TL;DR
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
Astral Codex Ten’s October 2026 links post is a curated roundup spanning AI-generated art and translation, a bot’s win in a forecasting competition, and debates about marriage and suburbs. The author warns that the linked material has not been independently checked in every case and may contain errors; the post itself also flags disputed statistics and unverified claims.
Astral Codex Ten has published its October 2026 links roundup, a collection of essays, claims and examples touching on AI art and translation, marriage statistics, and machine forecasting. The post reports that a bot named Laertes won the Metaculus Cup in September, while its author cautions that the links have not all been independently verified and may include material that is wrong or misleading.
The roundup’s most concrete news item concerns the Metaculus Cup forecasting competition. The author says that, in September, a bot won the contest for the first time. Citing an account in The Economist, the post identifies the winner as Laertes, developed by Jeffrey Liang, who reportedly spent less than 150 hours and a few thousand dollars on computing power and data. The roundup also names bots from Mantic, FutureSearch and Preseen among the high performers, and says human participant Ben Shindel placed first among people.
The author connects the result to a forecast made in an earlier article: that AI and human forecasters would reach parity by the end of 2026. The competition result is relevant to that prediction, but the roundup does not provide comparative scores or enough information to establish that parity has been reached. It also links to news of Mantic raising $25 million in seed funding; that funding claim is presented through the linked report, rather than independently documented in the roundup.
Other entries explore creative and social questions. The author highlights AI-assisted art inspired by Hieronymus Bosch and machine attempts to render Chinese literature and the Odyssey in English verse, while noting limits to the examples and, in one case, that the author cannot assess the Chinese translation directly. A separate discussion challenges a claim that only 18% of marriages are non-miserable: the roundup says the underlying survey categorized every response other than “very happy” as miserable, even though “pretty happy” was an available answer. The author reads the General Social Survey as showing 61% of respondents describing their marriages as “very happy.”
AI Enters Forecasting Competitions
The reported Laertes win gives readers a specific example of AI systems performing in a public forecasting contest, rather than only a general claim about future capability. It may be relevant to discussion of whether AI can make useful predictions, but a single competition result does not establish broad superiority or parity with human forecasters. The roundup offers no full score breakdown, contest methodology or head-to-head analysis, so those conclusions remain open.
The marriage entry offers a different lesson about how survey categories shape headline statistics. Treating “pretty happy” marriages as miserable changes the interpretation of the figures, according to the author’s reading of the source. Readers should distinguish the survey responses from the linked commentator’s definition and argument. Across the post, the author’s caveat about verification reinforces the need to check original sources before relying on a claim.
Popular right now
Dyson V8 Cyclone Cordless Vacuum
De'Longhi Magnifica Evo Fully Automatic Espresso Machine with Manual Milk Frother, Silver
Breville Bambino Espresso Machine, Brushed Stainless Steel (BES450BSS)
As an affiliate, we earn on qualifying purchases.
How the Roundup Frames Its Links
This is a monthly links post, not a single-topic investigation or a report that independently verifies every item. The author says commenters often identify evidence that two or three links in each roundup are wrong or misleading, and says corrections are made when found. The post therefore mixes sourced reporting, personal judgments, examples and claims with different levels of verification.
That distinction matters for the AI entries. The author describes a prior forecast of AI-human forecasting parity by the end of 2026, then points to September’s reported bot victory as a related development. Elsewhere, claims about an older Claude model completing text that resembles internal emails are presented as claims, and the author says the alarming interpretation that the model could leak internal messages is probably false. These are not equivalent kinds of evidence: a contest result, a model demonstration and speculation about a system’s behavior require separate scrutiny.
The roundup also links to arguments about American suburbs, the origins of modern polyamory and Jeremy Bentham’s utilitarian cookbook, among other subjects. Those entries broaden the collection but are not the central news development. The author’s comments on them are personal reactions and summaries, not substitutes for the linked sources.
“I haven’t independently verified each link.”
— Astral Codex Ten author, on the links roundup
Claims Still Need Source Checks
The roundup does not include the Metaculus Cup’s full results, scoring details or information needed to compare the bots directly with human participants. It reports that Laertes won and identifies other high performers, but readers cannot determine from this post alone how large the margin was or what the result says about forecasting performance beyond this contest.
The marriage figures are also contested. The author says the linked argument’s definition of “miserable” includes respondents who said they were “pretty happy,” and reports a different figure from their reading of the General Social Survey. The full wording, sample and analysis should be checked in the source before drawing broader conclusions. More generally, the author warns that some links may be wrong or misleading, and says not all have been checked independently.
Several other entries are explicitly tentative. The author says the AI translation example may appear impressive but cannot assess its fidelity to the Chinese original; the discussion of Claude’s apparent email completions is framed as a claim, not proof of access to or disclosure of internal messages. The supplied excerpt also ends mid-entry after the words “Fee fi fo fum, I smell the blood of an Eng”, so the final item’s full claim and context are unavailable here.
Corrections and Contest Details
The immediate next step for readers is to consult the linked Metaculus and Economist coverage for the competition results and details not included in the roundup. Further results from forecasting contests could help show whether the Laertes win reflects a sustained capability or a single performance; the October post does not say when additional comparative data will be published.
The author says corrections to the links post will be made as issues are identified and that important corrections may be highlighted later. Readers should check the original post for updates and compare disputed claims with their underlying sources. The roundup does not set out a correction schedule, confirm a later update, or resolve the disagreement over the marriage statistic.
Key Questions
What is the main development in the October 2026 roundup?
The post reports that Laertes, an AI forecasting bot, won the Metaculus Cup in September. It does not provide full scores or establish that AI systems have reached parity with human forecasters.
Who developed Laertes?
The roundup cites The Economist as identifying the developer as Jeffrey Liang, described as a Texas-based polymath. It says he estimated spending less than 150 hours and a few thousand dollars on computing power and data.
Does the roundup verify every linked claim?
No. The author says they have not independently verified each link and warns that commenters sometimes find material that is wrong or misleading. The post says corrections are made when identified.
Why does the post dispute the marriage statistic?
The author says a linked argument counted every response except “very happy” as miserable, including “pretty happy.” The author’s reading of the General Social Survey is that 61% described their marriage as “very happy”; the underlying survey and analysis should be checked directly.
Does the bot’s win prove AI forecasting is as good as human forecasting?
No. The roundup treats the result as relevant to a prior prediction of AI-human forecasting parity, but it does not provide the comparative results or methodology needed to establish that broader conclusion.
Source: astralcodexten.com
