For what little good it does in the moment, it seems clear that history will not look kindly upon many of the decisions being made in Washington. President Trump will bear the brunt of that eventual opprobrium, but it’s fair to assume that his explicit and tacit allies will earn more than a few paragraphs themselves, whatever language these histories might be written in.

What’s frustrating for Trump’s opponents is that this apparent inevitability seems to have only a limited effect on the decisions of those allies. There are some Republican legislators who’ve refused to confirm obviously unqualified nominees or unpopular legislation. But most haven’t, choosing instead to nestle in under the temporary shade provided by Trump’s umbrella of popularity.

This may well be because they don’t think that there will be some future analysis of the moment in which they are depicted on the wrong side of history. And they may be right. The longstanding idea that the arc of the moral universe bends toward justice does seem significantly less predictive than it used to.

But some legislators seem to understand that they are making decisions now that will color their legacies. Former Senate Majority Leader Mitch McConnell (R-KY) has voted against three of Trump’s most-contentious nominees — Tulsi Gabbard, Pete Hegseth and Robert Kennedy — and declined to vote for the nomination of Jeanine Pirro. But he’s an outlier; a review of 10 contentious confirmation votes shows that nearly every Republican supported all 10.

That was true from the least to the most conservative senators. The vast majority of the party caucus consistently sided with Trump’s picks. Of the 53 Republicans in the Senate, 43 of them voted for all 10 nominees. Another seven missed votes but voted yes on those they made.

Only one Democrat, John Fetterman (D-PA), voted for one of the 10 nominees (Pam Bondi).

What’s striking about the graph above is that the uniformity of Republican support also applies regardless of how states voted in 2024. Republicans from states that backed Trump only narrowly or not at all also supported the majority of or all of Trump’s particularly contentious nominees. The gray shaded area below indicates the range of 2024 margins in states that have a sitting Democratic senator. Several Republicans who sit in that range have backed Trump’s nominees regardless.

You’ll notice that one of these legislators is Thom Tillis, who announced that he wouldn’t seek reelection as he publicly opposed the spending bill advocated by Trump. But even this newfound independence didn’t keep him from supporting Emil Bove or Pirro. There’s a big difference between being a guy who declines to support a successful confirmation and being the guy who causes the confirmation to fail.

One conclusion from the data above is that the non-Tillis senators are more worried about political pressure from their right than from their left. Perhaps this is short-sightedness, an interest in getting past primaries and worrying about general elections later. Maybe it’s non-electoral fear, worry about how Trump (or his supporters) will respond to perceived disloyalty.

It does seem, though, that these legislators (and even legislators from less middle-of-the-road states) aren’t worried about the short-term cost of backing contentious or unqualified nominees, much less any long-term reputational damage. Their opponents could conceivably work to make Dave McCormick (R-PA) or Ted Budd (R-NC) the faces of confirming Emil Bove to a lifetime seat on the federal bench since, in practice, they were. But McCormick and Budd could take solace in knowing that they have a few years until they’re up for reelection. Not to mention that assuming ownership of a contentious confirmation — that of Brett Kavanaugh — didn’t keep Susan Collins from winning reelection in a state that voted against Trump last year.

As for those history books? Well, if written, they will include long lists of senators, granting them at least some sense of obscurity. And at this rate it will be illegal to publish them in the U.S. anyway.

Senate vote data: Blanche, Bondi, Bove, Gabbard, Hegseth, Kennedy, Lutnick, McMahon, Patel and Pirro.

This morning’s monthly jobs report estimated that the country didn’t add as many new workers in July as had been expected. More strikingly, it also showed that the number of jobs estimated to have been added in the reports for May and June were also lower than had been expected — and, in fact, lower than had been reported.

The Bureau of Labor Statistics’ initial estimates for May and June were increases of 139,000 and 147,000 jobs, respectively. After more data came in, though, the estimates shifted dramatically; it’s now estimated that the country added only 33,000 jobs in both months combined.

