Prediction Registry

Scheduled Regular Events

Scheduled Unique Events
Unpredictable Events
Other Events Introduction

Unpredictable Events

This category is intended for exploration of possible types of specification for events. The specifications must be made without knowledge of the data, of course, and we will be working toward schemes that allow the definition of criteria to categorize major world events before they happen, to identify some relatively small number where a coherent global reaction or engagement may be expected.

  • RDN, pred. 1998-08-08, while traveling in Halifax: Embassy bombings in Nairobi and Tanzania, 1998-08-07 07:35 UTC. On return home, prior to examining data, I made specific predictions described below. These terrorist attacks exemplify a tearing of the social fabric that would shock a global consciousness temporarily. For this particular event, we had only the general outline for predictions: a period of time, say 10 minutes, surrounding the point event, and a period, say a few hours, following the event during which the world becomes conscious of what has happened.

    An exploratory analysis based on 15-minute data segments from three eggs looked at the event-period from 07:15 to 07:45, and a three-hour consciousness-spreading period from 07:15 to 10:00. The associated probabilities indicate significant deviations for both time-periods.

    Hand calculation: >Significant deviations. Event: Chi-square 18.039, 9 df, p=0.035; Aftermath: Chi-square 69.536, 36 df, p=0.00066.

  • RDN, pred. 1998-08-22, no prior examination of data: Omagh tragedy, bombing in Northern Ireland, 1998-08-15 15:10 UTC (Irish Newspaper report times). Beginning at 14:30 with the terrorists' deceptive call warning of bomb in wrong place, people moved to the target area. We will examine the 45 min. from 14:45 to 15:30, which includes the time of the explosion at 15:10. Then also the three-hour period from 14:30 to 19:30, during which the news spread around the globe and the people in Omagh searched for loved ones and sought to rescue the survivors.

    Hand calculation: No significant deviations. Event: Chi-square 7.395, 12 df, p=0.830; Aftermath: Chi-square 39.469, 48 df, p=0.805.

  • RDN, pred. 1998-08-22, no prior examination of data: US Air strikes on the training camp in Afghanistan for terrorists, including those thought to be responsible for the African embassy bombings, and VX-gas factory in Sudan. 1998-08-20 17:30 (?) (CNN reports 13:30 EDT) simultaneous attacks. No prominent effect predicted, because this event, though widely known, does not engage people in a unified way. We will assess the period around the event, from 17:15 to 17:45, centered on the time of attack.

    Hand calculation: No significant deviations. Event: Chi-square 14.115, 12 df, p=0.235.

  • RDN, pred. 1998-09-04, no prior examination of data: Swissair Flight 111 crash. Time of crash approximately 03:20, Central European Time, September 3. I think that is 02:20, UTC, but must check. The prediction is for an event-spike around the time, and will specify the three 15-minute blocks with the crash time in middle block: 01:00 - 01:45. Then an aftermath period of three hours beginning with the same time. There is a two hour difference because of Summer time, so the event time is 01:20 UTC.

    Hand calculation: Significant or marginal deviations. Event: Chi-square 26.420, 15 df, p=0.034. Aftermath: Chi-square 75.027, 60 df, p=0.092.

  • RDN, pred. 1998-09-08, no prior examination of data: American event: Mark McGwire hits homerun #62, for new all-time record number in one season, eclipsing the great Babe Ruth and later Roger Maris. I was with Charles and Steven (visiting from LA) in Conte's having a pizza. The HR was hit at 98-09-08 21:18, and I think Steven even glanced up to see it on the tube. This time is EDT, which is, I believe, 4 hours from UTC. From the data from 01:00 to 01:30 on 98-09-09:

    Hand calculation: No significant deviation. Event: Chi-square 22.040, 18 df, p = 0.230. (The US data alone show 11.595, 9 df, p = 0.237).

  • RDN, pred. 1998-11-02, no prior examination of data: Note: the time estimates originally made from media reports were wrong, and this prediction now uses the corrected information from the USGS. An avalanche and "Lahar runout flow" at the Casitas volcano in Nicaragua during the onslaught of hurricane Mitch apparently killed thousands of people; current estimates suggest 2000 or more. The collapse occurred on Friday, October 30, beginning with an avalanche at a time between 10:30 and 11:00 am. The main damage was from a huge mudslide caused by the combination of the extreme rainfall and the avalanche, which occurred at 14:00. I will predict a period of deviation between 10:30 and 11:00, and another beginning at 14:00 and continuing for 3 hours until 17:00. The UTC times are 981030 1530 to 1600 and 981030 1900 to 2200.

    Hand calculation: No significant deviations. For the 30-minute period of the avalanche, chi-square = 5.40, p = 0.86. For the 3-hour period following the mudslide at 14:00 local time, chi-square = 66.20, df = 60, and p = 0.27.

