how to read it and where it comes from

. made with claude fable 5.2 and independently checked by gpt 5.6 sol as of 13 november 2026. not checked by a person. joscha bach didn't make or approve it.

how to read it

the board: time runs left to right in nine eras, and fields run top to bottom, from myth and religion down to ethics and AI risk. each card is a thinker, a work or an idea
a card: click it to open it. the summary of the idea comes first, then "start here": one page to read and, where one exists, the moment joscha bach talks about it, then what it connects to, then its other parts and, for joscha bach and his institute, quotes with their sources
connections: a solid line means the later point builds on the earlier one, a dashed line that it argues against it, a dotted line that the two are related. clicking a card lights up its lines and dims the rest
confidence: every summary and every connection ends with one of three words, highly likely, likely or possible, and usually a link to the page it was checked against. "hide less certain" keeps only what is highly likely: it hides every summary, section and connection below that level, and fades the cards whose main summary is below it
schools of thought: regroups the cards into families of ideas, each in year order, with the same lines between them. each card names its school. the schools are a model's judgement: claude fable 5.2 and gpt 5.6 sol each sorted every card independently; they agreed on 149 of 169; for the other 20 a third pass chose, and the card says where the other model put it. joscha bach and CIMC were placed by hand in functionalism, the tradition he places himself in

what the confidence words mean

every summary and connection carries one of three words. the word is not a rating of how sure a model feels. claude fable 5.2 and gpt 5.6 sol independently looked for and opened a source covering each item on 13 november 2026; the lower of their two verdicts is shown. most pages were searched for the relevant passages, not read end to end.

	a summary gets it when	a connection gets it when
highly likely	both checks found a source backing its main claims	both checks found a source stating the link itself; for links to joscha bach, a matched quote can establish it
likely	at least one check found the core supported but a claim unconfirmed; the card says which	at least one check called the link commonplace rather than directly documented, or found a qualification
possible	at least one check found no adequate source or a contradiction	at least one check found only an interpretation, or found the link wrong before it was corrected

nothing is marked certain. the source link on a card is the page used by the stricter check. items reworked in the second round carry one fresh check instead of two. two models can share errors and blind spots, so this reduces risk but is not a substitute for expert or primary-source review.

where it comes from

which points are on the map: names and ideas found in transcripts of joscha bach's public interviews, talks and papers, and the CIMC site. the transcripts were searched by name, not read end to end, so the list is incomplete. it leans towards computational views of mind, and traditions bach rarely mentions are thin or missing
summaries: written by claude fable 5.2 from its general background knowledge about each idea, then checked against a source as described above. not bach's words
connections: written by claude fable 5.2 from general knowledge, not taken from the transcripts, then checked against a source as described above. which idea 'builds on' which is often a judgement call
quotes: the ones marked "verbatim" were matched against the transcripts; "paraphrase" ones are claude fable 5.2's rewording
start here: the page to read is the most introductory of the pages the two source checks opened for the summary, encyclopedias first; it is chosen by rule, not by judgement. the joscha bach quote was picked by gpt 5.6 sol from the card's transcript mentions, as the one that says the most about the idea; it can only choose among real quotes. where all of them mention the idea only in passing, the card just names the talk. where the transcript has timings (most youtube and CCC talks), the quote was found in it and the link starts there, a couple of seconds early
years: approximate for older items
review: on 13 november 2026 claude fable 5.2 re-read every summary and connection and made about 45 corrections. the summaries of joscha bach's own ideas were then checked against his own essays, papers and talks, since the web pages the checks opened often did not cover what he says there; three details that are not in his texts were removed

cross-model verification

claude fable 5.2 and gpt 5.6 sol independently checked all 272 summaries and 226 connections against web sources. they disagreed on 46 summaries and 67 connections. the lower verdict determines the confidence shown on the map. the second pass prompted five summary corrections and seven connection corrections.

after those corrections, 217 summaries were highly likely and 55 likely; 134 connections were highly likely, 51 likely and 41 possible.

second round: sourcing what was less certain

every summary and connection below highly likely then went through two more steps. first, claude fable 5.2 looked for a page that states it; where none did, it made the smallest change the sources support: a softened or removed claim, a "builds on" turned into "related", a narrower note. it was told not to make a summary vaguer just to pass. then gpt 5.6 sol checked each result from scratch, with the same rules as before, and its verdict is the one shown. it was shown the pages claude had found but had to open and judge them itself; where it did not agree, three summaries were fixed again and rechecked.

27 summaries were reworded and 28 kept as they were with a new source. 34 connections were changed and 3 dropped: cell theory to michael levin, alan watts to illusionism, and kant to joscha bach, because nothing in bach's own texts credits kant for that idea. links to bach were checked against his talks and essays.

now summaries are highly likely and likely; connections are highly likely, likely and possible. the remaining ones are interpretive readings that the map keeps on purpose, such as calling aristotle's soul animist, which is bach's reading. this is still model review, not human fact-checking: a second model agreeing with the first is not the same as a person reading the source.

starting material

images

portraits and covers are thumbnails from wikipedia. each link goes to the article the image was taken from, where its author and licence are listed. some covers are non-free images used at thumbnail size.