a market dump has begun!

(or continued from recent weeks, depending what you’re invested in.)

whether it’s the dump we’ve been waiting for remains to be seen. it could turn into that. or it could stabilize in the days/weeks to come for a bit.

we believe a bounce soon is possible, given the strong rotational undercurrents propping up markets as of late.

semiconductors, AI, & the like like $SMH ( ▼ 4.79% ) have been appropriately getting whacked.

everything retail FOMO’d into is now getting FOMO’d out of.

at the same time, they’re still up nicely over the last year. $MU ( ▲ 12.98% ), for example, is down ~35% over the last month but up ~160% YTD and ~560% over the last year. people who didn’t buy the top with borrowed money—or who sold in May and went away—should be doing just fine.

over the last three months, just a few percentage points separated $SMH ( ▼ 4.79% ), $QQQ ( ▲ 2.6% ), $IWM ( ▲ 0.7% ), & $IGV ( ▲ 0.64% ), interestingly.

the last month or so is when software started to pull ahead and hardware started to plummet. the rest of tech/growth suffered too, though nowhere near as much as semiconductors.

even today, on Warsh Day, software $IGV squeezed out a green daily candle while $SMH ( ▼ 4.79% ) fell almost 5%. biotech $XBI ( ▼ 0.43% ) & healthcare $XLV ( ▼ 1.92% ) ended red too, though the damage was considerably smaller. (and i believe they were both green at one point, briefly.)

a euphoric, collectively-held (why it was held, who knows!?) version of reality has experienced rapid unscheduled disassembly after colliding with a more believable, pessimistic version of reality.

generic capex/announcement dopamine fails to have the same effect when repeated endlessly for years… what a surprise!

now, while markets grasp at straws and try to decide what to do next—whether to pick up the pieces of their allegedly-reusable Starship rocket or scramble to re-invest in capex-soft software—Warsh has given investors… nothing!

which is probably all investors deserve, if we’re being honest.

jury’s still out on how Warsh fares. not enough data to go on yet. we have a task force monitoring his performance and collecting data, though.

today (FOMC+1) will be the day to pay attention to, not yesterday.

in place of artificial intelligence, of which the technology is neither artificial nor intelligent (but it’s very cool!), markets have for years stuffed their collective shared-head with artificial confidence, which is now spilling like polyester stuffing from a well-used dog toy.

markets will find their (articial) confidence again. it’s just a matter of time.

copper starts changing as soon as the environment gets at it. air, moisture, sulfur compounds, & chlorides work on the surface until it becomes something new.

sometimes that new layer protects what is underneath. a 2025 conservation study describes sulfide patina as protective, historically valuable, & aesthetically valuable.

chloride-heavy corrosion can stay porous, work inward, pit the metal, & eventually produce bronze disease.

"patina protects copper" broadly holds true, but there’s nuance to it.

like copper patinas, some market damage burns off pressure or gives a weak story time to meet less cooperative evidence. some damage erodes internal structure.

we keep compressing the gaps where anybody might figure out which kind they are looking at, probably because waiting resembles wasted time from a safe distance.

before computers were machines, “computers” were people.

Charles Babbage watched human computers break complicated work into repeatable tasks and imagined machinery taking some of it over.

Ada Lovelace looked at his machine and saw that it might manipulate relationships beyond arithmetic, perhaps even music.

a century later, John von Neumann helped set out a stored-program design that drew on years of shared work and became a basis for modern computing.

none of them began from a blank page.

Babbage’s machine, which he hoped would replace human computers, did not materialize during his lifetime, which sort of reminds us of certain tech CEOs…

Lovelace's statement that a machine had "no pretensions whatever to originate anything" still resonates today.

Mr. Jambo is a hit.

the video crossed a million views within its first two weeks, and people are already making lower-quality descendants. one person transferred it to an actual VHS tape and back. others recorded it playing on a 1986 Commodore monitor and a GE Spacemaker CRT. somebody made You Can Call Me Mr. Jambo (parodying You Can Call Me Al). comments referencing Mr. Jambo can be found beneath many of Paul Simon's videos.

Jambo blossoms, nurtured by nutrients & microplastics from the decomposting remains of a Paul Simon Graceland compact disc.

the song's central joke does most of the philosophical work out in the open (if you enable subtitles).

Mr. Jambo speaks in isiZulu, telling the singer, "I don't speak English. Do you have a translator? Stop yelling at me."

Barry understands none of this, then sings that the meaning was whatever he had decided he heard. Mr. Jambo's request for a translator becomes wisdom about love, souls, & ancient scrolls.

markets in a nutshell, am i right?

somebody says something, we hear what we want or expect to hear, then our perception gets propagated to our own audience (of anywhere from 0 people to millions of people).

as briefly mentioned earlier, Paul Simon is context here.

in a Vulture interview, Mr.Jambo’s creator, Kyle Gordon, called Graceland a "stunning masterpiece" and said the parody was "definitely mostly Graceland." Graceland had done its own borrowing forty years earlier, recording with South African musicians while an anti-apartheid cultural boycott tried to isolate the country. Gordon keeps its sound alive, then uses it to make fun of the American urge to visit Africa & return with a personalized spiritual answer.

patina changes a surface while sometimes preserving what is underneath.

Mr. Jambo is doing something like that to this particular 1980s sound, right down to fans copying the copy onto obsolete media.

it preserves the inheritance by roughing it up, by coating it with a protective layer that also draws attention to it

anyway, back to the collective shared-head.

market information once had to travel through ticker tape, quotation departments, telephones, newspapers, brokers, & exchange members before an ordinary person could do much with it.

the chain was slow & unequal, with high costs and plenty of stale information.

it also forced people to pause occasionally, if only by accident.

