Author: Bob Weeks

  • Homicide in the States

    Homicide in the States

    Predicting state homicide rates from two factors: Political sentiment and the state’s total population.

    We tried to predict state homicide rates from two factors: how much of the vote went to Donald Trump in 2024, and the state’s total population. The relationship with population is very weak. There’s a slight hint that states with higher Trump vote shares might have higher homicide rates, but this finding is right on the edge of typical “significance” and should be taken with a grain of salt. Overall, these two factors, by themselves, do not explain much of the difference in homicide rates across states.

    Click for larger

    Regression Statistics

    Below is a table of the regression coefficients, their standard errors, t-values, p-values, and 95% confidence intervals:

    TermEstimateStd. Errort valuep valueLower CIUpper CI
    (Intercept)~0.53~2.87~0.190.853~-5.24~6.31
    x (Trump %)~10.20~5.08~2.010.051~-0.02~20.42
    size_values5.98e-086.61e-08~0.900.371-7.34e-081.93e-07

    Model Fit Statistics:

    • R-squared = ~0.084
    • Adjusted R-squared = ~0.044
    • F-statistic (df=2,46) = ~2.10 (p = 0.134)
    • Number of Observations = 49

    Detailed Explanation of the Results

    Intercept: The estimated intercept is approximately 0.53, but with a large standard error and a high p-value (0.853). This indicates that if both predictors (Trump vote share and population) were theoretically zero, the predicted homicide rate would be around 0.53 murders per 100k. However, this is not statistically distinguishable from zero.

    Share of Trump Vote (x): The coefficient is roughly 10.20 (p = 0.051), which is borderline significant at the 5% level. The positive sign suggests that, when state population is held constant, higher projected Trump vote share is associated with a higher homicide rate. Specifically, for each 1.0 increase in Trump vote share (that is, going from 0% to 100%, which is not realistic but a linear extension), the model predicts an increase of about 10 homicides per 100k. However, it’s important to note that 0.051 is slightly above the traditional 0.05 significance cutoff, so this result is borderline and should be interpreted cautiously.

    State Population (size_values): The coefficient is very small (~6.0e-08) and not statistically significant (p = 0.371). This means that once the model takes the share of Trump vote into account, there’s no clear evidence that the total population size by itself helps predict the homicide rate in a linear sense.

    Model Fit:

    • The R-squared (0.084) and adjusted R-squared (0.044) are relatively low, indicating that only around 8% of the variation in homicide rates is explained by these two predictors.
    • The overall F-statistic is not significant at conventional thresholds (p = 0.134), which further suggests caution in drawing strong conclusions from this model.

    Interpretation and Caution:

    • The data set is relatively small (49 states), and we have an NA for one case.
    • High multicollinearity or other issues could be at play (the condition number in the background is quite large), possibly because population can vary widely and might interact with many other state-level factors.
    • The borderline p-value for the Trump vote share means the result could go either way with more data or slightly different model specifications.

    In summary, with this simple linear model, we do not see strong or definitive evidence that either state population or projected Trump vote share is a powerful linear predictor of homicide rates. There is a weak (borderline) indication that higher Trump share might relate to higher homicide rates, but the effect fails to reach the conventional threshold of statistical significance at the 5% level, and the overall explanatory power of the model is low.

  • Opinion | Unmarked Vans. Secret Lists. Public Denunciations. America’s Police State Has Arrived.

    One-sentence summary: Masha Gessen argues that recent immigration enforcement tactics under the Trump administration resemble those of a secret-police state, with unchecked detentions, legal disregard, and public denunciations creating a pervasive climate of fear and surveillance.

    In this opinion piece, Masha Gessen contends that the United States is exhibiting the characteristics of a secret-police state through the escalating actions of immigration enforcement agencies under President Trump’s second term. The article opens with the disturbing video of Mahmoud Khalil’s arrest by plainclothes agents in an unmarked van – a scene reminiscent of authoritarian regimes – and highlights similar detentions of individuals like Tufts graduate student Rumeysa Ozturk and Brown professor Rasha Alawieh.

    Gessen describes a pattern of arbitrary enforcement, where even legal visa holders and U.S. citizens face detentions with no explanation or due process. Courts have issued orders to block certain deportations or removals, but the executive branch has ignored these rulings, undercutting the judiciary and legal protections. A growing number of reports detail ICE presence at schools, libraries, workplaces, and subways, leading to widespread fear among immigrant communities and a near-emptying of certain neighborhoods.

