Wikipedia:Articles for deletion/Earthquake prediction

Earthquake prediction

Earthquake prediction (edit | talk | history | protect | delete | links | watch | logs | views) – (View log | edits since nomination)

(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

Note: This merge proposal was originally opened on the article's talk page. Following the March 2026 RfC, formal merge discussions are now held at AfD rather than the historical Proposed article mergers process (PAM). I've moved the discussion accordingly per WP:TPO.

Moved from Talk:Earthquake forecasting § Proposed merge with Earthquake prediction

 – FaviFake (talk) 15:36, 15 April 2026 (UTC)

Proposed merge with Earthquake prediction

I propose that Earthquake prediction be merged into Earthquake forecasting. Modern scientific consensus suggests that deterministic prediction is increasingly viewed as a specific, limiting sub-case within the broader discipline of probabilistic forecasting.

Rationale

   Conceptual Overlap: The distinction originated in 1935 (Wood and Gutenberg) to differentiate probabilistic "forecasts" from "predictions" (specific events within "narrow limits"). While the 1990s Nature debate highlighted a split between those who view deterministic prediction as unrealistic and those who search for rupture precursors, both fields rely on the same underlying science.

   Methodological Unity: Both concepts utilize the same stochastic point process architecture, earthquake catalog analysis, and rigorous hypothesis testing standards (e.g., CSEP).

   Scale, not Category: Deterministic prediction represents an extreme end of a "sliding scale" of hazard assessment rather than a separate discipline. It is more accurately represented as a specialized section within the forecasting article.

   Redundancy: Maintaining two separate articles leads to significant content duplication regarding seismicity models and statistical methods.

Proposal

I propose consolidating "Earthquake prediction" as a dedicated section within "Earthquake forecasting." This serves to align the encyclopedia with scientific reality: the pursuit of individual event "alarms" has largely been superseded by the development of rigorous, testable algorithms evaluating time-dependent probabilities. Upon merging, Earthquake prediction should become a redirect to the relevant section in this article. NeriB (talk) 08:01, 15 April 2026 (UTC)

Comment: To me, this nomination sounds like it was written by an AI. FaviFake (talk) 15:37, 15 April 2026 (UTC)

@FaviFake: Not AI... Why say so? I'm a researcher in this topic. --NeriB (talk) 06:45, 16 April 2026 (UTC)

Sure. Remember to WP:SIGN your comments though. FaviFake (talk) 17:03, 15 April 2026 (UTC)

Broken formatting and choppy, bold-heavy bullet-point-like text make people's AI antennae twitch. Stepwise Continuous Dysfunction (talk) 21:21, 15 April 2026 (UTC)

Comment:

This is a ridiculous proposal. For starters, Earthquake prediction (EP) is a large (around 200k bytes), mature article, while Earthquake forecasting (EF) is a relatively small (26k bytes) article of quite dubious character. EF is dubious because, aside from about five sentences (lacking relevant citations) that refer to earthquake forecasting, it is almost entirely extracted (with light editing) from Earthquake prediction, with the "Notable forecasts" section consisting solely of the lede from the UCERF3 article. While I would welcome a serious article on earthquake forecasting, the current EF article is effectively just a subset of the EP article. Incorporating a whole item into a subset of itself is, as I said, ridiculous.

(An aside: as the EF article says nothing about "probabilistic assessment of general earthquake hazard", the {For} link ought to be redirected to, say, UCERF3.)

Re NeriB's "Conceptual Overlap" rationale: the EP article clearly distinguishes the two concepts. More subtly, I would argue that prediction and forecasting are /not/ "both fields [that] rely on the same underlying science." EP has largely been a search for some basis - scientific or not- for predicting earthquakes, and notably unsuccessful. There simply is no "same underlying science." As to "Methodological Unity": perhaps this applies in some narrow academic context, but in the actual practice and history of earthquake prediction "rigorous hypothesis testing standards" is notably lacking.

