Month: January 2024

Nutshell summary: Good news!

Long, complicated explanation: Buckle up for this one. No doctors were involved in this analysis.

As you know, for years I've been getting M-protein results to check the level of multiple myeloma in my bone marrow. This is done via an SPEP test—Serum Protein Electrophoresis. A successful treatment of my cancer would mean that the M-protein level was "undetectable." Not completely gone, but so low the SPEP test couldn't detect it.

I always assumed this meant an SPEP result of 0. But no: I finally got tired of never getting an explanation of how this stuff really works so I looked up the clinical literature myself. In the relevant units, it turns out that SPEP can only detect M-protein down to about 0.10.

But there's more. I've also been getting a second test all along: Serum Immunofixation. It's a binary test (normal/abnormal) so I've never paid any attention to it. Obviously I've been abnormal ever since I was diagnosed.

Not anymore. It turns out that immunofixation is only accurate down to a level of about 0.05. Below that the cancer level is undetectable.

Now, as you might recall, the CAR-T procedure I had last April can produce a range of outcomes. The best outcome is "complete," which turns out to mean a cancer level that's undetectable by both SPEP and immunofixation. Conversely, if you're below the SPEP level but still above the immunofixation level, your outcome is "partial complete."

That's where my test results have been for months. But this month something changed in the immunofixation results. Instead of "IgG Kappa monoclonal gammopathy"—i.e., multiple myeloma—the result was "Band of restricted mobility of IgG Kappa. Consider retesting if clinically indicated." This suggests that the immunofixation result is right on the edge of being undetectable. Here's a chart:

September was the last time I got an SPEP result. Now, in January, I seem to be very near to not getting an immunofixation result either. If this is true, it means I've finally reached the fabled "complete" response from the CAR-T procedure. It took eight months instead of six weeks, but in the end it came through.

Now, it's worth noting that no doctor has confirmed this for me. In the case of my general oncologist I think it's because he's not familiar enough with CAR-T to know these details. In the case of my transplant physician, I imagine he knows but just never felt like explaining in detail.

So that's that. It's good news no matter what, and if my amateur interpretation is correct it's very good news. The next step would be a test for MRD—Minimal Residual Disease—which is more sensitive than either SPEP or immunofixation. No one has suggested this yet, probably because I was nowhere near needing it before now. Maybe someday.

In any case, who knows? The fact that it took so damn long to reach this level might mean it will also take a very long time before it starts to go up again. My multiple myeloma has always been fairly slow reacting. Check back with me in 2028 or so and we'll know.

This is just a reminder:

It's worth noting that corporate profits spiked in mid-2020 and finished their rise by early 2021. It's not obvious from the chart, but this is before inflation—either in consumer prices or producer inputs—had risen more than barely. Since the start of the pandemic, inflation has gone up about 20 points while corporate profits have soared nearly 100 points. At the same time, average markups over cost have increased from about 11% to 16%.

The increase in corporate profits has never been a big cause of inflation. That was due mostly to supply chain disruptions and government assistance. But it's certainly added a bit to it. Based on the markup data, I'd guess that prices increased 4-5% just thanks to higher markups. That's nearly a quarter of the inflationary surge.

And now that inflation is over and it's obvious that corporations overshot their price increases? They'll lower prices, right? Ha ha.

Today brings some surprisingly good news about the economy—or at least about people's views of it. The University of Michigan's consumer sentiment survey took its second big jump upward this month after a big jump in December:

According to the Wall Street Journal, this is the biggest two-month spike since 1991. Consumer sentiment still isn't as high as it was throughout the economic expansion of the teens, but it's getting there.

The University of Michigan press release says the index has gone up even among Republicans. Maybe this means the fact that the economy has been great is finally breaking through the Great Wall of Fox News. There's only so long you can deny reality, right?

The Guardian has a dramatic headline today:

A groundbreaking study shows kids learn better on paper, not screens.

I apologize in advance for going down a very long and very complicated rabbit hole on this, but you can always scroll past it if you get bored. Needless to say, I got curious and decided to read the paper it was based on. My brain nearly exploded.

In a nutshell, a team of researchers hooked up a bunch of kids to EEG helmets and measured their responses to reading cues. Some kids were tested after reading narrative passages on a screen while others were tested after reading the passages in print. So far, so good.

But here's where it starts to get complicated. The kids read the passages and were then presented with "probe words." The words were either related to the passage, unrelated, or sort of related.

