store locations.
    • Severely restrict information on detainees to make families think missing family members are detained and not dead.
    • Very slowly allow the concealed bodies to enter Iran's normal death statistics, by pressuring families to agree to dub the deaths accidental or by natural causes.
    Our Protester Death Total Methodology:

    • Estimate deaths in Tehran from news accounts and unofficial hospital counts.
    • Estimate deaths outside Tehran using media reports of the severity of repression outside the capital. Allow for poor reporting of deaths outside Tehran area.
    • Include deaths both by shooting and torture / ill-treatment in detention.
    • Allow that maximum concealment would have been in the capital -where media presence and strong public communications systems made secrecy more vital.
    • Factor in that reports say relatives are perusing large numbers of photos of corpses when trying to identify missing relatives who may be deceased.
    • Allow for exaggeration and bias in reports.
    See Also:


    More on Methodology

    Postscript. The UK, Guardian catches up with this story:

    Original Article & Audio at:
    http://seaofgreenradio.blogspot.com/2009/07/245-protesters-killed-to-date-in-iran.html

    http://seaofgreenradio.blogspot.com/
    Short url:
    http://bit.ly/seaofgreennews
    Follow on Twitter:
    http://twitter.com/seaofgreenradio
    twitter.jpg
  2. 23:08 texmexchex this graph doesn't make sense http://iran.whyweprotest.net/news-current-events/14168-245-protesters-killed-date-iran.html
    23:09 freeIran134 It is comparing the probability of a death toll, to the possible death toll,
    23:11 texmexchex yes but it doesn't make sense
    23:11 freeIran134 how?
    23:11 *** YipMan joined #iran
    23:12 texmexchex according to the way the graph is labelled, there is approx 80% prob that 250 were killed, and a roughly equal % prob that 300 were killed
    23:12 texmexchex these two probabilities alone total to more than 100%
    23:12 texmexchex if it were a cumulative prob. graph, on the other hand,
    23:12 texmexchex the graph would never decrease
    23:13 texmexchex it would be monotonically increasing
    23:14 texmexchex or monotonically decreasing i suppose
    23:15 texmexchex with 100% probability that 0 or more people were killed, and eventually 0% as # of people increases
    23:15 *** Sporadic joined #iran
    23:15 freeIran134 i see, i guess it is kinda wierd
  3. dont worry about whether it makes sense or not. its good propaganda, and we should use it to our advantage.
  4. The number comes from all the reliable death counts added together. Where the graph came from and how they made it, I do not know. They should explain that.

    Kind of stupid for the Pasadran to block access to this site for many. You can link to this thread elsewhere (like facebook), and they can see the title there, but not the thread when they click on the title. So you fucking IRG jackasses who block this site make it so people can see the graph, but not the thread talking about it. LOL.

    Once again, you shoot yourselves in the foot and don't even notice it. You fools do that a lot, you know that? Evil eats itself.
  5. "The number comes from all the reliable death counts added together. Where the graph came from and how they made it, I do not know. They should explain that."

    Whatever the original numbers were, that the graph itself is nonsensical can be concluded by inspection of the graph itself.
  6. FintanDunne Member

    Methodology

    If it's good propaganda then let's just say 500+ were killed.
    Or hundreds of corpses are stacked in piles - as fact.
    I'm not interested in propaganda.

    You can't add probabilities like that?!

    More on Methodology

    The probablility axis of the graph reflects the cumulative confidence
    level in the underlying news reports supporting a fatality total.

    For example a single news report might say that some number of
    protesters were shot in, say Shiraz. How reliable is that source or
    news org? Have their other reports been later confirmed by other
    known high confidence news orgs? How many other reports are
    there which reference Shiraz fatalities? What's is the average
    confidence in the acumulated reports?

    Or take the Le Figaro article detailing an interview with two Iranian
    doctors who fled to France. The article said they claim that an
    unofficial tally by medical staff in Tehran counted 92 fatalities
    in Tehran area hospitals. What other claims were made by the
    doctors in that article? Did those claims turn out to be substantial
    or not? How does a claim of 92 fatalities in the Tehran area mesh
    with other news reports of hospital fatalities?

    And so on. Slowly building a picture of the true number of fatalities
    which is weighted to reflect the confidence in the underlying reports.

    We've been building the underlying data for the last month. It would
    take weeks to write it all up as a formal report. For now, we can just
    present the conclusions.
  7. "You can't add probabilities like that?!"

    If it's a discrete probability mass function (and I see no reason that it wouldn't be, since there's 0 probability that, say, 200.3 protesters were killed), then you certainly can add probabilities like that.