It’s important to note that such revisions are normal. The initial jobs numbers are just estimates and based on data for only part of each month. Those numbers are adjusted over the next two months and then for each fiscal year. Sometimes the adjustments lead to a higher estimate, sometimes lower. Sometimes there is a stretch in which jobs numbers are adjusted up. Sometimes there’s a stretch in which they’re adjusted down.

The BLS is transparent about all of this. Here, for example, are the initial and final estimates for each month since January 2017 (excluding the covid months in which the numbers shifted enormously and therefore broke the scale of the graph).

You’ll notice that the shifts for May and June are particularly large. In fact, they are among the ten biggest downward revisions over the past 20 years.

The other eight months? March and April 2020, September and October 2008, December 2008 and 2020, January 2009 and March 2021. In other words, all months in which there were shocks to the economy: the financial crisis in 2008-2009 and the pandemic in 2020-2021.

So was there a shock to the economy in May and June that might explain those months’ inclusion in the top ten? Well, it depends: Do you accept the possibility that Trump’s ad hoc, sweeping application of tariffs to imports from a huge swath of the world in April might have had negative downstream effects?

Trump doesn’t, but, then, he wouldn’t. Instead, in a post at Truth Social, he blamed a Biden appointee at the Bureau of Labor Statistics, announcing that she would be fired.

The numbers (and in particular those revisions) were “RIGGED in order to make the Republicans, and ME, look bad,” Trump wrote in another post. It was just like how, “on November 15, 2024, right after the Election … the Jobs Numbers were massively revised DOWNWARD, making a correction of over 818,000 Jobs — A TOTAL SCAM.”

Except that (as he curiously noted in today’s first post about the BLS) the annual revision he refers to didn’t happen right after the election. It happened in August 2024, which can be demonstrated by the fact that Trump complained about the change then.

It was a “MASSIVE SCANDAL,” he wrote, claiming that the Biden administration had been “caught fraudulently manipulating Job Statistics to hide the true extent of the Economic Ruin they have inflicted upon America.” He accused the administration of having “PADDED THE NUMBERS with an extra 818,000 Jobs that DO NOT EXIST, AND NEVER DID.”

So, uh, which is it? Is the BLS aiding a president (as claimed in 2024) by reporting too-high numbers and then revising them down or is it hurting a president (as claimed in 2025) by reporting too-high numbers and then revising them down? Or is the BLS just making estimates and then revising those estimates as more information about the job market becomes available?

That is not the throughline for Trump, of course. His throughline is that if the numbers say bad things about his opponent that’s good and that if they say bad things about him, that’s bad. And now that he has the ability to control who’s saying those things, he is now pledging to replace the Biden appointee with someone “much more competent and qualified.”

“Important numbers like this must be fair and accurate,” he wrote on Truth Social. “[T]hey can’t be manipulated for political purposes.”

But, of course, that’s exactly what Trump is demanding. He doesn’t want numbers that reflect the best estimates of government experts. He wants numbers that make him look good and he’s going to seek out a bureaucrat who will give him that.

Trump's 1st term BLS appointee🔥:

We call on Congress to respond immediately, to investigate the factors that led to Commissioner McEntarfer’s removal, to strongly urge the Commissioner’s continued service, and ensure that the nonpartisan integrity of the position is retained https://t.co/cYyLm9cV2p

— Brendan Duke (@Brendan_Duke) August 1, 2025

It is another step in the sweeping Trumpification of the federal government. Its power and its resources are being systemically and relentlessly redirected toward his interests and his vanity, with even temporary embarrassments provoking a sharp backlash.

There was something Trump could probably have done earlier this year to avoid the bad numbers in today’s jobs report. It isn’t that he should have fired the BLS official. It’s that he should have been more wary about toying with the U.S. and global economy by imposing those tariffs.

Here’s hoping that we’ll still be able to objectively evaluate the effects of that toying once Trump’s appointee is in place.

Let’s start deep in the weeds and then walk out.

On Thursday, FBI Director Kash Patel — chosen for that role in large part because of his commitment to undermining the investigation into Russian interference in the 2016 election — declared on social media that his team had uncovered a document, “along with thousands of other documents, buried in a back room at the FBI.” A classified annex to the report compiled by Special Counsel John Durham, Patel insisted that the document revealed “evidence that the Clinton campaign plotted to frame President Trump and fabricate the Russia collusion hoax.”