Brad Barber & Terrance Odean followed 1,607 investors who switched from telephone-based to online trading during the 1990s. before switching, the group had beaten the market by more than 2% annually. afterward, they traded more, speculated more, & lagged it by more than 3% annually.

Toomas Laarits & Jeffrey Wurgler used browser data from a much more recent, approximately representative sample and found that the median individual investor spends about six minutes researching a ticker before trading it, usually immediately beforehand. the mean was around half an hour, meaning some poor people did enough work to drag the average upward while the median investor read half a headline, saw a green number, & hopped on board.

Robert Shiller's Narrative Economics treats stories like epidemics. they move from infected people to susceptible ones, then slow as the pool of new hosts dries up. AI has plenty of real technology inside it, but it plays by the same narrative playbook. once everybody has heard the original strain, continued growth requires mutation—a fresh promise capable of finding new hosts and/or new money. without either, the susceptible population dwindles & the epidemic curve rolls over. (see: $SMH ( ▼ 4.79% ), $DRAM ( ▲ 7.36% ), etc.)

Han, Hirshleifer, & Walden modeled investors converting others to strategies more readily as reported returns grew. active, volatile, lottery-like strategies win the storytelling contest. Bali, Hirshleifer, Peng, Tang, & Wang found that heavier social-media discussion around lottery-like stocks predicted more retail buying, especially among Robinhood users, followed by lower returns.

Robinhood Social puts the story & the button in the same feed. users can see verified entries, exits, daily and one-year P&L, comments, & likes, then trade manually without getting out of bed, just a mere seconds after doomscrolling on Instagram. now, the moment Warsh stutters, somebody can panic & self-immolate on the public Robinhood Social feed. no more forced pause from unintentionally slow, laggy technology.

Warsh's contribution today was a Federal Open Market Committee decision to hold the target range at 3.50%-3.75%, with three members dissenting in favor of a quarter-point increase. in his opening statement, he said "prices reacted in real time to incoming information" and that market participants are learning to "play the ball, not the referee."

which sounds healthy enough at face value?

speed creates additional, different problems once we move from the person tapping buy to the infrastructure powering financial systems.

the U.S. moved standard securities settlement from T+2 to T+1 in May 2024, reducing the time during which prices can move, counterparties can fail, & somebody has to fund an unfinished trade. the SEC adopted T+1 without requiring market-wide T+0 because the day between trade & settlement also contains netting, reconciliation, funding, error correction, & other boring procedures that become visible mainly when they fail.

T+0 can mean settlement by the end of trade date, while real-time gross settlement means settling each transaction individually as it happens. that difference changes when cash has to move & how much the system needs at once. a DTCC, SIFMA, & ICI paper warned that retail and foreign investors might need to pre-fund positions under some T+0 designs, while real-time gross settlement would require cash and securities to move throughout the day instead of allowing the system to offset obligations first.

DTCC moved the experiment forward on July 15 when more than thirty firms used DTC-tokenized assets in live production transactions across private and public networks. the event covered collateral pledges, securities lending, Treasury/repo delivery-versus-payment, equities, & central-counterparty margin over several hours. the tokens remained digital twins of securities held at DTC and could move between traditional and tokenized form. DTCC plans to launch its Tokenization Service in October.

Tobias Adrian's April IMF note, Tokenized Finance, describes where the risk goes. atomic settlement and automated collateral can reduce counterparty exposure and reconciliation costs, while removing time previously used to mobilize liquidity, net exposures, correct discrepancies, or intervene before the transaction becomes final. "Together, these features shift the locus of risk from institutions to infrastructure," he writes.

that infrastructure can then do something extremely efficient & extremely stupid. "A faulty price feed or coding error can rapidly trigger cascading liquidations before authorities respond."

with more efficient systems, stress moves faster, while discretion arrives at its usual human speed.

one of cultivated meat's first dedicated research projects was funded by NASA. (of course.)

around 2000, Morris Benjaminson's Touro College team used NASA funding to grow goldfish muscle explants as a possible food source for long space missions. their peer-reviewed paper appeared in Acta Astronautica in 2002 and described an in-vitro muscle-protein system for "food products for Space travelers."

this was among the field's earliest laboratory research, according to a 2019 history of cultivated meat. a roughly contemporaneous bio-art project complicates any claim that it was absolutely the first cultivated-meat project or grant, which is why we have spent more time qualifying the goldfish-meat fact than most investors spend researching a trade.

where were we? rotation!

it has been underway for a while, even as different portfolios experienced different versions of the dump. what we’re watching now is where that money & attention go next.

if some of it rotates back into the overly-hot AI stocks, markets can subsist on rainwater & crackers a tad longer. if the AI narrative & its stocks descend, hand-in-hand, into the carbonite freezing chamber, the outlook turns considerably bleaker.

we still don’t know whether today marked the beginning of the larger dump, another stage of the one already underway, or the moment before a bounce. tomorrow gives us another observation. so will the day after that.

markets will find their artificial confidence again, before anybody can prove they deserve it.

we would prefer people spend more than six minutes deciding what to do when it returns!

see you next time.

(jambo! 👋)

AURUM NOSTRUM NON EST AURUM VULGI 🪱

🌐 SOFTWARE

🤖 HARDWARE

🧬 BIOTECH

🏗️ INFRASTRUCTURE

💰 FINANCE

🏛️ SOCIETY

⏳ PERSPECTIVES

RIBELO KONTRAŬ TIRANOJ ESTAS OBEO AL DIO ⚕