    Further, as I pointed out, if it's a cumulative density function, it would be monotonic.

    The graph doesn't make sense.
  8. Small correction:

    "You can't add probabilities like that?!"

    If it's a discrete probability mass function (and I see no reason that it wouldn't be, since there's 0 probability that, say, 200.3 protesters were killed), then you certainly can add probabilities like that.

    Further, as I pointed out, if it's a cumulative distribution function, it would be monotonic.

    The graph doesn't make sense.
  9. And if it were a probability density function, the y-axis doesn't make any sense. You should be able to integrate under an interval to get the probability of the outcome lying in that interval. Integrating under the graph, as is, leaves a total probability far greater than 100%.
  10. FintanDunne Member

    That's right.
    and if my grandmother was an orange,
    then elephants would be kangaroos.

    I've seen your technique many times on the net: nitpick relentlessly
    on obscure technical detail to divert from the substance of the issue.

    My agenda is to quantify the protester deaths methodoically.
    What's yours?

    I've explained what I set out to graph and I think
    most people get the general idea of rating on the
    basis of confidence in the underlying news reports.
  11. "I've explained what I set out to graph and I think
    most people get the general idea of rating on the
    basis of confidence in the underlying news reports. "

    I get the general idea of confidence ratings.

    But the graph doesn't make sense. If your defense is that it has "truthiness", as Colbert would put it, I think that speaks for itself.
  12. I offered the criticism in good faith, expecting constructive engagement as a response rather than self-righteous bluster and denial. I offer further constructive criticism: consider creating a graph like the one on display here: Probability density function - Wikipedia, the free encyclopedia
    Note that the probability is denoted within the interval rather than (incorrectly) on the y-axis. The y-axis is probability density, not probability.
  13. I think whoever posted that graph did not understand it, and took it at faith. They're probably not understanding what you are pointing out wrong with it, either.

    And on that note, I'm guessing that whoever made it is probably in the same boat.
  14. FintanDunne Member

    I am the author of the graph and the one who posted it.
    I have a scientific background and understand the points.

    The graph is not a scientific probability function. It shows
    the probable final casualty total -based on existing news
    reports, as of 15th June. I've explained that clearly.

    Iranian casualties are of little interest to maths nerds, who
    goof off on something less real than actual dead people.
  15. Sounds good to me.
  16. Whatever the intended purpose or meaning of the graph, the graph as it stands is meaningless.
    • The graph purports to plot probability vs number of casualties, as its own axis labels make clear.
    • It is clearly demonstrable that the sum of probabilities described by the graph total more than 100%. Even considering only the possible casualty totals of 250 and 300, the graph shows their sum to be greater than 100%.
    • You originally responded to this by "You can't add probabilities like that?!." If the y-axis is indeed probability (as is labelled on the graph), and not probability density, then yes, you can add probabilities like that.
    • If the y-axis is mis-labeled, and is intended to instead show probability density, then fixing the label would be the first step, but the tick-values on the y-axis would also be incorrect.
    • If the plot is instead intended to show cumulative probability, then the shape of the plot is obviously incorrect, as CDFs are monotonic.

    You defend your meaningless graph as follows:
    • Claiming that it's not "scientific."
      • You are expressing information with numbers that is not true, as is evident from the graph itself. This is misrepresentation. Claiming that the graph shouldn't be held to a rigorous standard when you yourself have decided to put numbers on it is disingenuous. If you want the standard of judgment to be lower, simply remove the false numbers.
      • You use terms like "methodology" more than once in this thread, giving the lie to your assertion that you are not engaging in misrepresentation.
    • Dishonestly invoking the justice of your cause as an excuse for the misrepresentation you have engaged in.
      • You use comments like:
        • "My agenda is to quantify the protester deaths methodoically. What's yours?"
        • "Iranian casualties are of little interest to maths nerds, who goof off on something less real than actual dead people."
      • That your intentions in authoring this piece were good does not refute the demonstration of your misrepresentation.
      • That you cynically invoke a humanitarian tragedy to cushion your ego does not flatter you.
      • Engaging in misrepresentation is actually a disservice to the cause you purport to serve.
  17. FintanDunne Member

    The UK, Guardian catches up with this story.