    The article emphasizes the opaque operations of the Department of Homeland Security, including revoking visa statuses without notification and detaining foreign nationals for minor infractions. Secretary of State Marco Rubio has openly celebrated the revocation of legal statuses, suggesting a political campaign of intimidation rather than lawful enforcement.

    A major concern raised by Gessen is the emergence of secret lists and citizen surveillance. The ICERAID app invites users to report suspected undocumented immigrants and even self-report in exchange for potential rewards, effectively encouraging public denunciations. Private groups like Mothers Against College Antisemitism and Betar U.S. are compiling lists of students and professors they deem antisemitic, many of whom are later detained or punished by state authorities, with little evidence provided.

    The piece closes by comparing this environment to those seen in totalitarian states, where citizens fear random targeting, surveillance, and denunciation by neighbors or strangers. The psychological toll includes anxiety, self-censorship, and isolation. Gessen warns that while some may still have the ability to speak out, the country is already functioning as a secret-police state.

    Gessen, Masha. “Opinion | Unmarked Vans. Secret Lists. Public Denunciations. America’s Police State Has Arrived.” The New York Times, 2 Apr. 2025, www.nytimes.com/2025/04/02/opinion/trump-ice-immigrants.html.

    Key takeaways:

    • Plainclothes arrests and unmarked vans evoke fear tactics of past authoritarian regimes.
    • Individuals with legal immigration status and even citizens are being detained without due process.
    • Courts have attempted to intervene but have been largely ignored by federal enforcement agencies.
    • DHS is altering visa statuses without transparency, affecting foreign students and professionals.
    • Secret lists and apps like ICERAID are enabling citizen-led surveillance and denunciation.
    • Groups with extreme ideologies are influencing state actions and targeting individuals for supposed antisemitism.
    • The current climate fosters fear, isolation, and self-censorship, echoing life under secret-police states.

    Most important quotations:

    • “It’s the unmarked cars.”
    • “We don’t give our name.”
    • “We’re looking every day for these lunatics.” – Marco Rubio
    • “Give us a person and we’ll find the infraction.”
    • “The United States has become a secret-police state. Trust me, I’ve seen it before.”

    Word count of summary: 603
    Word count of input: 2,538

    Model version: gpt-4
    Custom GPT name: Summarizer 2

  • Ivermectin Shows That Not All Science Is Worth Following

    One-sentence summary: Despite widespread interest in ivermectin as a COVID-19 treatment, many studies supporting its use are riddled with methodological flaws, data inconsistencies, and outright fabrications, revealing deeper issues within the scientific publication process.

    Ivermectin, an antiparasitic drug that is highly effective in treating conditions like river blindness and scabies, gained attention during the COVID-19 pandemic as a potential treatment or preventative measure. While dozens of studies and a few meta-analyses suggested some efficacy, deeper scrutiny by a team of volunteer researchers engaged in “forensic peer review” revealed significant problems in many of the studies cited to support ivermectin’s benefits.

    This team, which independently analyzes scientific studies for statistical anomalies, inconsistencies, and data manipulation, reviewed about 30 influential ivermectin studies. At least five were found to be so flawed that they should be retracted-one already has been. For example, a study from Egypt counted deaths that occurred before it began, and another from Lebanon featured duplicated data. Despite these issues, such studies are rarely retracted or even formally investigated, as the peer review system was overwhelmed during the pandemic and lacks the rigor to detect fraud or errors in rushed publications.

    The most concerning aspect is that the studies with the most favorable outcomes for ivermectin often turned out to be the most unreliable, while more rigorously conducted trials tended to yield inconclusive or modest results. The pandemic created a publishing environment where even flawed studies could gain widespread attention quickly, bypassing normal scrutiny and being amplified by social media.

    Furthermore, systemic issues plague scientific publishing: journals are slow to respond to critiques, authors resist transparency, and studies with questionable data are often left unchallenged. This creates a body of literature that appears scientifically legitimate but is fundamentally flawed. While not necessarily indicative of a conspiracy, the situation reflects a larger failure in managing scientific integrity. The prevalence of poor research, especially in high-stakes scenarios like a pandemic, points to a need for greater institutional oversight and a cultural shift toward skepticism and accountability in evaluating scientific claims.

    The article argues that rather than following all science unquestioningly, it is essential to assess the quality of the research, especially during times of crisis. Many flawed studies are published in obscure or predatory journals and are designed to pad academic résumés rather than advance knowledge. However, in the pandemic era, even such marginal studies can gain traction and influence public discourse, leading to widespread misinformation.