  • RDN, pred. 1998-11-29, no prior examination of data: A few days ago, there was a period when we did not know what might happen in Iraq. The US was ready to throw bombs at a problem created when Saddam Hussein refused to cooperate with UN inspectors. Although most of us went about our business, I think there was a deep concern, an emotional unrest, strongly felt opinions, a very broad, even if confused and weakly directed attention to the imminence of a horrible rent in the social fabric. Though the decisions that ultimately were made seemed in the hands of very few, it is clear that very many were interested. Almost all of this interest was not personal but global, not self-directed, but integral to the group intelligence of humanity. I have not looked at the EGG data to see whether there is anything to see, but this is an example of a situation where I think a prediction of a composite deviation in the data, suitably circumscribed, is warranted. My own feeling is that the global engagement became very strong over a period of a few hours and that it would have peaked very near the point in time when the decision was made by Saddam to cooperate and the US not to bomb. I invite anyone who is interested to submit a precise formulation for such a prediction, to be tested against the extant data. This is an event of the sort I intend to specify as an instantiation of the GCP hypothesis, and I do have a general plan for making the prediction, but let us see what others think. (In August, 1999 I was finally able to do the analysis, with somewhat imprecise specification of the time, due to the long lapse. News report BBC at 21:44 GMT says the crisis is over, so arbitrarily I decided on 19:00 to 22:00 as the "11th hour" period. Given the imprecision, this prediction might properly be withdrawn from the list, but the outcome is essentially null, (Chisq 46.94 on 48 df, p = 0.516) so it would have little bearing either way.)

  • RDN, pred. 1998-12-19, no prior examination of data: The US and Britain proceed to bomb Iraq beginning on Wednesday the 16th, at 22:00 UTC (01:00 Thursday the 17th in Iraq). I predict that the hour surrounding the actual beginning of bombing will show deviations (21:30 to 22:30 UTC).

  • RDN, pred. 1998-12-19, no prior examination of data: Today, the House of Representatives voted articles of impeachment, after some days of debate. It is the second impeachment in US history, and is extraordinary in many ways. Certainly it is momentous, and even though confusing and difficult to understand, there should be some resonance simply because it is an important historical event. I predict a composite deviation increasing during the morning and peaking at about 13:30 (13:24 announced) when the vote approving the first article was complete. Specifically, I predict that the three hours 12:00 to 15:00, centered on that time will show significant deviation. (17:00 to 20:00 UTC)

    Hand calculation: A non-significant deviation. For the three-hour period, chi-square = 71.9, 84 df, p = 0.824.

  • RDN, pred. 1999-01-12, [Note: the correct date is 1999-02-12, RDN] no prior examination of data: When the Senate votes, there will be a moderate deviation in the hour following the conclusion. Whether the vote is for acquittal or impeachment, attention will be focused and a relief that it is over will unify most of us.

    Notes on the analysis parameters, made in response to question: The registry has a mistake in the date -- it should be 1999-02-12. The one-hour event goes from 12:30 to 13:30 Washington time, which is 17:30 to 18:30 GMT. There are 10 eggs, so 40 15-min periods, and hence 40df.

    Hand calculation: A non-significant deviation. For the one-hour period, chi-square = 41.20, 40 df p = 0.417.

  • RDN, pred. 1999-02-24, no prior examination of data: Earthquake in Columbia, centered on Armenia, 140 mi from Bogota. Monday 1999-01-25, Early afternoon, apparently at 13:19 or 13:20, which is 18:20 UTC. Predict from hour before to three after the event. 15 min blocks.

    Hand calculation: A non-significant deviation. For the four-hour period, chi-square = 157.724, 144 df, p = 0.205.

  • Dr. G, pred. 1999-03-11, Dow Jones will go over 10,000 today or tomorrow. RDN suggests predicting the EGG will show some reaction, but Dr. G is dubious. We will look, however, at the event one hour. Predict increase variance in the hour before, flat in the hour after. Analyze by 15 min blocks.

    Note: This event was not analyzed at the time, and the oversight was only noted in October 2002, when Eckhard Etzold reported it. The analysis was done on October 26 and included in the formal Results table as event number 24. The prediction was for a two hour period centered on the market closing time, which was 21:00 GMT. The date was 990329, and there were 12 eggs running, times 4x2=8 15-min periods, for 96df. The analysis period was 20:00 to 22:00 GMT.

  • RDN, pred. 1999-03-24, no prior examination of data: NATO attacked Yugoslavia beginning at about 14:00 EST, 19:00 UTC. I predict that the hour from the actual beginning of bombing will show deviations (19:00 to 20:00 UTC).
    Hand calculation: A nominally significant deviation. For the one-hour period of first attacks, chi-square = 65.72, 48 df, p = 0.045.

  • Hollis Polk: prediction on 04-22, no examination of data. The shooting spree in Littleton, Colorado, where two students killed many of their fellows and a teacher, a total of 15 dead as of this date may be a global event because it is so unexpected, unexplainable, and tragic. HP suggests the time period from 1:30PM - 4:30PM EDT on 4/20/99. "I think they said that it began around 11:30 mountain time and went on for about 3 hours. Though more and more people heard about it as time went on, so if the effects get stronger toward the end of the time (can you do that?), it might be worth looking at 24 hours beginning 1:30PM EDT." Formal prediction is 17:30 to 20:30, UTC on 99-04-20. After hearing news, it appears that the tragedy began earlier, with the first calls coming in at 11.
    Hand calculation: The data are strikingly flat, i.e., the deviation is much less than expected. For the period predicted by HP, Chisquare is 118.067, on 132 df, p = 0.801 (0.20); for the five hour period from 10 am to 3 pm, Chisquare is 178.704, on 220 df, p = 0.981 (0.019).
  • RDN, pred. 1999-06-24, no prior examination of data: The war in Yugoslavia began to wind down after 10 weeks of bombing. No exact time could be established for its end, but my criterion was that Milosevic must make a public declaration. This happened on 10 June, 1999, with the first report I could find at 15:00 UTC, from the BBC Online Network, a service of BBC News. I also found an article published at 05:22, UTC that described world media waiting to be able to use headlines saying "Peace at last". For the former, the prediction was made for a three-hour period, beginning an hour before the announcement and continuing for two hours after (1400 to 1700, UTC). For the latter, the predicted period was 0500 to 0800, UTC. Both were designated as formal predictions. For the Milosevic announcement, the data for 17 eggs over the three-hour period (15-minute blocks) showed Chisquare = 239.84, 204 df, p = 0.042. For the "Peace at last" headline story, the three hour period (for 16 eggs with intact data; a 17th egg could not be included) showed Chisquare = 198.16, 192 df, 192 df, p = 0.365.