It seems to me NeriB may have a very close-up view of modern seismological consensus; namely, that the only prospect of earthquake prediction is by means of probabilistic assessment, other approaches and claims having failed. But the scope of Earthquake prediction is not (per WP:CRYSTAL) where scientific study is headed, but of the history of the pursuit.

In summary: the proposal is ridiculous, and the rationale nonsensical. The proposal should be rejected, and Earthquake prediction kept.

• KEEP ~2026-23582-61 (talk) 01:37, 17 April 2026 (UTC)

@~2026-23582-61: The current "dubious" state of the Earthquake forecasting article is a primary argument for a merge, not against it. Consolidating the technical depth of the prediction article into the forecasting framework allows us to fix the lack of rigor in the latter while accurately reflecting the modern scientific shift. Regarding the "same science": both fields/ goals utilize the same methods - wether it's stochastic modeling or physics based modeling or something else; the distinction is primarily the operational threshold for an "alarm." Even if the history of EP is a history of failure, it represents the pursuit of a deterministic goal within the broader discipline of forecasting. A merge ensures this history is preserved while being framed within the modern consensus. NeriB (talk) 20:06, 18 April 2026 (UTC)

NeriB: The dubiousness of the EF article is that - aside from about five sentences - there is essentially nothing there but what has been copied from the EP article (and UCERF3). Do you not see that? Will you grant that if the EP and EF articles are merged, and then the duplicated parts removed, the result is just the original EP article? (Plus the UCERF3 lede and the TEC section, but to simplify this discussion let's just ignore those two bits.) If we simply removed the duplicated parts of EF first, then THERE IS NOTHING LEFT TO MERGE. The only effect would be to rename EP to EF, which isn't going to fly. Sorry, but the dubiousness of EF is actually a primary argument AGAINST a merge.

But perhaps what you really want (?) is to "fix the lack of rigor" in EF by importing the "technical depth" of EP? Sorry; that "depth" is specific to the topic, and reflects the work done documenting the topic. What you are proposing is akin to saying: "Having not done my homework, I'd like to steal the work of the kid next to me, and put my name on it."

(Incidentally, I think earthquake forecasting could be a distinct - and separate! - article. Think in terms of explaining how the UCERF forecasts are generated.)

Underlying all of this, you seem to be running on a pair of misunderstandings. First, that EP is - and ought to be! - a subfield of EF, as they (allegedly) "share the same method", etc. But! they do not. E.g., within EP I see no use of "stochastic point process architecture", and very little "rigorous hypothesis testing standards".

Second, it seems you do not recognize that forecasting (at least as I have seen it in UCERF) is not about particular earthquakes, but long-term seismic hazard. EP is about near-term evacuation of buildings (etc.); EF is about (e.g.) seismic refitting of buildings.

But let's come back to the bottom line: Earthquake forecasting is basically a stub article, and sucking in - and then deleting - a larger article does no good. ~2026-23582-61 (talk) 22:59, 20 April 2026 (UTC)

• Note: this discussion has been included in the list of Science-related AfD discussions. Spiderone^{(Talk to Spider)} 15:55, 15 April 2026 (UTC)
• KEEP - There are practical differences between prediction and forecasting. Going back to Jordan et al. (2011) "Predictions and forecasts both make statements about future earthquake activity based on information available at the time; that is, they provide prospective (before-the-fact) rather than retrospective (after-the-fact) earthquake information. In this report, the Commission distinguishes between a prediction and a forecast using a strict dichotomy. A prediction involves casting an alarm — an assertion that one or more target ruptures will occur in a specified subdomain of space (subregion) and future time (subinterval). Predictions are therefore prospective deterministic statements: if a target event occurs in the alarm subdomain, the prediction is a true alarm; otherwise it is a false alarm (or type-I error). If a target event occurs in a subdomain without an alarm, the error is a failure-to-predict (or type-II error). A prediction can also be cast as an anti-alarm, a deterministic statement that no target rupture will occur in a subdomain [33]. Forecasts are prospective probabilistic statements: they specify the probabilities that target events will occur in space- time subdomains. The probability in a particular subdomain is a number P that ranges between 0 (no chance of a target event) and 1 (certainty of a target event). A time-independent ICEF FINAL REPORT - 30 MAY 2011 forecast is one in which the subdomain probabilities depend only on the long-term rates of target events; the events are assumed to be randomly distributed in time, and the probabilities of future events are thus independent of earthquake history or any other time-dependent information. Time-independent forecasts are useful for long-term seismic hazard analysis." Mikenorton (talk) 10:18, 18 April 2026 (UTC)