    Ultimately, the author warns that without institutional support for rigorous review and accountability, it falls to a small group of volunteers to identify and expose problematic science-an untenable situation given the stakes involved.

    Heathers, James. “Ivermectin Shows That Not All Science Is Worth Following.” The Atlantic, 23 Oct. 2021, www.theatlantic.com/science/archive/2021/10/ivermectin-research-problems/620473.

    Key takeaways:

    • Ivermectin gained popularity as a COVID-19 treatment based on a flawed body of research.
    • At least five out of 30 reviewed ivermectin studies had serious data problems or methodological flaws.
    • Studies with the most dramatic pro-ivermectin findings were often the most unreliable.
    • Peer review systems were overwhelmed during the pandemic, allowing flawed research to be published and amplified.
    • Volunteer “forensic peer reviewers” often uncover problems journals fail to catch or address.
    • Scientific publishing lacks institutional mechanisms to catch and correct flawed research effectively.
    • Poor research can influence public health discourse when it gains attention on social media.
    • The current system permits low-quality studies to be published with little accountability.

    Most important quotations:

    • “The problem is, not all science is worth following.”
    • “If five out of 30 were guaranteed to explode as soon as they entered a freeway on-ramp, you would prefer to take the bus.”
    • “Most problematic, the studies we are certain are unreliable happen to be the same ones that show ivermectin as most effective.”
    • “There is no net.”
    • “We have now reached a point where those doing systematic reviews must start by assuming that a study is fraudulent until they can have some evidence to the contrary.”
    • “An unbelievable, inaccurate study… may bubble over into the public consciousness as soon as it appears online.”
    • “It is a testament to how badly the scientific commons are managed that their products are fact-checked for the first time by a group of weary volunteers.”

    Word count of generated summary: 799
    Word count of supplied input: 2,774

  • Adult Literacy and Numeracy in Counties

    Adult Literacy and Numeracy in Counties

    Analyzing literacy and numeracy rates in counties alongside political sentiment.

    PIAAC (the Program for the International Assessment of Adult Competencies) defines different levels for adult skills in literacy and numeracy, ranging from below Level 1 up to Level 5. Level 3 is often considered a baseline or minimum desired level of proficiency for dealing with the demands of everyday life and work in a modern society. In more detail:

    Level 3 in Literacy: At Level 3, tasks typically require integrating, interpreting, or synthesizing information from dense or lengthy texts. Individuals can identify important details, make some inferences, and navigate moderate complexity in reading materials.

    Level 3 in Numeracy: At Level 3, tasks may require multiple steps and the interpretation of numerical information in various formats (e.g., tables, charts, or text). Individuals can handle somewhat more complex calculations, interpret data, and apply reasoning to solve problems in real-life contexts (budgeting, scheduling, analyzing simple graphs, etc.).

    When a chart shows the share of adults “at or above Level 3,” it means the percentage of people who reach at least this medium-range skill level. Individuals below Level 3 may struggle more with everyday tasks — like understanding multi-paragraph instructions or making sense of typical quantitative information — whereas those at or above Level 3 can manage those tasks more successfully.

    I gathered county-level data from PIAAC and plotted it along with the percent of voters in the county that chose Donald Trump in 2020. The PIAAC data is for 2017. I also asked ChatGPT to help me explain the chart. (Click chart for larger)

    These two scatterplots each show U.S. counties as bubbles, where:

    • Horizontal axis (x-axis): The percentage of votes cast for Donald Trump in 2020 for each county.
    • Vertical axis (y-axis): The percentage of adults in the county who score “at or above level 3” in either literacy (left chart) or numeracy (right chart).
    • Bubble size: Proportional to the county’s population. Larger circles represent more populous counties.

    In other words, each point corresponds to a county. The position on the x-axis shows how strongly that county voted for Trump, while the position on the y-axis shows the share of adults who have at least level-3 skills in literacy or numeracy (based on data from PIAAC). Some key takeaways:

    1. Negative relationship: In both charts, you can see that as the percentage of Trump votes increases (moving right on the x-axis), the average proportion of adults scoring at or above level 3 in literacy/numeracy tends to go down. In other words, there’s a negative association between Trump’s vote share and higher literacy/numeracy skills in these data.

    2. Wide spread but a clear pattern: While there is a cluster around the middle, many points in the lower-right quadrant indicate that counties with higher Trump vote shares often have a smaller share of adults at or above level 3. Meanwhile, counties with lower Trump vote shares (points farther left on the x-axis) tend to show higher percentages of adults meeting level-3 thresholds.