  • Prediction, RDN: John F. Kennedy, Jr. with wife and sister-in-law, apparently crash small plane on a flight to Martha's Vineyard from northern New Jersey. The flight began at 8:30 pm, July 16, EDT, and the crash probably occurred about 9:30. JFK Jr. has been a much loved center of public attention, certainly in the US. This will be a long term "event" depending on the search process, and there very likely will be a major ceremony of memorium. The first prediction is that the three hours beginning at 9:30 (01:30 UTC, on 17 July) will show deviation. This acknowledges a possibility that some ripples of effect are created at the time of an event, before any large numbers of people are conscious of it. The time of the crash, plus and minus 15 minutes will be assessed as a possible reflection of an "earth consciousness". We also will explore data variations possibly correlated with the developing reactions of the public (especially in the US). Analyze by 15 min blocks, medium effect predicted for the two formal predictions.

  • Prediction, RDN: India Train Crash, 990801, 20:00 UT. Prediction made after George deBeaumont mentioned there had been a terrible traincrash in India, with apparent deviations for several hours after. (Given this prior information, my predictions cannot be included in the formal database, but the analysis is useful exploration.) The collision happened at 0130 local time (2000 GMT) when an express train bound for Delhi hit another train head-on at Gaisal station, in the state of West Bengal. At least 300 are believed dead, and officials fear the toll may be as high as 500. I am putting a formal prediction into the registry prior to my examination of the data and with no detailed information from George, because I want to have some representation of events that are not centric to the US and Europe. The prediction is that the four-hour period beginning at the time of the crash will show deviation (this is modeled on the Embassy and Casitas events). The time period from 990801 20:00 to 23:59 will be analysed in 15-minute blocks, medium effect (would be high, but compensating for prior knowledge). Results, preliminary (only 16 of 17 eggs could be included): Chisquare 302.94 on 256 df, p = 0.023. When the geographically proximate eggs are selected: India, Fiji, New Zealand, and North American west coast, the effect is larger, with Chisquare 127.52, 96 df, p = 0.017.
  • Prediction, RDN: Earthquake in Turkey, near Istanbul. Magnitude 7.8, and terrible destruction with many hundreds of lives lost, especially in the suburban, poorer areas. I am making two predictions, based on models of Embassy bombing, previous natural disasters, and the India train crash. For the globally spreading knowledge I think a longish aftermath is necessary, so am going to predict a four-hour period, from 03:00 Turkish local time, which I think is UTC+3 This means from 00:00:00 to 03:59:59 UTC, on 17 August 1999, 15-minute blocks. (The UTC/Turkey correspondence needs to be checked for accuracy.) I also want to make a prediction that is for the immediate earth-consciousness, a half hour from 23:45:00 on 16 08 to 00:14:59, 17 08, centered on the time the earthquake hit at 03:02 local.

  • Japanese Nuclear Accident, Predictor, RDN. On Thursday, September 30, 1999, according to a BBC article published at 22:40 UTC, a major nuclear accident released large quantities of radioactive materials at a processing plant in Japan, causing some injuries at the plant, and evacuations of the surrounding area. A GCP prediction of high effect was made for the time surrounding the actual event. The leak was detected at 10:35 local time (0135 UTC). The prediction is for the period from 10 to 11 (0100 to 0200 UTC), with 15-minute blocking.

  • Typhoon in India, October 29, 1999, Predictor, RDN. From "Giant cyclone slams into India's coast. BHUBANESWAR, India (AP) -- A powerful cyclone, the second in two weeks, slammed into India's eastern coast today as authorities packed villagers into trucks in an effort to evacuate 20,000 people. The cyclone bearing winds up to 155 mph hit along 300 miles of coastline in the states of Orissa and West Bengal at midday today (29 October)." This is a large-scale and continuing disaster, and it is difficult to identify a corresponding, simple prediction. We will make two predictions which are somewhat exploratory in nature. An earth consciousness prediction addresses the beginning of the event with a prediction of two hours surrounding midday, to be analysed in 15-minute blocks. The GMT times are 04:30 to 06:30, 29 October, 1999. A second prediction addresses the extended development of the disaster for 24 hours, with one-hour blocks, from 04:30, 29 October to 04:30, 30 October, 1999. Both predictions are formal, medium.
  • Prediction, Brenda Dunne: November 12 earthquake in Duzce, Turkey, 170 Km east of Istanbul. The magnitude 7.2 quake followed a 5.7 event on the previous day. Widespread destruction, and 450 or more dead. The time was 18:59 local (16:59 UTC). The prediction is more complex than usual, namely, positive deviations from expectation in the two hours preceding the quake (14:59:00 to 16:58:59 UTC), and negative deviations of Z^2 from expectation during two hours following the quake (16:59:00 to 18:58:59 UTC). The blocksize for analysis is 15 minutes, and the expectation category is medium.