  18. How should I read this graph? I thought by looking at it that there was a 40% chance that 200 people died, is that true?
  19. The author has been asked this several times, but studiously avoids providing an explanation of substance. The conclusion seems to be: don't read it. The graph is a drawing; the probability numbers are just there to make the author look smart.
  20. FintanDunne Member

    Suppose that it's my fault for not editing the original post to add stuff
    that has been on the original blog entry for days. I figured people were
    going to the full blog entry to read about it. So I just posted it in a reply
    five posts down the first page of this thread.
    Here:http://iran.whyweprotest.net/40811-post6.html

    Maybe people have not read that or gone to the blog. I've added it to the
    main post too because people may not read down the thread.

    This is from that post and on the blog too:

    That's right. It gets very unlikely that fatalities are below 200 across Iran.

    Simlarly, it's unlikely (40%) that the figure is over 400.

    I don't mind people querying the methodology. I object to attempts to portray
    it as a formal statistical probality function and then attack it. I never said it
    was such. I clearly indicated it was a graph of likelyhood for the lay person:

  21. So, which is it, FintanDunne? Is it a graph of likelihood that the fatalities are below a given amount, or above a given amount? You can't even keep the story straight in those two sentences.

    Pity, then, about your own over-ambitious "methodology" announcement:

    That would be a very ambitious undertaking even if the person in question actually knew what he was talking about, given the limited information available.

    The "lay person", such as yourself? Most lay people don't presume to communicate what they can't even understand.

    Nonsense. Cumulative confidence can not go up and then down. You are confused, as you demonstrate in the two sentences referenced above.

    If the graph were cumulative confidence that the fatality count was below a given number, then the plot would increase going left to right.

    If the graph were cumulative confidence that the fatality count was above a given number, then the plot would decrease going left to right.

    You don't have a clue, FintanDunne. Please stop stating fiction as fact.
  22. FintanDunne Member

    It's a graph showing both.
    Rather obviously.

    Like this:
    Most people are of average intelligence.
    So it's unlikely that a random person you meet will be very dumb or very smart.
    One is well above the average - the well other below.
    Both are unlikely.

    Simple.

    Keep pretending it's complicated.
  23. That graph makes perfect sense to me. There is a 40% chance the causalities are 200, a 90% chance it is 275, and a 40% chance it is 400. What is so hard to figure out about that. It is pretty straight forward.

    The criticisms are based on a lot of assumptions that are not warranted. Its a simple fucking graph. Straight forward, obvious, easy to understand.

    I don't even know why this thread is still alive. They guy keeping it alive is either a trol making up shit to fuck with the OP, or someone trying to impress us all with his big brain, but no one cares.

    Really. Drop it. The causalities are probably a lot higher now.
  24. :) well said-

    either way sooner or later the truth and facts will come out
    and those that have hidden them and abused their positions will be fully accountable
    that is a dead sure certainty
    Love never hides or lies or attacks......

    thanks Fin and the gang
    erin etal
  25. Right on. Martin Luther King said "Not only do the ends not justify the means, but the ends are inherent in the means."

    Deception will give birth to a lie. Do we really want that? I don't. I am not interested in propaganda either.
  26. just mocking the chart, not the deaths

    [IMG]
  27. So the y-axis is both the likelihood that the fatalities are below the x-axis amount, and simultaneously the likelihood that the fatalities are above the x-axis amount? This is clearly false. As an example (none should be needed), this would imply that there is simultaenously a 80% likelihood that there are above 250 fatalities and a 80% likelihood that there are below 250 fatalities. Clear contradiction.

    Let's consider one of your two contradictory sentences:
    This statement suggests that, to read the graph, one follows a particular value on the x-axis (400, in this example) up to the plot (40%), and concludes that there is a 40% likelihood that there are over 400 fatalities. By this reasoning, there is an 80% likelihood that there are over 300 fatalities. Fine so far. But now take the value of 250 fatalities, also 80%. So there is an equal likelihood that there are over 250 fatalities and that there are over 300 fatalities. This suggests that there is 0 likelihood that there are between 250 and 300 fatalities. And it gets worse: take another number: 200. Following the same reasoning, there is a 40% likelihood that there are over 200 fatalities. How can there be an 80% likelihood that there over 300 fatalities yet a 40% likelihood that there are over 200 fatalities? This is a logical contradiction.

    Going back to your contradictory statements:
    we can also imagine a different meaning intended: that you read any value to the left of the peak as the likelihood that fatalities are less than that x-axis value, and any value to the right of the peak as the likelihood that fatalities are more than that x-axis value.

    Two problems with this:
    1. Having the meaning of the y-axis change depending on the x-value makes the graph meaningless. At the least, two y-axes and two plots should be used on the graph.
    2. More importantly, the plot is still self-evidently false. This interpretation of the graph would yield an 80% likelihood of less than 250 fatalities, and an 80% chance of more than 300 fatalities, a clear logical contradiction.