    3. Bubble sizes and population: Larger circles often appear to the left (that is, in counties where Trump’s vote share was lower) and frequently have higher literacy/numeracy. These big circles represent large, more populous counties. Conversely, many smaller circles — which represent less populous counties — are distributed across the rest of the charts, including in areas where a higher share of votes went to Trump.

    Overall, the charts illustrate that — within these data — counties that leaned more heavily toward Trump in the 2020 election are on average associated with somewhat lower proportions of adults at or above a level-3 threshold in literacy or numeracy, and vice versa.

  • Trump’s Tariffs Make Absolutely No Sense

    (Unlocked gift link included)

    One-sentence summary: Jason Furman argues that Donald Trump’s proposed “reciprocal tariffs” are based on flawed economic reasoning and would damage the U.S. economy, worsen global trade relations, and ultimately empower China.

    In this opinion piece, economist Jason Furman critiques former President Donald Trump’s expected announcement of “reciprocal tariffs,” a policy aimed at matching other countries’ tariff levels on U.S. goods. Furman argues that the rationale behind this policy is deeply flawed, starting with the misconception that trade deficits are inherently bad and stem from foreign trade barriers. He illustrates this point by likening trade to everyday exchanges, emphasizing that buying more from a partner than one sells is not inherently problematic.

    Furman explains that imports benefit American consumers through variety, quality, and affordability, and help domestic companies stay competitive by lowering input costs. He dispels the idea that tariffs influence trade deficits, citing that these imbalances are largely determined by national saving and investment rates. Because the U.S. invests more than it saves, it attracts foreign capital, which in turn drives up imports and creates a trade deficit – a sign of economic strength, not weakness.

    Reciprocal tariffs, Furman argues, would reduce both imports and exports, hurt economic growth, and risk inflation and job loss. Even without retaliatory measures from other countries, tariffs would make foreign goods more expensive and reduce foreign demand for American goods by strengthening the dollar. Furman warns that Trump’s approach is not limited to equalizing trade barriers but selectively targets products with higher foreign tariffs, ignoring areas where U.S. tariffs are already higher.

    He also critiques Trump’s claims that foreign VATs discriminate against U.S. exports, clarifying that these taxes apply equally to domestic and imported goods in those countries. Furman notes that demanding changes to VAT systems would require countries to alter foundational aspects of their tax codes – a highly unrealistic goal.

    The real economic costs are substantial: Trump’s first-term tariff hikes averaged a 1.5-point increase, while 2025 actions have already added another six points, with more expected. These could reduce GDP growth by 0.5 percentage point, raise inflation, and cost households roughly $1,000. Lower-income Americans will feel the burden most, even as potential tax cuts benefit wealthier households. The stock market has already lost $3 trillion since February due to tariff-related uncertainty.

    Furman concludes by warning of geopolitical consequences. As the U.S. retreats from global trade, China stands to benefit by deepening its ties with other countries. This shift in trade alliances could pave the way for a broader realignment of global political power, increasingly centered around China.

    Furman, Jason. “Opinion | Trump’s Tariffs Make Absolutely No Sense.” The New York Times, 31 Mar. 2025, www.nytimes.com/2025/03/31/opinion/trump-tariffs-economy.html.

    Unlocked gift link:
    https://www.nytimes.com/2025/03/31/opinion/trump-tariffs-economy.html?unlocked_article_code=1.8E4.oaZr.ZUB2kW7HA-uk&smid=url-share

    Key takeaways:

    • Trump’s proposed reciprocal tariffs are based on a misunderstanding of trade deficits and global economics.
    • Trade deficits are more influenced by investment and saving behaviors than by tariffs.
    • Imports benefit consumers and support U.S. businesses by providing cheaper inputs and greater product variety.
    • Tariffs will likely lead to less trade, higher prices, slower growth, and economic uncertainty.
    • Lower-income households will bear more of the costs, while benefits will skew toward wealthier individuals.
    • Trump’s approach selectively targets high-tariff goods while ignoring where U.S. tariffs are higher.
    • Foreign VATs are not discriminatory; changing them is implausible.
    • The policy could diminish the U.S.’s global economic role and strengthen China’s influence.

