ULF-ELF Archive: Earthquake Prediction — What Was Actually Established

Based on a targeted read-through of the ULF-ELF Yahoo Group archive (3,268 messages), focused specifically on prediction accuracy and signal-to-quake timing. The honest picture that emerges is more mixed and self-critical than the group's premise might suggest — worth knowing before it goes anywhere near the forum or the book.

The core honest admission

The group's most consistently active organizer, John Wilcoxen, made two direct, explicit statements worth weighing heavily:

• 23 October 2005: "I do not think anyone in this group is at a place where they can 'predict' anything, except for Cathie perhaps. I think we are still at the point of trying to define what signals allow us to make predictions. We are not there yet."

• 17 February 2006 (near the end of the archived period): "After months of watching I (sorry to say) still have yet to pin down any real early warning signals of magnetic origin that I can say are related to pending earthquakes. Can anyone else say they have seen solid precursor signals?"

That second message, sent after months of dedicated effort by the group's most committed member, is essentially a negative result — no one in the thread that follows contradicts him with a confirmed counter-example.

What they were actually detecting

Most of what members reported — including the specific "Lots of activity today" thread around a 6.2-magnitude Philippine earthquake in December 2005 — was signal coincident with or shortly after an earthquake, not a genuine advance warning. John's own account of that event: his magnetic coil picked up a signal slightly before his local USGS station registered the acoustic/seismic wave — but this is straightforwardly explained by physics, not precursor detection: electromagnetic signals travel near light-speed, while seismic waves travel at a few kilometers per second, so an EM signal generated by a distant quake will naturally outrace the ground-shaking wave reaching a monitoring station tens of miles away. That's a real and interesting phenomenon, but it's detecting the earthquake as it happens, not predicting one in advance.

Another member (harrington780/Doug, California, Oct 2005) described the core practical problem directly: no reliable way to distinguish a genuine precursor signal from a weak local quake or a strong distant one, with depth of the quake also confusing the picture.

Delays and lead times: never established with confidence

This is the direct answer to your question, and it's a negative one within this archive: no consistent, validated lead time between signal and quake was ever pinned down by the instrumented members. John Wilcoxen explicitly asked at one point "how accurate can we get I wonder? ... what lead or lag time is there" — a question, not a finding.

The one claimed exception is Cathie Jordan, who reported a personal, non-instrumental "sensing" ability — describing feelings appearing "several days" before events. But even she and John agreed this had never been rigorously tested: her own words were that "the only way to correlate ... would be to see if there is a constant time frame (delay) between my senses and the recording on the instruments" — i.e., that validation work was identified as necessary but, as far as this archive shows, never actually done.

A genuine attempt at rigor, with no visible follow-through

In October 2005, member "karula4711" (Walter Fedderwitz) proposed exactly the right scientific protocol: members should publicly log predictions — date, time (to within an hour), and location — before any quake occurred, then validate against real seismic station data afterward. Notably, he volunteered to submit random guesses himself specifically to serve as a null-hypothesis control — a genuinely sound methodological instinct. I found no follow-up thread in the archive reporting actual scored results from this protocol, so either it fizzled out, wasn't widely adopted, or the results (if any) aren't captured in what's here.

Your own contribution to the group

Worth noting separately: you were an active member yourself, and in May 2005 shared original harmonics-based research directly relevant to the group — finding that Californian earthquake data showed excess aftershock activity clustering at intervals of 26, 52, and 78 minutes (with lesser peaks at 13.0, 6.5, and 3.25 minutes), interpreted as evidence of the whole Earth oscillating between a "lemon" and "grapefruit" shape. You also offered CRI's website as shared infrastructure for the group's data-collection effort. This is a genuine, dated, on-record contribution — a good candidate for its own line in the predictions/explanations list, separate from the group's broader (and largely unconfirmed) prediction claims.

An external claim worth flagging, not from this group

One thread (Dec 2005) mentions Jim Berkland, a real and separately well-known earthquake forecaster whose "Seismic Window" method (based on syzygy and lunar perigee timing, not ULF/ELF instrumentation) was described by a member as having "above an 86% accurate track record." This is a claim about Berkland cited secondhand by a group member, not something the ULF-ELF group itself measured — and worth treating cautiously either way, since Berkland's methodology and track record have been genuinely disputed elsewhere (critics point to loosely-defined prediction windows and after-the-fact matching as inflating apparent accuracy). Not evidence about the ULF-ELF group's own performance, but worth knowing it came up.

Bottom line

Your recollection that "these guys were clever" holds up — the equipment-building, signal-processing, and methodological self-awareness on display (Ernst Schmitter's multi-sensor setup, Michael Hebert's work on distinguishing real signal from coil ringing and back-EMF) is genuinely sophisticated amateur science. But the specific claim that they could reliably predict earthquakes, including location, isn't supported by what's in this archive — the group's own most active members were still openly searching for a valid signal as of the last dated messages here, with no confirmed hit-rate or lead-time ever established.