Patel’s post quoted writer John Solomon, who called the annex a smoking gun that, if “authenticated by further investigation” would show a plot by Hillary Clinton and her campaign to invent a Trump-Russia link. Oh, and rest assured, George Soros got a mention, too.

Solomon, along with Patel, had been tapped by Donald Trump in 2022 to serve as the then-former president’s representatives for records held by the National Archives. The point of that appointment appeared even at the time to have been to cherry-pick available material to make a case that the Russia investigation was a political hack job.

The effort simply took a few years to manifest.

The current allegation

The problem here is that the annex had been investigated further, by John Durham. The New York Times writes that Durham had come to the conclusion that the email at the heart of the claim was “probably manufactured” by Russian intelligence, the source of the email in the first place. Journalist Marcy Wheeler, who has tracked all of this closely for years, thinks that probably overstates Durham’s conclusion — but paints a compelling picture for why he should have. Either way, Durham relegated it to an annex when, if his probe had shown it to be legitimate or useful, it would have largely made the case he was seeking to prove.

The other thing is: none of this is new. Not only because Durham looked at it during his investigation, itself an effort to formalize a means of undercutting the public understanding of the overlap of Trump and Russia in 2016. But the core claim was made public back in 2020, in the waning days of Trump’s failed effort to secure a second consecutive term in office.

At that point, John Ratcliffe (now the director of the CIA) was serving as Director of National Intelligence, despite his thin resume. But he was a Trump ally, which one might fairly assume is why he first produced the allegation that Patel revived this week. There was a July meeting in which Clinton’s campaign discussed tying Trump to Russia, see, and that was how all of this started! Oh, also, this was a product of Russian intelligence analysis and the intelligence community “does not know the accuracy of this allegation or the extent to which the Russian intelligence analysis may reflect exaggeration or fabrication.”

It didn’t get much traction in part because of that (warranted) caveat and in part because of the other tumult of the campaign. (Trump had just been diagnosed with covid-19, for example.) It also didn’t get much traction because there wasn’t the same investment by Trump’s allies in debunking the Russia allegations — and because the effort to cast all of the Russia-Trump investigation as a “hoax” was still relatively young. Yes, Trump was calling all of it a hoax since his first months in office. But it took a while for that idea to percolate out to his base and, more importantly, for the right-wing media universe to distill the disparate, cherry-picked elements of that purported hoax into something that served as a convincing post hoc explanation for what had happened. By 2025, by now, Trump had both the burning desire for vengeance and the inoculation of his base necessary for the appointment of unabashed loyalists to intelligence positions and for a full-on assault on the reality of what had happened.

The context

It’s important to note that, in addition to the evidence of a Clinton-led plan to impugn Trump being unreliable, there was already a broad, public discussion about Trump’s ties to Russia at the point Clinton’s team is alleged to have invented the whole thing. If there was a conversation within the campaign that inspired the information stolen by Russia that then made its way to U.S. intelligence, it was probably simply one in which she and her advisers recognized that the Trump-Russia question was a liability for her opponent.

This conversation allegedly happened on July 26, 2016. On July 25, there was an article in Vice discussing possible efforts by Russia to aid Trump’s campaign. On July 21, there was a Politico article describing Russian enthusiasm for Trump. The month prior, The Washington Post had reported that Russia was behind the hacking of the Democratic National Committee’s network. On July 22, WikiLeaks began publishing that material, to the Democrats’ and Clinton’s disadvantage.

That was all public. Behind the scenes, there was a lot more going on. The FBI was already looking into Carter Page, an adviser to Trump’s campaign, when Page traveled to Russia in early July and gave a speech praising Vladimir Putin. Trump had tapped Paul Manafort to manage the Republican convention and then his campaign — someone who was also on the government’s radar for his ties to Russia. (Manafort, who volunteered to work for free, would later be revealed to have passed campaign polling information to a man linked in a bipartisan Senate report to Russian intelligence.)