  • Prediction, Reinhilde Nelson: July 25th, 2000 at 15:02 GMT. The Concorde crash upon takeoff from Paris will engage more attention than usual because the Concord is a high-profile plane, beautiful and symbolic of the romance of flying. It is the first crash of a Concorde, which has been considered among the savest planes since 1969. This was a chartered flight with mostly German tourists on their way to a cruise. All 100 passengers and 9 crew were killed, as were four people on the ground. Analysis will be of the half hour following the crash and a four hour aftermath, with the formal result being the half hour period, assessed at one-second resolution, with the 4 hour period to address developing awareness of the news. Expectation high. Secondary analysis will determine whether European eggs show greater reaction.

  • Prediction, Roger Nelson: The Kursk submarine accident. We will look at the half hour beginning with the first explosion using raw, 1-second data, and then the following days until the resolution (access and confirmation that all are indeed dead) in 1-minute blocks. Expectation high.
  • Prediction, Roger Nelson: The 7.6 magnitude Earthquake in Central America on January 13 2001 captures wide attention, partly because hundreds, probably more than 1000, died. El Salvador was hardest his, with 45000 houses destroyed. Though I was traveling in Vienna, and little in touch with the news, I heard about it immediately. The formal prediction is based on the Turkey event in August 1999, which was of similar magnitude. The time was 17:33:29 UTC, and the prediction is for the half-hour surrounding this moment. This reflects the "earth consciousness" notion that something might already be happening before the actual event is felt. Analysis will be using seconds resolution, and expectation level is medium.
  • Prediction, Roger Nelson: (Written on Sept 12, after some preliminary examination of the data recorded on this frightening day. I was distracted but quite clear that this was formally a GCP event, and my prediction was not based on the early analysis.)

    On September 11, 2001, beginning at about 8:45 in the morning, a series of terrorist attacks destroyed the twin towers of the World Trade Center and severely damaged the Pentagon. Commercial airliners were hijacked and flown directly into the three buildings. The first crashed into the North tower at 8:45, and about 18 minutes later the second airliner hit the second tower. At about 9:40, a third airliner crashed into the Pentagon. At about 9:58, the South tower collapsed, followed by the North tower at 10:28.

    The formal prediction for this event was not registered before any analysis, but because it is formulated on the basis established for the terrorist bombing in Africa in August 1998, there is no possibility of data selection based on prior examination of the data. The 1998 prediction specified a period "beginning a few minutes before the bombing, and including an aftermath of a few hours." The actual time was from 10 minutes before the bombing to three hours after. In this case we will specify 10 minutes before the first crash to four hours after, which makes the aftermath following the last of the major events, the collapse of the second tower, about the same as the period in 1998. The confidence level is high, and the resolution is seconds.

    Dean Radin made a prediction that the variance would show strong fluctuations: "I'd predict something like ripples of high and low variance, as the emotional shocks continue to reverberate for days and weeks." Although this is not sufficiently specific to be included in the formal database, it is effectively a prediction that the variance around the time of the disaster should deviate from expectation. At least I wish to make that more specific version of the prediction, with medium confidence, seconds resolution.

    Doug Mast made a specific formal prediction for a deviation of the Chisquare "over the time periods 1000 to 1003 GMT, corresponding to a European organized mourning ( and the time period 1200 to 1203 EDT (1600 to 1603 GMT) corresponding to the beginning of the Washington service and many organized mourning events in the Eastern US." He specified seconds resolution and medium confidence.

    Note added Sept 21: At the outset, I did not know how to calculate the probability of the huge changes that appear in the graph of cumulative deviation of variance I generated as a test of Dean Radin's prediction. In fact the extreme deviations of the data conform precisely to the qualitative description Dean first made, although the "ripples of high and low variance" are actually more like the rise and fall of a great wave. Given the powerful indications from all of the contextual analyses of the September 11 events showing that it elicited a uniquely powerful response from the EGG network, I am going to assign it a probability of 0.001, based on the result found in Dean's independent finding of Sigma ~ 3.5 for the same data processed via his sum of Z². This is at the conservative end of the range seen in the explorations that have been made. These contextual analyses are accessible via links from the primary analyses.

  • Prediction, Roger Nelson:
    Date: Mon, 8 Oct 2001 12:27:01 -0400 (EDT)
    From: rdnelson 
    To: Roger D. Nelson 
    Subject: prediction
    On October 7 2001, apparently at 12:30 EDT, the US began attacks
    on Afghani sites, followed at about 13:00 by food/medicine
    drops.  I am traveling, and have no access to the prediction
    registry, so am writing this note from the public library in
    North Platte NE to establish the parameters.  The prediction
    is for a period beginning at 12:00 and continuing for 8
    hours (1600 to 2400 GMT) seconds resolution, medium
    confidence, standard analysis.
  • Prediction, Joseph Menahem:

    Note added June 26 2004: For consistency, since this is the only instance of a single egg being used, we have decided to use the standard full-network specification for the re-analysis being done by Peter Bancel in 2004. This changes the result somewhat, to a stronger outcome.