  @Mikenorton: The Jordan et al. (2011) report actually reinforces the case for a merge. It treats prediction as a "strict dichotomy" of output but places it within the same prospective framework as forecasting. As the report notes, prediction is treated as a type of forecasting with very narrow, up to infinitesimal, error margins (they state prediction as deterministic, i.e. no errors). This can also be noticed by the following quotation: “At the present time, earthquake probabilities derived from validated models are too low for precise short-term predictions of when and where big quakes will strike; consequently, no schemes for "deterministic" earthquake prediction have been qualified for operational purposes." This confirms that prediction is effectively a specialized goal (an infinitesimal error margin) within the field of probabilistic forecasting.
  This is further emphasized by Mizrahi et. al. (2024): "What is commonly referred to as earthquake prediction, namely the deterministic description of the location, time, and magnitude of a future earthquake “within narrow limits […], so that a planned evacuation can take place” (Main, 1999), is currently not possible. The term earthquake forecasting refers to a probabilistic assessment of earthquake occurrence, over a range of magnitudes, and within some specified time frame and region (e.g., Kagan & Jackson, 2000)."
  Merging avoids duplicating the methodology, history and validation standards that both fields fundamentally share. NeriB (talk) 20:06, 18 April 2026 (UTC)

• Earthquake forecasting#Elastic rebound is a copy-paste WP:DUPLICATE of Earthquake prediction#Elastic rebound, Earthquake forecasting#Characteristic earthquakes is a duplicate of Earthquake prediction#Characteristic earthquakes, and Earthquake forecasting#Seismic gaps is a duplicate of Earthquake prediction#Seismic gaps. So while these two concepts are not identical, if most of the content is identical, they should still be merged for clear organization and presentation. If these are kept separate, I'd encourage appropriate split of the relevant sections. Reywas92^Talk 14:50, 18 April 2026 (UTC)

  @Reywas92: Exactly. A unified article under the "Earthquake forecasting" would allow for a much cleaner organization of the theories, methodologies, history and applications of both goals. NeriB (talk) 20:06, 18 April 2026 (UTC)

  Agreed. FaviFake (talk) 10:04, 19 April 2026 (UTC)

Disagree. "[M]ost of the content" of Earthquake forecasting is identical to a very small portion of Earthquake prediction only because that is where the former was copied from. (Which content is not even relevant to EF.) As I have said earlier, when the duplicated material is removed from EF, there is essenially nothing left to merge. Also, there has been no demonstration here that prediction and forecasting have theories, methodologies, etc. in common, perhaps because the EF article has no EF specific content. ~2026-23582-61 (talk) 23:08, 20 April 2026 (UTC)

This doesn't make sense, if "the EF article has no EF specific content" and it's not "relevant to EF" then why is there an EF article? What should the EF article have? Even the leads overlap. I do not think this level of copy/paste duplication is acceptable, particularly as it confuses the reader as to what exactly each topic is and how they are different. If you think these should be separate pages, please outline which sections/content each one should or shouldn't include so that there is not so much overlap. Or does it need to start over from scratch? — Reywas92^Talk 01:35, 21 April 2026 (UTC)

I agree, the current EF does not make sense. And: good questions. For one of which the answer is: start over.

I maintain that, though similar, EP and EF as topics are sufficiently distinct to warrant separate articles, and also that there is enough material to write a proper (albeit modest) article for EF. By "proper" I mean not a compilation of "what everyone on the Web says", but diving into how operational forecasting is done. See the sources in UCERF3.