Then there was the meeting at Trump Tower. A contact of the Trumps emailed Donald Trump Jr. in June 2016 to inform him that the Russians had dirt on Hillary Clinton. Trump Jr. responded with enthusiasm and a meeting was set up. The evening the meeting was set, Trump clinched the delegates needed to be the Republican presidential nominee. In his victory speech, he announced that he would “give a major speech on probably Monday of next week and we’re going to be discussing all of the things that have taken place with the Clintons.”

The contact who reached out to Trump Jr., by the way, knew the family from the time that the then-Trump-owned Miss Universe pageant was held in Moscow.

Do you think Putin will be going to The Miss Universe Pageant in November in Moscow – if so, will he become my new best friend?

— Donald J. Trump (@realDonaldTrump) June 19, 2013

The 2016 meeting, involving Trump Jr., Manafort and a Kremlin-linked Russian attorney, didn’t provide the promised dirt. The speech delineating “all of the things” about Clinton didn’t happen.

The bigger picture

It’s not clear that the Obama administration’s FBI or intelligence agencies knew all of the details in the summer of 2016, but they certainly knew the broad strokes. They knew about Page and Manafort and they knew that the Russians had hacked the DNC. When they learned in July that another adviser to Trump’s campaign had told a foreign diplomat about the existence of Clinton emails that had been stolen by Russia, the FBI launched a counter-intelligence probe. That was in late July, only days after Trump gave a press conference in which he talked about how he wanted to be friendly with Russia and asked Russia to try and hack into Clinton’s private email server. (They did.)

By the beginning of August 2016, in other words, key triggers for the Russia investigation were in place. There was reason to suspect that Russia was trying to influence American politics and the election. There was reason to be concerned about links between Trump’s campaign and Russia. American spies had evidence by early August 2016 that Putin was trying to help Trump win.

A 2019 report by the Justice Department inspector general determined that the launch of the FBI investigation into all of this was reasonably predicated. The investigation, which evolved into a special counsel probe during Trump’s presidency after Trump fired FBI Director James Comey, unearthed a lot of other details — and some dead ends. Robert Mueller’s report ran into the hundreds of pages, finding that, while there was not criminally provable coordination or collusion between Trump and Russian actors, the Trump campaign embraced Russia’s help.

The backlash

Trump was reportedly frustrated by post-election reports about Russia’s efforts because his victory in the 2016 campaign had been so narrow. He wanted the world to believe it reflected his popularity, not a quirk of the Electoral College. Having people saying that even that was a function of Russian interference was clearly grating. Even before he took office, he was decrying the Russia probe as a “witch hunt.” All he needed was a lattice of “evidence” that would prove his point.

As with his claims about the 2020 election being stolen, his allies in the right-wing media universe were happy to oblige. Various theories were elevated and embraced as proof that the Russia probe was political rather than justified by what was known at the time. Various FBI and intelligence officials were identified and pilloried. Picayune pieces of the probe were elevated as essential and damning. Phrases like “dossier” and “insurance policy” and “FISA warrants” became familiar vocabulary on the right, shorthand distillations of why, to their eye, the Russia investigation was all a set-up. Instead of disproving the validity of the Russia probe, though, each of those phrases simply provided something for those looking to dismiss the Russia probe as a hoax to latch onto, the way 2020 election deniers might latch onto “ballot dropboxes” or “absentee ballots.”

Now Trump is president again, unfettered in his desire to use his power to attack his enemies. His longstanding hostility to Barack Obama and Hillary Clinton and his wish to pretend that what happened in 2016 was all a set-up made it almost inevitable that his administration would try to gin up some sort of conspiracy. What’s surprising, really, is that it depends so heavily on already debunked claims. There was an actual investigation into all of this, initiated by his then-loyal Attorney General William Barr and led by a sympathetic U.S. attorney, John Durham! But Durham only managed to bring charges against three people and to get a plea agreement with one — the one whose actions had first been identified by that 2019 inspector general report.