    This is no problem of technicalities or nit-picking: this is a problem of the information shown on the graph being self-contradictory, regardless of the convoluted interpretation you attempt to contrive after the fact. Again, FintanDunne, please stop trying to pass off fiction as fact.
  28. FintanDunne Member

    Jesus H !! lol :) :)

    Sure, dude.
    Anything you say.
  29. You have just demonstrated your ignorance. How can there be a 40% chance of 200 fatalities and a 90% chance of 275? These two likelihoods alone add up to more than 100%. And if we interpret your statement to mean 40% chance of less than 200 fatalities, 90% chance of less than 275 fatalities, and 40% chance of less than 400 fatalities, this is still self-contradictory. How can there be a greater chance that fatalities number less than 275, than the chance that fatalities number less than 400? Another contradiction. And if we interpret your statement to mean a 40% chance of 200 fatalities or less[/], 90% chance of 275 fatalities or less, and 40% chance of 400 fatalities or more, this is still self-contradictory. How can there be 90% chance of 275 fatalities or less, yet a 40% chance of 400 fatalities or more? These probabilites add to more than 100%, a clear contradiction.

    Assumptions that the graph not contradict itself? If you overlooked any of the contradictions mentioned above, the graph is clearly not simple enough for you.

    No, the OP, FintanDunne, posted false data, and you have impressed us with your stunning ignorance.
  30. "Dude," all of the interpretations of the graph that you suggested lead to the conclusion that the information on the graph contradicts itself.

    The graph is demonstrably false. The question remains, why did you, FintanDunne, post false data?
  31. CurtMonash Member

    No, you haven't been clear, at least not up until the point in the thread I'm quoting.

    And by the way, I have a Harvard Ph.D. in game theory, a discipline somewhat related to probability. I also taught a university statistics course when I was 18 years old. So yes, I'm a math nerd.
  32. CurtMonash Member

    That adds up to 170%, which is obviously nonsense, and it gets worse from there.

    The explanation is not clear.
  33. FreedomAgent Member

  34. CurtMonash Member

    FintanDunne,

    There's only one thing I can think of that makes sense and bears a relationship to your graph. Do you mean to say that:

    Probability of {the death total is (EITHER less than the lower point marked X% OR greater than the higher point marked X%)} = X%?


    If so, you are guilty of not saying that clearly. ;) But you are NOT guilty of posting a meaningless graph.

    ----------------------------------------------------

    Also, it would be helpful to see some examples of how you assigned probabilities to specific claimed events. I get that they're very subjective; I'm just interested -- as I'm sure other people would be -- in seeing how your subjective assignment of probabilities works. Have you posted something like that anywhere?
  35. tex-IRAN Member

    Really? Then why doesn't the graph peak at 100% or extremely close to it, rather than 90%? That still doesn't wash.
  36. CurtMonash Member

    Carelessness?

    Or, of course, they may not really be probabilities.

    A little story -- when I was a young stock analyst at PaineWebber in the early 1980s, we hired the market strategy team of Charles Pradilla and Ed Kerschner. They threw around some graphs with "probabilities" that were something like but not exactly like the probability the market would go up. I tried and tried to understand how these were calculated. Explanations weren't clear. Finally, they admitted these weren't really probabilities in the ordinary sense of term.

    But despite that bogus dressing, they were excellent, highly accurate market strategists anyway ...
  37. tex-IRAN Member

    Carelessness? So not only careless enough to misrepresent what the graph conveys, and not only careless enough to pigheadedly refuse to acknowledge the problems with the graph, but careless enough to not even have the plot in the right place on the graph? Not far from my accusations all along, and not at all contradicting my conclusions about FintanDunne making up research.

    Ya think?
  38. tex-IRAN Member

    Oh yeah, not to mention, careless enough to contradict your imaginative explanation himself.

    For example:
  39. dude, just admit you made a mistake and move on. further denying it just destroys your credibility.
  40. AStranger Member

    For anyone who has ever learned statistics, one of its first lessons would be on probability distributions. In this case, the graph shown by the OP obviously violates one of the rules of probability distributions. I don't see why so much arguments and explanations are required. Let's just ask the moderator delete this entire thread and move on.
Thread Status:
Not open for further replies.

Share This Page

Customize Theme Colors

Close

Choose a color via Color picker or click the predefined style names!

Primary Color :

Secondary Color :
Predefined Skins