An ER article should start - in the lede - by distinguishing between EP and EF, just as the EP article does. To that end both articles rely on the same sources, and so the initial text might be very similar. Every section after that should be scrubbed. If no one is up to rewriting it EF could be left as a stub (but with a link to UCERF3 as an example of EF). ~2026-23582-61 (talk) 02:58, 22 April 2026 (UTC)

@~2026-23582-61: The fact that "prediction" is the preferred colloquial term among the public does not justify maintaining an unscientific separation in an encyclopedia. Priority here should be scientific accuracy, not reinforcing popular nomenclature. EP is not a separate discipline from EF; it is the aspiration to minimize forecasting errors to a deterministic level. Since both rely on the same physical basis and scientific methods, as you yourself stated, maintaining two articles is technically redundant.

If the mathematical models, catalog analysis, and hypothesis testing (e.g., CSEP) are identical, there is no justification for two separate articles. Moving the content into a single unified article is the only way to ensure readers understand that "prediction" is a specific, limiting operational goal within the broader framework of forecasting. Leaving EF as a stub while the physics remains duplicated in EP only serves to confuse the reader and creates a perpetual maintenance burden. NeriB (talk) 06:58, 22 April 2026 (UTC)

Relisted to generate a more thorough discussion and clearer consensus.
Relisting comment: This needs further discussion from more participants
Please add new comments below this notice. Thanks, Star Mississippi 02:39, 23 April 2026 (UTC)

@NeriB: Who said anything about a "preferred colloquial term among the public"? The use here of "prediction" is not based on colloquial or popular usage, but on multiple reliable scientific sources. Your objection is a strawman argument; please be more careful.

Note that I have NOT stated - actually deny - that 1) "both [EP and EF] rely on the same physical basis and scientific methods", or that 2) "maintaining two articles is technically redundant"; your assertions are FALSE.

Regarding having two separate articles: I am fine with that. What I object to is, first, someone creating an article by copying whole sections from other articles where "earthquake" and "forecasting" are mentioned. That disserves the topic, and WP as a whole.

Second, I object to "fixing" such bastardy by sucking in - and then deleting! - a much bigger article. About all that amounts to is a back-handed way of getting a rename. If you really think that EP should be a child of EF, then how about first creating the parent? Would you do that?

You continue to insist that EP and EF share - even have identical - scientific methods, etc. Can you provide examples? Alternately, can you point to any mathematical models, catalog analysis, etc. used in any of the notable earthquake predictions included in EP?

It seems to me your breadth of view here is limited entirely to CSEP. Your statements certainly make more sense in that limited context (and we could have a discussion about that). How about writing an article about CSEP?

(For any passers-by, CSEP is the "Collaboratory for the Study of Earthquake Predictability". For an introduction see https://arxiv.org/pdf/1003.1092/, or https://doi.org/10.1038/s41597-025-05766-3. Or even Zechar's dissertation, cited in EP.)

~2026-23582-61 (talk) 23:17, 24 April 2026 (UTC)

@~2026-23582-61: The objection regarding "colloquial usage" misses the core issue: while the word "prediction" appears in scientific literature, it is not a distinct operational discipline. As established in the sources I’ve cited (Jordan et al. 2011; Mizrahi et al. 2021), any deterministic alarm is simply an operational threshold reached by a probabilistic model. Earthquake prediction is an aspiration to achieve a level of precision in forecasting that allows for deterministic action; it is not a separate science. The mathematical models and catalog analyses found in the history of EP are fundamentally statistical assessments of seismicity rates—the same building blocks used in forecasting—differing only in error margins rather than underlying geophysics. Despite claiming a methodological split, the objector has provided no peer-reviewed citations (WP:RS) to substantiate an operational method that exists outside statistical seismology, leaving the burden of proof (WP:Verifiability) entirely unmet. NeriB (talk) 06:56, 27 April 2026 (UTC)

There is the core problem: you have your head stuck so deep into statistical seismology you don't allow anything that "is not a distinct operational discipline", you insist (repeatedly) that "mathematical models and catalog analyses" are inherent in EP ("that both fields fundamentally share" - your bolding), and yet you have not - cannot? despite several requests - point to any such elements in the past history of EP. By your argument the article currently titled "Earthquake prediction" is thus NOT about EP. (E.g., where is the mathematical modeling?) Yet you would make this entire article a mere section within "Earthquake forecasting", which currently exists as several duplicated bits of (mainly) the EP article. Never mind that NONE of this material even hints at your "scientific methods".