All that’s happening now is that the people in charge of the bureaucracy are unbound by either honesty or shame. There’s no effort to be objective, just to present as Trump-friendly a case as possible. And since so few Trump allies ever saw or read analyses of the evidence the first time around — these were not details that got a lot of coverage on Fox News, for example — they can be presented as new and meaningful when they are neither.

Claims that the Russia probe was somehow a “hoax” have themselves always been a hoax, relying on after-the-fact cherry-picking and relying on hindsight. They have always depended, and depend now, on the willingness of Trump’s allies and supporters to be disinterested in nuance in favor of adherence to their partisan desires.

The reemergence of the “Russia hoax” narrative at the moment, of course, isn’t coincidental. It comes as Trump is trying to deflect attention from his relationship with Jeffrey Epstein, a situation in which his longstanding effort to align himself with Real Americans against The Elites crumbles. In addition to serving as a distraction, the Russia conversation puts him back on the right side of that equation. The Elites (Obama, Clinton, whoever) were out to get him because he was fighting for Real Americans!

Why not focus on that, instead of who stole which teenager from who?

---

Update: Patel and his allies have responded to criticisms that they are cherry-picking and misrepresenting by … cherry-picking and misrepresenting things to backstop their initial picked cherries. Wheeler, again, went into the weeds to explain what they’re getting wrong, however willfully.

A few years ago, during the last iteration of this website, I made a little interactive aimed at giving me a sense of an otherwise incomprehensible number: the number of Jewish deaths that occurred during the Holocaust. I started a timer that incremented by one every second, allowing me (and anyone who happened onto the site) to experience the slow accrual of six million seconds and, by extension, the scale of that particular horror.

It is 11:09 a.m. on July 31, 2025 as I write this. If I’d started a similar counter 6 million seconds ago, it would have begun about half an hour after midnight on May 23 of this year. If I started the counter now, it wouldn’t reach 6 million until late on the evening of Oct. 8.

Even put into those terms, it’s hard to grasp. It’s hard to understand how long ago May 23 was, for example. Perhaps it’s more useful to consider 6 million minutes, which would put us back in early March 2014. Or 6 million hours, which stretches back to the year 1341.

Our brains are simply incapable of understanding such large numbers. And for good reason. What need did early man have for the number “one million” — much less “one thousand” or even “one hundred”? For so much of human history, a history that itself extends back an incomprehensible length of time, there was simply no circumstance in which a one followed by six zeroes offered any utility. So there was no evolutionary advantage in being able to immediately grasp what 1 million meant.

Much less 1 billion. The need for humans to conprehend a thousand million is even more recent. It’s generally driven by discussions of money, first in the public and then in the private sector. It wasn’t until 2018 that “billionaire” began to appear more frequently in English-language books than did “millionaire.” And we can see that “trillionaire” — a million millions — is just starting to make its appearance.

To appreciate the scale of a billion, enterprising humans have turned to other representations. This effort to convey the wealth of Jeff Bezos using grains of rice — each representing not $1 but $100,000 — has remained with me.

@humphreytalks

Rice. Part 2: Jeff Bezos net worth. #rice #billion #billions #amazon #jeffbezos #money #personalfinance #xyzcba

♬ original sound – Humphrey Yang

And that’s when Bezos’s wealth was much more modest than it is today. The current richest man in the world, according to Bloomberg, is Elon Musk, whose net worth is three times what Bezos’s was at the time the video above was made. It’s almost literally incomprehensible.

This problem doesn’t simply apply to numbers, of course. Earlier this week, “Today in Tabs”‘s Rusty Foster theorized (convincingly) about how our evolutionary lack of intellectual sophistication has hobbled our ability to accurately assess the role of “artificial intelligence” systems.

“The essential problem is this: generative language software is very good at producing long and contextually informed strings of language, and humanity has never before experienced coherent language without any cognition driving it,” Foster wrote. “In regular life, we have never been required to distinguish between ‘language’ and ‘thought’ because only thought was capable of producing language, in any but the most trivial sense. The two are so closely welded that even a genius like Alan Turing couldn’t conceive of convincing human language being anything besides a direct proxy for ‘intelligence.’ ”

We can’t understand “one billion.” We also can’t understand that a thing that talks just like a human is just parroting human speech in the way we would understand it if that speech were coming from, say, a parrot. (I remain steadfast in my belief that a great deal of A.I.-centered rumination could be sidestepped if more people read the essay “The Soul of the Mark III Beast.” Amusingly, when I was trying to remember the title of this essay a few weeks ago, I tested an A.I. to see if it could remind me. Instead, it invented a non-existent essay and offered that as its answer.)