    On Wednesday October 17, 2001 An Israeli minister was shot dead in a Jerusalem hotel today by a suspected Palestinian gunman, prompting fresh fears for the shaky truce between Palestinians and Israelis that was agreed only three weeks ago. The Syria-based Popular Front for the Liberation of Palestine immediately claimed responsibility for murder of the far-right tourism minister, Rehavam Ze'evi. "It created a big emotional reaction here."

    The local time in Israel is 2 hours ahead of GMT.  The
    assassination occurred on the 17th at about 7:00 a.m. 
    I think the first major news came out at 8:00 a.m. --
    at least I heard it the first time.  The final and
    formal announcement about the minister's death came at
    10:00 a.m.  
    The second major related event in my opinion is the
    funeral which took place on the next day and which was
    broadcast live on the T.V. and radio from the
    morning till late afternoon.  Huge number of crowd
    came to visit the coffin on the parliament yard from
    9:00 a.m. (18th) till about 12:00, where also some
    significant speeches took place and then on the
    cemetary his son called directly to the prime minister
    who attended there to do what he has to do
    (interpreted as a call for revenge - 4:00-5:30 p.m.). 
    Joseph Menahem lives in Israel. He specified deviations in data from the egg he hosts. The formal prediction is for a deviation in the local egg's data beginning at 07:00 and continuing to 10:00 (05:00 to 08:00 GMT). Standard analysis (for one egg), seconds resolution, medium confidence. We will also do an exploratory analysis of the funeral event, from 07:00 to 10:00 on the 18th, and we will explore the full network reaction to the assassination events.
  • Prediction, Mon, 12 Nov 2001:
    Nick Reiter wrote to ask if I could explain what was happening on the real time display of egg data, which looked unusual to him. He said that he was watching due to the unfolding tragedy of the plane crash in Queens. I responded that I intended to make a formal prediction for the event, and asked if he would be so kind as to get the real times for the accident and a reasonable guess about the end of the high concern period (while people were thinking O God, another attack). I proposed a period lasting for two or three hours, or however long it had taken for people to become fairly certain it was an accident.

    The newswire and media references cited the crash as happening at 0917 EST. Rick said his own guess was that the period of highest anxiety would extend from about 0945 EST to perhaps 1500 EST. By that time, it seemed like there were quite a few references to the FAA and FBI stating that it was unlikely that a terrorist bomb was involved. I told Rick I heard reference while I was driving home at about 13:30 to increasing confidence that it was an accident, and asked if he "would be comfortable with a period from 09:00 to noon? Or do you feel it should go on to 15:00?" He said 0900 would be a good starting point, but that we might want to extend the period of prediction / analysis out to 1300, just to capture what may have been the people tuning in at lunch to hear something hopeful.

    We agreed that the formal prediction would be from 0900 to 1300 EST (1400 to 1800 UTC). Resolution seconds, medium confidence. Explorations may be done to see if there is a longer aftermath.

  • Prediction, Sun, 9 Dec 2001. From email, Peter French, NZ:

    In case it is significant to GCP research activities ... on Friday, 7 December 2001, "New Zealand awoke" (I first heard it on the radio at 6.00am, Friday, 7 December) to the "shock news" that a famous "New Zealander", Sir Peter Blake, had been killed.

    He was perhaps particularly well-known in our largest city Auckland (for his yachting activities there) - Auckland being a city where two REGs operate. So, it is possible that the output of those REGs has been affected.

    The timing of the incident is not yet clear: A london paper reports on Thursday, 6 December, 2001
    Emsworth (England).- Sir Peter Blake, K.B.E., was shot and killed by armed intruders who boarded the "blakexpeditions" vessel Seamaster, anchored off Macapa at the mouth of the River Amazon, early this morning (UK time).

    The group of seven or eight armed and hooded intruders boarded Seamaster at approximately 10. 15pm local time, in Macapa at the mouth of the Amazon.

    Sir Peter was aboard his Seamaster vessel in the Amazon river mouth when it was attacked by river pirates. The boat was in the region carrying out conservation and scientific research. Brazilian police say the bandits slipped on to rob the vessel, and stole watches, cameras, and an inflatable dinghy.

    National MP Murray McCully, who is the former America's Cup Minister, has described Sir Peter as a beacon who stood out in a world of grey murk. He says Sir Peter was a man who made a remarkable impact on the lives of New Zealanders and others around the world. Conservation Minister Sandra Lee says Sir Peter's untimely and tragic death is hard to comprehend. She says there are times in our history when we awaken to a horrible nightmare, and the tragic murder of Sir Peter will always be remembered as such.

    I think the time of the murder was 22:15 on the 6th in East Brazil time zone, which is 00:15 on the 7th GMT, and morning on the 7th in New Zealand. We will set the prediction time as 00:00 to 04:00, on 7 Dec, GMT. Seconds resolution, medium confidence.

    On May 6 2002, I received a note from Faeria French saying the date was in error, "I now think that the time was 22.15 on the 5th (not the 6th)." This will become the time of the formal analysis.

    Note: This was analysed but inadvertently was not included in the formal Results table until the error was noted by Eckhard Etzold in October 2002. It was then included as event #96.

  • Prediction, 19 Jan 2002. With help from Luke Ouko, Nairobi.

