alternatively, “Another Gopher Incident”?
the groundhog day analogy works on 2 levels—
the “deja vu” type of “groundhog day”: we have experienced model-catalyzed periods of AI delusion many times before. definitely over 10 times, but i’ve lost count.
the “few more weeks of this crap” type of “groundhog day”: the groundhog (AI ppl) saw its shadow (GPT-5.6 impressed them), so we'll have a few more weeks of winter. (AI mania)
the positive reception to GPT-5.6 out of the gates, reinforced by SK Hynix’s successful US debut, should buy the AI trade a few more weeks, carrying it through the end of the World Cup and into earnings season. earnings may then determine whether enthusiasm gets another leg higher or exhausts itself.
people believed this was intelligent in the late 1960s
hype-based narratives need novel dopamine hits to stay exciting to users, and especially to investors. like Big Tech earnings, each new announcement not only has to be bigger & better, it has to exceed high expectations more & more each time.
or, attention/stocks can consolidate flat/down, resetting (to some extent) their dopamine receptors.
as the latest Eliza Effect plays out, this time for GPT-5.6, societal & economic & cultural sores continue festering beneath the dirt we’ve spent years (or decades) packing into every wound instead of seeking proper treatment.
all that to say:
the unstoppable AI monster truck probably hasn't hit "E" yet, but it's running on fumes, and the longer we ignore maintenance & repairs, the more work we'll have to do to fix 'er up & get back on the road later.
the tech is gr8, for the record. but there’s a Reality Tidal Wave steadily growing in the distance getting ready to wipe the AI Beach slate clean, so we can start rebuilding from scratch—the post-dissolution re-emergence of structure we discussed in W28.Y2026. (this will be a good thing!)
the tidal wave is getting tougher to avoid. but… it isn’t here yet. the party can continue for a bit.
we’re still expecting a conviction-testing dump, rather than these relatively ordinary selloffs, before EOY. but that’s a fairly wide window…
sometime in late-July, early August seems reasonable, but we won’t make the same mistake as perma-bears and stick our heads in the sand prematurely. it’s much better to keep your head on a swivel and be ready to adapt.
we (FAYE) might look for short-term trading opportunities in the days ahead. just for fun. to keep busy amidst the market’s whiplash. it might be fun to grab some small slices of (probably AI or AI-adjacent) stocks that could rally into and/or out of earnings.
if those trades work out, we could do some dumping/trimming late july or early august(?).
we’ll keep you posted.
although we originally planned to sit in cash (outside of our longest-term investments) and do nothing, GPT-5.6 being a crowd pleaser & other ephemeral AI dopamine hits are tempting us to dabble with tiny, more speculative positions for a few weeks.
and maybe we’ll add to our long-term holdings if they dip enough... TBD.
👆 all the AI psychosis sleeper cells in San Francisco (& online) whenever a new model comes out.
(GPT-5.6 stealing the bulk of the spotlight from $META ( ▼ 2.46% )'s Muse Spark 1.1 is amusing, although. the latter, despite coming from Meta, excites our team more, as it should be cheaper to use—but still more expensive than the latest Kimi/MiMo/DeepSeek models… i think it’s comparable to GLM-5.2, so that’d be the most fun to compare Muse Spark 1.1 to. Meta appears determined to keep access to its new models inside its own API, which may prevent us from testing it, though.)
see you next time.
AURUM NOSTRUM NON EST AURUM VULGI 🪱