The theory here is that you have a stronger EEG response to new things. Thus, if you had good comprehension of the passage, you should have a weak response to related words since you've already seen them and a strong response to unrelated words because they're new to you.

So the authors made a prediction for reading in print vs. reading on a screen. Here's the prediction:

In the print scenario, response would get steadily stronger as words became less related to ones the kids had seen before. But in the screen scenario, there would be no difference between moderately related and unrelated words.

Why? I have no idea. There's a bit of handwaving late in the paper, but I can't say that I found it very convincing. Note that "chimera words" are those that are moderately related to the text:

Whether chimeras are perceived as related or unrelated to the text may depend on the strength of the encoded memory traces established during text discourse processing. Therefore, perception of chimera words as unrelated words would be consistent with shallow discourse processing as hypothesized in digital text reading, whereas chimera words perceived as related would be consistent with deeper discourse processing as observed in print reading.

Hmmm. Anyway, here are the results:

These look pretty similar! But it's true that in the screen scenario the chimera words and the unrelated words are nearly identical at the critical 400 μsec mark. In the print scenario, there's a small difference.

On the other hand, in the print scenario there's no difference between chimera and related words. This doesn't confirm their prediction.

And related words have a noticeably weaker response in the screen scenario. Unless I missed it, the authors don't discuss this.

But there's more! The researchers also performed some ordinary comprehension tests. Check it out:

There is virtually no difference between print and screen, and what difference there is is dwarfed by the error bars.

NUTSHELL SUMMARY: A ton of work went into this. However:

• The sample size is small (59 children).
• How do we know if a word is "moderately" related to a text passage? The authors picked words using software and then had a hundred random adults rate them. There's a good chance the results are all but meaningless.
• The authors' prediction is extraordinarily indirect. How sure are we that the difference between high and low comprehension will show up specifically in the response difference between moderately related words and unrelated words?
• In any case, the prediction was only barely borne out.
• And in the print scenario the response is virtually the same between all words. This definitely does not support the hypothesis.
• And the absolute response to related words is weakest in the screen scenario, which means the kids were very familiar with them. This is fairly direct evidence of greater comprehension in the screen scenario.
• Finally, in an ordinary test of comprehension, there's no difference at all between screen and print. This was true at the time of the test and on a test administered a day later.

Color me unconvinced. Maybe I'm missing something, but the prediction is so roundabout and the results are so minuscule that I have a hard time believing there's anything here.

POSTSCRIPT: Ah. At the bottom of the Guardian writeup we're told that the author "participated in the fundraising" for the study. That certainly explains the enthusiasm of the headline.

During most of the time that Obamacare has been around, more people have benefited from its Medicaid expansion than from buying insurance on the marketplaces. But that's no longer true. Final numbers aren't in yet, but Andrew Sprung reports that enrollment in the marketplaces is already up 25% compared to last year and up nearly 80% compared to 2020.

Why? Several reasons, probably, but the main one is the expansion of subsidies passed in 2021 and then extended last year. I haven't posted an update on the number of uninsured people recently, so here's the latest:

After Obamacare passed, the uninsured rate dropped from 16% to around 10-11% and stayed there for several years. Then, in 2021, when the expanded subsidies were passed, the uninsured rate started dropping again. It was at 8.6% in mid-2023, and that was before the big jump of this year. It will be a while before we have updated numbers, but when we do I'll bet that the uninsured rate at the start of 2024 will be down to 8% or so. If the Medicaid holdout states would cut the crap and accept the Medicaid expansion, we'd probably be down to 7%.

That's still 7% too much, but it's a helluva lot better than a decade ago.

A couple of days ago I wrote about the new bipartisan bill to expand the Child Tax Credit and lower taxes on business. The topline numbers were simple: over ten years the CTC would be increased by $33 billion while taxes would be lowered by $33 billion.

But David Dayen says this is misleading. We now have an official scoring of the bill and here's how things look year-by-year over the next decade:

The CTC is extended for three years and then goes away. It costs $33 billion over three years and $33 billion over ten years.

The tax cuts are different. They allow corporations to get some tax credits now instead of waiting for them, but eventually they have to pay the taxes they avoided. So during the first three years they save $177 billion. Then they start paying up, and over ten years the net benefit drops to $33 billion.

Which do you think is the fairest way of comparing the two? If you compare over three years, the tax cuts are five times bigger than the CTC. If you compare over ten years they're the same size.