In the statistical seismology article it would be fine to explain that in that specific context EP is considered an aspirational end-point by those working in that field. But you haven't done that. Nor have you addressed any of the points I have raised about why this proposal is ridiculous. And that is the bottom line here: this proposal to merge/delete (effectively rename) EP is, as I said at the start, ridiculous. ~2026-23582-61 (talk) 20:47, 28 April 2026 (UTC)

Relisted to generate a more thorough discussion and clearer consensus.
Relisting comment: I request that previous participants stop bludgeoning the discussion and let other people express their views.
Please add new comments below this notice. Thanks, Sandstein 17:56, 1 May 2026 (UTC)

Comment: The two sections on "Elastic rebound" and "Seismic gaps" should be removed from the Earthquake prediction article as they are not directly relevant to the topic as defined, something that I will do once I've fully checked that none of the citations are used elsewhere in the article. This will reduce the duplication in the articles under discussion. I suspect that the "Characteristic earthquake" section could do with heavy trimming or complete removal, as we already have a section on "1985–1993: Parkfield, U.S. (Bakun-Lindh)" in the "Notable predictions" section. On checking back through the history of both articles, I see that the repeated sections were originally added to the Prediction article when it covered both topics, but these were not changed when the forecasting article got created, which in hindsight they should have been. Mikenorton (talk) 12:16, 3 May 2026 (UTC)

Those two sections are relevant to the basis for making predictions; I suggest not removing them. Reid's formulation of elastic rebound is especially notable historically as a major step in understanding earthquakes.

The "duplication in the articles" is entirely due to the copying of those sections in December 2016 from the then current EP article (and one section from UCERF3) to the EF article. If duplication is so burdensome as has been alleged here, then the simplest resolution is to revert the duplication in EF.

These sections were - and are - part of EP's "Predictions methods" section (split into "Precursors" and "Trends", etc.), whose purpose was (initially) to illustrate methods deemed notable because they either 1) were the basis of a notable prediction (e.g., Parkfield), 2) were "highly regarded" as possible precursors (e.g., dilatancy–diffusion, M8, AMR, etc.), or 3) figure prominently in the popular mind (e.g., animal behavior), and therefore ought to be addressed lest anyone think "science" or WP is ignoring them. Note that these sections are typically very brief, in the form of a paragraph explaining what it is, and another assessing how well it has done.

[Unfortunately, there has been substantial bloat by the addition of methods that do not meet these criteria, and have not notably contributed to prediction. If you want to do some heavy trimming I suggest removing the following sections: Earthquake_prediction#Freund_physics (or its elaborating sub-sections), Earthquake_prediction#Nowcasting, Earthquake_prediction#Machine_learning, and the last two paragraphs of Earthquake_prediction#Animal behavior. Also "EMP induced seismicity" - causing earthquakes is not predicting them. The VAN stuff ought to be trimmed, but that is a different barrel of worms.]

What any of these sections would do in a forecasting article depends on the scope taken, and how the article is written. E.g., the "forecasting" (that aspires to prediction?) done in the context of CSEP and RELM is based solely on supposed patterns in the seismic data. Any consideration of (e.g.) seismic gaps is buried deep in the model being tested, and not really relevant unless the various forecasting models are being examined.

Similarly, UCERF is a prime example of forecasting. But note: it "forecasts" the seismic risk expected to accrue over some period time. It does not predict, or even "forecast", any particular event. And it seems to be an adequate article without needing to mention seismic gaps and such.

Before getting into a lather about supposed overlaps of supplementary content in EP and EF lets see what might needed in an article that actually has content about earthquake forecasting. ~2026-23582-61 (talk) 21:29, 4 May 2026 (UTC)

The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review). No further edits should be made to this page.