The new answer engines — ChatGPT, Perplexity, Gemini, Claude — don't rank a list of links; they cite a few trusted sources inside the answer.
But they mostly don't crawl alone — they ride the classic indexes. Bing feeds ChatGPT, Google feeds Gemini, Brave feeds Claude (Perplexity runs its own). So you must be crawlable & indexed first — then EEAT & authority, schema, and FAQ structure decide who gets cited.
Indexes sweep outward across the web; the pages they reach become indexed (discoverable). When an engine cites a source, a gold beam travels back through its index to the engine — the source flares, crystallizes, and its gem grows a little with every citation.
Why it drifts: linked sources spring together while every source gently repels the rest — so the web self-organizes into clusters, the way a real knowledge graph groups related pages.