So the question is, do you believe that companies will ever have to pay those taxes they avoided? Or will Congress inevitably extend the tax breaks? David thinks the latter, and not without good reason.

What's even worse is that after 2026, not only does this expansion of the CTC go away, but so does the expansion from the Trump tax cuts, which expires in 2025. At that point the CTC plummets, which David thinks is a problem:

Settling for less now—much less, as the JCT score makes clear—makes an unequal trade more likely in a future under divided government, which is quite possible. After 2025, the CTC is chopped in half. If this deal becomes a “tax extender” precedent, in 2026 Republicans would get their favorite business tax credits and Democrats would get a little more refundability on a $1,000 CTC.

This is all speculative, which makes it very hard to have a firm opinion. What makes it even harder is that the biggest share of the tax cuts comes from pulling in the R&D tax credit, which is generally well thought of by both liberals and conservatives. So even if it's bigger than the CTC, maybe that's not such a bad thing?

Another benefit of the bill is that it's paid for by ending the Employee Retention Credit, a widely abused corporate benefit passed during the pandemic. I have no idea why it's still around, but in any case it would be good to get rid of it.

So this is a tricky one. But there's one thing that hasn't changed: it's still widely believed that this bill has little chance of passing. So maybe it doesn't really matter much.

Peter Biskind says the golden era of cable/streaming TV is over thanks to cost cutting and a reduced taste for innovation. Maybe so. But along the way he says this:

In 2023, HBO had a whopping four shows up for outstanding drama series, including the eventual winner, “Succession,” as well as “The White Lotus,” “The Last of Us” and “House of the Dragon.”

This might seem like proof that the original disrupter is back on top. But there’s a problem: The HBO we once applauded as the avatar of quality programming no longer exists in its previous form. As a result of the current mania for mergers, WarnerMedia, which owned HBO, merged with Discovery in the deal that created Warner Bros. Discovery. In the process, its streaming arm was rebranded. HBO Max, its new moniker, was reduced to, simply, Max, a hollow handle suggesting, at best, “more” or “most” — but certainly not quality.

Can someone explain this to me? Everyone went absolutely batshit over Warner's rebranding of HBO to Max. It's inexplicable. Rebranding happens all the time. Maybe the new name is good, maybe it's bad, but it's just a name, not a destruction of all that was good and pure about the company.

And yet, that's how people took it. What's the deal with this?

Here's your quick recap on the E. Jean Carroll affair. In 2019 Carroll claimed that Donald Trump had raped her in a dressing room a couple of decades ago. Trump denied it and said a bunch of nasty things about her, so she sued him for defamation. She won and was awarded $5 million, but Trump kept right on repeating his lies. So she sued him again and the trial is now underway.

So what's his defense? Among other things, his lawyer suggested today that Carroll had benefited from Trump's slander:

Alina Habba concluded her cross-examination by suggesting that E J Carroll had not suffered because of Donald Trump’s comments — didn’t she have opportunities like having a Substack and TV appearances? Was she making more money? Was she better known?

“So, your reputation in many ways is better today isn’t it Ms Carroll?” said Habba.

What else? Well, Trump's lawyers have also suggested Carroll is a perv (she writes a sex advice column) and that all the threats and backlash she received had nothing to do with Trump. Rather, they were all from people who already hated her.

Pretty desperate, eh? But what can they do? Trump has already been found liable for sexual abuse, battery, and defamation, so those are no longer up for question. And obviously Trump said the stuff he said. What choice do his attorneys have aside from throwing mud on the walls and giving Trump a stage to mutter and roll his eyes?

Adam Ozimek points to something interesting today. It turns out that when CBO estimates the budgetary impact of increasing the number of work permits for highly-educated immigrants, it only calculates the cost—which comes to about $4 billion over ten years. But what about the benefits?

Following the population-change approach of incorporating all of the direct budgetary effects of changing the number of people in the United States, the analysts at PWBM estimate that a proposal similar to Section 80303 in H.R. 4521 would raise federal spending by $4 billion and increase federal revenues by $133 billion from 2025 to 2034, for a net decrease in the federal budget deficit of $129 billion.

Over the following decade, the difference between estimates under the conventional approach and the population-change approach is even larger, from a $74 billion increase in the budget deficit under the conventional approach to a $634 billion decrease under the population-change approach.

CBO generally doesn't include dynamic effects in their scoring, but they're allowed to do it for "major" immigration proposals. This isn't big enough to be considered major, which is why someone else had to do it. But it's still true.

Sunrise over downtown Los Angeles.