    Date: Sun, 20 Jan 2002 16:36:43 
    From: Luke Ouko 
    Subject: Volcanic Eruption ! 
    Hi, The volcanic eruption was big news. More so for me because I have worked 
    in Goma before.  I will probably go there some time during the course of the 
    week to help with some volunteer work. I will get you some links you can look 
    at to get a better picture. It is unfortunate that big media organizations 
    like CNN do not seem to have found much to report.
    Date: Mon, 21 Jan 2002 05:05:19
    From: Luke Ouko 
    Subject: Re: Volcanic Eruption !
    Nyiragongo Volcano(Democratic Republic of Congo)
    1.52 S, 29.25 E.
    Nyiragongo volcano in eastern Congo erupted early Thursday morning (17th
    January), sending out plumes of ash and three rivers of lava that destroyed
    14 villages near the Rwandan border and drove thousands from their homes.
    There have been 47 fatalities so far. Injuries have been caused by lava,
    firearms, and hit by vehicles leaving the area. The risk of disease and
    malnutrition may cause many more deaths in the coming weeks.
    Mount Nyiragongo is just 10 kilometres (six miles) north of Goma, a city 
    of 350,000 people. With as many as 400,000 persons potentially homeless,
    the shelter needs will be considerable.
    The earliest news to an international audience was not available until
    some time on Thursday night off BBC mainly. Most other media only picked up
    on the story on Saturday.
    Date: Tue, 22 Jan 2002 15:50:27
    From: Luke Ouko 
    Subject: Re: Volcanic Eruption !
    Hi Roger,
    I am sorry that this information is coming so late. I spent a lot of time
    trying to sort out the staff I needed to prepare for my trip to the DRC.
    The times of the start of the disaster vary between 4.55 AM and 5.05 AM
    depending on who you talk to. But I guess this should be close enough of
    an estimate.

    The prediction will specify a short time before, and an aftermath period of a few hours, for a total of four hours. Specifically, 4:45 to 8:45, local time. Nairobi is 3 hours East, and Goma is probably in the same timezone, so the UTC interval for the formal analysis is 01:45 to 05:45. For exploration, we will also look at the whole UTC day, expecting to see some general effect of the spreading news and concern. Seconds, Medium.

  • Prediction, Jaan Suurkula, 1 Feb, 2002, concerning the death of Astrid Lindgren. With spcification by Roger Nelson.

    On Monday Jan 28,  before noon, the famous Swedish child book
    author Astrid Lindgren died.
    She is perhaps most known for "Pippi Longstocking" but has altogether
    written over 80 books, many of which were bestsellers. She was one of the
    12 most read authors in the world and was translated to 80 languages on
    all continents and won many prizes. She was very much loved
    by former and present children on all contintents.

    She apparently died before noon local time (GMT+1), on 28 January 2002. The announcements to the media came between 12 and 1pm, and we will set the length of the prediction as four hours. The formal prediction will be for 11:00 to 15:00 GMT, covering the time a little before she died to a few hours after, during which the first news spread around the world. Standard Stouffer Z analysis, medium, seconds, all eggs.

  • Prediction, Roger Nelson: Earthquake in Afghanistan, March 25 2002

    Source USGS NEIC 
    Date and time 2002 03 25 14:56:37 UTC 
    Depth 33.0 kilometers Magnitude 6.1 
    Location 35.97N 69.18E, HINDU KUSH REGION, AFGHANISTAN 
    Reference 100 miles (160 km) N of KABUL, Afghanistan 

    The original reports on the USGS website indicated from 800 to 1000 people were killed, 4,000 injured, 1,500 houses destroyed in the Nahrin area. Approximately 20,000 people homeless. Felt strongly in much of northern Afghanistan. Also felt in the Islamabad-Peshawar area, Pakistan and Dushanbe, Tajikstan.

    Of course this event in the context of the war and distress in Afghanistan connected with the terrorist attacks in September 2001 became a focus world-wide attention. Prediction is the standard for natural disasters, with a period prior to the event and an aftermath of a few hours. Specifically, the analysis period will begin half an hour prior to the main tremblor, and continue for 3.5 hours after for a total of 4 hours. The time is 14:26 to 18:26 UTC. Standard method, medium, seconds.

  • Prediction, Roger Nelson: Passover Bombing, March 27, 2002

    For months I and others have wondered how a possible Global Consciousness might be responding to the horrible ongoing violence and intransigence in the middle east, with Palestinian suicide bombers and Israeli reprisals killing innocents, and destroying hopes for a peace in the region. But how to identify a "global moment" in this unending turmoil. Finally I decided to take the Passover bombing as a representation of the situation. I have not looked at the data.

    According to news reports, 28 people were killed and 140 injured, many critically, in the attack. It was described as the worst atrocity since the outbreak of the current conflict in Septemper 2000 in terms of carnage and its symbolic circumstances, coming on the first night of Pessah at a communal Sedar.

    The bomb was detonated at about 7:15pm (19:15) in the dining room of a seaside hotel in Netanya, Israel. Israel is 2 hours ahead of GMT so the corresponding GMT time was 17:15. We will take the usual period for disasters comprising a short time before to a few hours after the event. The formal prediction is for 17:00 to 21:00 GMT, standard analysis, seconds resolution, medium expectation.