None of this is meant to be accusatory. Instead, it’s meant sympathetically — rather necessarily since I am a human who similarly lacks the capacity to immediately understand these things. It’s useful, not embarrassing, to remember that the human mind is clever enough to have invented things that it itself cannot fully comprehend. Man made a rock too big for Man to lift.

There’s a corollary here that I will mention because it has directly impacted my work at various times over the years. In addition to failing to understand scale, we often fail to understand change. For example, I would imagine that most Americans don’t know that the population of the U.S. increased by more than 20 percent just since the year 2000.

If you are 25 years old, in other words, there are now six people for every five that were in the U.S. when you were born. If you are 50, there are now eight people in the U.S. for every five that were here when you were born. So when people (Donald Trump) suggest that there’s some baffling conspiracy behind Joe Biden getting more votes in 2020 than Barack Obama did in 2008, you might remind them that the U.S. population grew by almost 10 percent over those 12 years. That context is important.

It’s important for people to understand our natural limitations. But it’s also important for writers to understand them and to accommodate them. It’s important to convey information about the world in a way that recognizes that this information may be misconstrued.

When we’re talking, for example, about something like the scale of deaths from covid or among children in Gaza, presenting that information as something other than a five- or six-digit number gives readers a more concrete sense of what’s actually happening.

Please get ready to witness a truly ridiculous measure of public opinion. I don’t want to oversell it, but I do want you to know that, when I saw it, I exclaimed out loud, and so I want you to be prepared in the event that you are driving (always a good time to read stuff on the internet) or have a heart condition.

I’d like to begin by establishing some context. As you may know, murder in the United States rose steadily from the 1960s through the 1980s until peaking in the early 1990s. There are a lot of theories about why the peak occurred then, but federal (and other) data demonstrates that the peak of murders in America came in the first half of that decade. That’s true both in terms of raw numbers and, more importantly, of the rate at which murders were occurring as a function of population.

It’s also true despite the surge in murders that began in Donald Trump’s first term in office during the coronavirus pandemic.

The rise and fall of murder is largely a story of the rise and fall of murders in American cities since that’s where a large percentage of Americans live. But data compiled by analyst Jeff Asher — whose work on the subject doesn’t receive the attention and credit it should — shows how that peak emerged and faded. The number of murders in various U.S. cities during the first half of this year is among the lowest on record in those places.

AND YET.

YouGov asked Americans this week if they thought the murder rate in American cities had risen or fallen since 1990. And most Americans, completely incorrectly, said that they thought murders had increased. A third thought the rate had increased a lot, which is the opposite of true!

Why? The partisan split on responses offers a hint: Republicans — who remain skeptical of cities, to put it generously — are much more likely to think that murder has increased. Which, again, it hasn’t.

Fascinatingly, younger Americans, people who didn’t live through the surge in crime that unfolded in the 1980s and early 1990s, are more likely to understand that murders have receded. Older Americans, who are also more likely to be Republicans, are more likely to be incorrect about the trend.

I’d indicated the year 1990 on the initial graphs so you can go back and compare murder rates/counts at that point with the rates now. But let’s put a fine point on it. Here’s how the number of murders in the cities Asher has tracked changed over the first half of each year relative to 1990.

That’s only some cities, but it’s not cherry-picking. In New York City, for example, the number of murders so far this year is down 85 percent relative to 1993. Last year, murders were down 83 percent relative to 1990. You have to ask yourself: When avatars of the right like Charlie Kirk proclaim that the city is intensely scary, is that a reflection of the city or of Charlie Kirk?

Murders, happily, are down in American cities. But there’s a lot of investment and political utility in suggesting that they are not. Americans, unfortunately, tend to believe it.