Trust Center

Newsroom

The operational state of this publication: what the data layer is doing right now, which figures have drifted away from the articles built around them, and how every piece scores on the things that can actually be measured. Published rather than kept internal, because a site that asks you to check it should show you its own queue.

LiveRead from the API when this page loads. With no JavaScript, the figures below are the ones baked in at build time.

—

Market provider

checking…

—

Market session

—

Upstream calls

this instance

—

Upstream errors

Drift queuePublished figures the issuing agency has since revised, and what we decided about each one.

Figures we published that the agency has since withdrawn
Where	Published as	Agency figure now	Status
The economy is sending two signals at once. The Fed is listening to the louder one.Nonfarm payrolls, July 2026	−23,000	+21,000	Correct as published
Jobs — July 2026Nonfarm payrolls, July 2026	−23,000	+21,000	Correct as published

Open means nobody here has ruled on it yet, and is the only status that wants work. Correct as published means the piece described the evidence that existed when it ran and we are not rewriting it — a quietly corrected argument is no longer a record of what we argued. Copy updated means the text now carries the newer figure. Every ruling is recorded against the exact figure it was made about, so rewriting the sentence returns the item to Open rather than inheriting a decision about a number that is no longer on the page.

Baseline gapNot an editorial matter. The hand-written fallback in the repository against what the data feed returned.

The display fallback, compared with what the data feed returned
Field	Fallback in the repository	Feed
S&P 500	7,786	7,638
Nasdaq Comp	26,729	26,523
Dow	53,732	51,683
10-yr Treasury	4.7	4.94
VIX	15.81	15.44
WTI Crude	82.4	107.02
Bitcoin	62,829	80,844
Inflation (CPI)	3.4	3.35

These are the values the site shows when a feed is silent. A gap here means the fallback is old, which every affected row already discloses with its own observation date and a dagger. It is listed for the operator, not queued for an editor, and it was previously reported as though articles had been written around these numbers. They were not.

Content auditWeakest first. Every score is computed from the article itself — hover any bar for the component and its working.

All 11 articles, scored out of 10
Article	Purpose	Words	Claims	Sources	Age	Score
What 3.4% inflation actually does to money sitting in a savings…Calc 2 · Inte 1 · Opin 1	CONVERSION	529	4	1	36 d	5.6
$10,000 a year ago: bitcoin, gold, and the part return charts never…Calc 2 · Inte 1 · Opin 1	REACH	583	4	3	35 d	6.8
Why small caps hit a record on the day the S&P 500 fellFact 1 · Inte 2 · Spec 1	IDENTITY	628	4	4	34 d	6.9
Market Autopsy: how Silicon Valley Bank died in 44 hoursFact 1 · Inte 1 · Opin 1	IDENTITY	683	3	2	36 d	6.9
$10,000 in NVIDIA in 2014 is $5.4 million. The interesting number is…Data 1 · Calc 2 · Inte 1 · Opin 1	REACH	682	5	3	6 d	7.0
24.6%Fact 1 · Inte 1	REACH	406	2	2	35 d	7.2
Everyone agrees earnings are booming. Here is the strongest case that…Data 1 · Calc 1 · Inte 2	TRUST	672	4	3	34 d	7.8
The economy is sending two signals at once. The Fed is listening to…Fact 1 · Data 1 · Inte 1 · Opin 1	TRUST	1,178	4	12	6 d	8.7
The Fed's rate is 3.5%. Your mortgage is 6.67%. Here's what sits in…Calc 2 · Inte 1	UTILITY	525	3	4	34 d	8.7
What a $5.45 trillion price tag actually assumes about NVIDIAData 1 · Calc 2 · Inte 3	IDENTITY	951	6	5	34 d	9.0
Before you buy NVIDIA ahead of August 26Fact 1 · Calc 1	UTILITY	551	2	5	34 d	9.8

Mean score 7.7. The scale rewards sourcing, labelling density, the share of claims that are evidence rather than judgement, having something inspectable, a complete Signal box, and recency. It says nothing about whether the writing is any good — no arithmetic can, and pretending otherwise would be the false precision this publication exists to argue against.

Rate-limit budget

The market strip polls on the cadence the API reports, not on a fixed timer: 15s while US markets are open, 600s when they are closed. Polling a daily series every fifteen seconds would be a request every reader pays for and nobody benefits from, which is why the interval follows the source rather than the habit. At the open-market rate that would still be 4 calls a minute per reader if each visit hit the upstream API — a thousand concurrent readers would exhaust a 60-a-minute free tier in under a second.

It does not, because the API responses carry s-maxage=15, stale-while-revalidate=60. The CDN serves one upstream call to every reader inside the window, so upstream usage is a function of time, not of traffic: at most four calls a minute regardless of whether one person is reading or a hundred thousand are. The backtest endpoint caches for six hours for the same reason — the answer to a question about 2014 does not change during the day.

Stale fields right now: none