    Hala Azzam stimulated this prediction, and later suggested an element reflecting the Israeli response, namely, "... April 3rd, at 2:00 am when the Israeli troops entered the Jenin Camp and the days before and after that." For a concise period, I decided to use the Jenin incursion time plus four hours, with the same parameters as the Netanya boming prediction. Thus the prediction period is 00:00 to 04:00 April 3, 2002.

  • Prediction, Roger Nelson: School shooting, Erfurt, Germany, April 26 2002, based on an early notice given me by Eckhard Etzold.

    Sadly, the infection of random violence in schools is not confined to the United States. On 26 April, at about 11:00, an angry student began shooting people and before killing himself, he had killed 18 teachers, staff, and fellow students, and torn a substantial rent in the social fabric of trust.

    The formal prediction is for a period from 11:00 in Germany which is 09:00 GMT, extending for three hours (modeling on the Littleton tragedy) to 12:00 GMT. Standard analysis, seconds, medium.

  • Prediction, Roger Nelson: Bombing in Bali on 12 October 2002.

    Late in the evening of 12 October, at 10:30 local time, two bombs exploded in the Kuta Beach resort area of Bali. The first was small, and caused people consternation, but the second was a large explosion that destroyed several buildings and set others on fire. Nearly 200 people died, either from the explosions or in the collapsing buildings and fires. Most were young people enjoying music and dancing at clubs, and many were foreign, westerners from Australia, Europe, and the US. At the time of this writing, no perpetrators are known, but suspicion falls on the terrorist network of al-Quaida, and more generally on Muslim extremists. The event has been referred to as Indonesia's 9/11.

    The formal prediction will specify a two and 1/2 hour period beginning at 10:00. This comprises a pre-event period of 30 minutes, and an aftermath period of two hours, similar to other terrorist attack predictions. 22:30 in Jakarta is 15:30 UTC on the same day. The prediction period is thus 15:00 to 17:30, UTC on 12 October. Standard analysis (and when possible also the variance analysis as additional exploration), seconds resolution, high confidence.

  • Prediction, Roger Nelson: The GCP will show significant structure if the US government attacks Iraq in a pre-emptive war.

    This prediction, as written on Sept 21 2002 does not have a date, and it is a prediction I would rather not test because I deeply believe that the Bush Whitehouse is making a tragic mistake by pressing for war with Iraq. However, Bush himself seems determined to the point of obsession to make war and at this time it is all too likely.

    In the eventuality, we will apply all three of the major analyses developed in the context of the 9/11 attacks. 1) cumulative correlated meanshift, 2) cumulative device variance, 3) long-lag autocorrelation. Number 1 will be tested within a window of four hours beginning one hour before the attack. Number 2 will use the context of 24 hours for a 1000-iteration permutation analysis (10K or more if necessary for precision) of the maximum absolute excursion. Number 3 will test lags up to 2 hours in a 24-hour context. In addition to the formal assessments, we will explore, and will report the explorations as indicated by their salience. Seconds resolution, standardized analytical recipes.

    May we never be able to test these predictions.

  • Prediction, Oleg Avchenko, first suggestion on October 25:

    "Dear Roger! Now we have probably maximum (pike) of tragedy."

    The hostage crisis in Russia where Chechen rebels were holding 800 people in a theater was violently resolved when government officials stormed the building after pumping poison gas into it. Sadly, about 120 of the hostages died from the effects of the gas, and another 150 are described as in critical condition. The rebels killed only two people, and all 50 of them were killed in the attack. Oleg provided timing details:

    Dear Roger! The big request to make the analysis of data EGGs
    for October, 26 2002  during time from 3.00 till 5.00 GMT. Storm of 
    a building of theatre began at 3.30 GMT and it proceeded approximately 
    1.00. After storm, near 5.00 GMT the TV has informed about clearing 
    hostages. The peak or a maximum of tragedy, in my opinion was near  
    4.00 GMT.

    An hour or two later, Oleg wrote to apologize for a mistake, and indicated that the storming of the building actually began earlier, at about 02:30. The formal prediction is for deviation during the period from 02:30

  • Prediction, Roger Nelson, noted on October 25, written on 31st, prior to accessing the data, concerning the Wellstone crash.

    Along with vast numbers of other Americans, including people on both sides of the political spectrum, I was deeply saddened by the news that Senator Paul Wellstone, the passionately liberal Democrat whose re-election campaign was vital to control of the Senate, was killed in a plane crash in northern Minnesota on Friday along with his wife, daughter and five others.

    The twin-engine private plane went down about 10 a.m. in freezing rain and light snow near the Eveleth-Virginia Municipal Airport, about 175 miles north of Minneapolis.

    After worrying about the US-centric nature of the event, I decided to make a formal prediction after all, because there is a poignancy about the event that I think is one of the elements of a "global event". The crash occurred at about 10:00 Minnesota time, and I will make the event duration two hours, from 15:00 to 17:00 GMT, on 25 October, 2002. Standard analysis and medium confidence.

  • Prediction for November 28, Hotel bombing and missile attempt in Kenya. Al Pasternak suggested this as a global event that would raise the world tensions and stress level, in a note on Dec 11. There has been no prior examination of the data.

    Thanks for the information, Al.  I will add a formal prediction attributed to you 
    to the registry for deviations in our standard analysis of the data on 28 November 
    2002 from 0500 GMT to 0800 This will cover both incidents, and give a half hour 
    prior and 2.5 hours aftermath for the bombing, which is similar to other terrorist 
    attack predictions.  I appreciate your suggestion for this to be a formal event, 
    and doubly appreciate the careful identification of the timing.  -- Roger
    On Wed, 11 Dec 2002, Al Pasternak wrote:
    > > At approximately 08:00 on Thursday the 28th November 2002 two heat seeking 
    > > missiles were shot at Flight No. 582 taking off from Mombassa Airport in Kenya.
    > >  Officials said at about 8:30 a.m. local time a vehicle pulled up to
    > >  the entrance of the hotel and was refused entry. The men reversed
    > >  huge explosion that tore through the building.
    > I confirmed this from two different sources. 0830 Kenya time is three hours 
    > ahead of GMT thus the time of the attack at the Paradise Hotel was 0530GMT
    > Al Pasternak

  • Prediction for the suicide bombing in Grozny, Chechnya on 27 December, 2002. This is yet another horrifying incident in a series showing that we have allowed an idea to take root that suicide bombings aimed at killing and maiming civilians are acceptable. This one was apparently the work of the Chechnyan separatists, who loaded trucks with thousands of pounds of explosives, and rammed them through the security around the pro-Russian government building at about 2:30 in the afternoon. There were about 200 people working there, and some 55 of them were killed and most of the rest badly hurt.

    The prediction is for three hours from 11:30 to 14:30 (Grozny is three hours earlier than GMT). Standard analysis, medium confidence.

  • Prediction for homocide/suicide bombing in Tel Aviv Market on Jan 5 2003. An Israeli journalist named Gadi Guy, doing an interview for a Hebrew web magazine called "Weekend", asked me whether we had looked at the violence in Israel. I said yes, but only a sampling of events. We agreed that he should provide information for an event he thought likely to have an effect on the GCP network. He sent the following as one of his interview questions:

    5. On January 5, 2003, at approximately 17:30pm GMT, a bomb exploded in a crowded market in Tel Aviv, killing more than 20 people and injuring over 70, causing widespread fear and anger all over the country. Has this event registered on any of your sensors, especially the one in Israel?

    The formal specification mirrors those for previous, similar events. We agreed that I would explore more broadly, in particular look at the Israeli egg, but the formal prediction would be for three hours beginning at 17:00, half an hour before the explosion. Standard analysis, seconds resolution.

  • The Columbia shuttle disaster on Feb 1 2003, though most poignant to Americans, is a global event. Some of the astronauts were from other countries; one was born in India and another was the first Israeli in space. People around the world feel sympathy and loss.

    The shuttle was scheduled to land in Florida at 0916 EST (1416 GMT), but a few minutes before 09:00 problems with wing sensors were reported, and at 09:00 the last communication was received. Within a few minutes, residents of north Texas reported loud "bangs", and vapor trails. The news was pretty immediate, and though people expressed hope, there was little, and then about 9:30 or so, NASA put the flags to half mast.

    I'm going to set the prediction period from 08:00 to noon, for the formal, standard analysis, seconds. That includes an hour of precursor time, and three hours following the explosion for the news to spread. I will also do explorations of other timespans and may look at variance deviations because so many people are interested to know what the GCP network did. Several eggs report late, so the full network will not be available for the immediate analysis.

  • Prediction, Roger Nelson, made evening of Feb 18 2003 for Subway fire in Taegu, South Korea, Tuesday at about 10:00 Korean time, which is 9 hours earlier than GMT. More than 130 people were killed and 99 were missing in South Korea on Tuesday after flames and smoke engulfed two crowded subway trains following an arson attack, officials said. The mayor of the southeastern city of Taegu said a 56-year-old male with a history of mental illness was suspected of starting the blaze at the end of the morning rush hour. Many struggled in vain to escape the inferno that reduced the trains to metal skeletons and sent black, acrid smoke belching into the sky for hours after the fire started.

    "My daughter called me twice at 9.57 a.m. crying 'mother there's smoke everywhere, but the door won't open!" said a woman at a makeshift crisis centre outside Taegu's Joongangro Station. "Everything is gone," said Sung Bo-hun, who was inside the subway until 7:40 pm (1040 GMT). "You can't recognise the people inside. It is all black and grey."

    Our prediction will specify a standard analysis of a 12 hour period from 09:45 Korean time (00:45 to 12:45 GMT) Feb 18 2003. This is intended to include a few minutes before the fire was set, and a longish aftermath period for the news to spread around the world. Our Korean egg is not running, but there are eggs in Japan and Vladivostok whose repsonse we will examine.

  • Prediction, Roger Nelson. Event on March 12, 2003

    Serbian Prime Minister Zoran Djindjic, who fought to transform his country from pariah nation to pro-Western democracy, was assassinated on Wednesday and the government swiftly declared a state of emergency. The prime minister died from his wounds at 13:30 (12:30 GMT) at Belgrade emergency center. According to TimesOnline he was shot at 12:35 local Belgrade time (11:35 GMT).

    The formal assessment period will be from 11:30 to 14:30 GMT, to include a little time before the murder, and a couple of hours for the news to spread. Standard analysis, seconds.

Scheduled Regular Events

Scheduled Unique Events
Unpredictable Events
Other Events Introduction

GCP Site Map

[next page]