    Most important quotations:

    • “Every step in this chain of reasoning is wrong.”
    • “Imports are good, not bad.”
    • “There is generally no correlation between a country’s tariff levels and its overall trade balance.”
    • “Even if other countries don’t retaliate… the situation is still bad.”
    • “The enormous increase in business uncertainty that tariffs have engendered means anything could happen.”
    • “Lower-income families will pay a higher fraction of their income in tariffs.”
    • “Give these countries a choice between economic relations with the United States and with China, they would probably choose the latter in a heartbeat.”

    Word count of generated summary: 798
    Word count of supplied input: 1,651

    Model version used: gpt-4-turbo
    Custom GPT name: Summarizer 2

  • The Secret History of America’s Involvement in the Ukraine War

    (Unlocked gift link included)

    One-sentence summary: A secretive, high-stakes U.S.-Ukrainian military partnership forged in 2022 shaped the course of the Ukraine war, yielding early gains but later unraveling amid diverging goals, internal rivalries, and shifting American policy.

    Adam Entous’s investigation, based on over 300 interviews across multiple countries, unveils the covert partnership between the United States and Ukraine that shaped the trajectory of the war against Russia from 2022 through early 2025. The collaboration began shortly after Russia’s full-scale invasion, when top Ukrainian generals were secretly flown to Wiesbaden, Germany, where U.S. Army Europe and Africa headquartered a clandestine operation known as Task Force Dragon. There, alongside coalition allies, U.S. officers and intelligence agencies helped plan Ukraine’s defenses and counteroffensives, supplying intelligence, training, and advanced weaponry. This tight-knit alliance enabled a technologically outmatched Ukraine to hold off and even repel Russian forces in key battles, such as Kherson and Kharkiv.

    Throughout the war’s early stages, the U.S. supplied Ukraine with an unprecedented arsenal, including HIMARS, Javelins, Patriot systems, and intelligence that formed the backbone of targeting efforts. U.S. and Ukrainian officers worked side by side, planning major strikes and coordinating real-time battlefield operations. American intelligence often guided Ukrainian artillerymen, with cooperation so deep that NATO officers described the U.S. as part of the “kill chain.”

    Yet the partnership also strained under cultural misunderstandings, mismatched expectations, and diverging goals. Ukrainian leaders often viewed American support as insufficient or too cautious, while U.S. officials grew frustrated with what they saw as political interference and disorganization in Ukraine’s military command. These tensions reached a breaking point during the failed 2023 counteroffensive, when internal rivalries-particularly between Generals Zaluzhny and Syrsky-and President Zelensky’s strategic preferences led to a fragmented, ultimately ineffective campaign. American planners had urged caution and patience, but Ukraine pressed ahead, dividing resources among multiple offensives, especially focusing on Bakhmut, which yielded high Russian casualties but no strategic breakthrough.

    As the war dragged on into 2024, Ukraine’s battlefield gains dwindled. U.S. support persisted but became more conditional and measured, crossing previous red lines gradually – authorizing long-range missile strikes into Russian-held Crimea, expanding CIA intelligence-sharing, and later permitting strikes inside Russian territory itself. These decisions were often reactive, prompted by Russia’s tactical advances or fears of political fallout from perceived inaction.

    The firing of General Zaluzhny in early 2025, replaced by the more politically aligned General Syrsky, signaled a shift in Ukraine’s internal dynamics. At the same time, the partnership evolved toward greater Ukrainian autonomy in operations, though friction remained – especially as Ukraine used coalition weapons in unsanctioned offensives into Russia, breaching previously agreed-upon rules.

    Operation Lunar Hail, a covert long-range missile campaign against Russian military assets in Crimea, showcased the enduring yet fragile collaboration between Ukraine and its Western allies. Despite successes, Ukraine’s reliance on Western support – increasingly uncertain under President Trump’s incoming administration – exposed the country’s vulnerability. Trump’s victory brought immediate fears of waning U.S. commitment. Biden’s outgoing administration rushed to solidify support and authorize broader targeting flexibility, but the geopolitical future of Ukraine remained uncertain.

    Ultimately, while the U.S.-Ukraine partnership delayed Russian advances and yielded critical battlefield victories, it failed to deliver a decisive outcome. Political rivalries, mismatched ambitions, and limited resources eroded early momentum, and by early 2025, Ukraine’s territorial gains had stalled, with both sides entrenched and the war’s end nowhere in sight.

    Entous’s detailed narrative reveals how a secret alliance, forged in crisis, became both a tool of resilience and a mirror of its participants’ contradictions, ambitions, and limits.

    Entous, Adam. “The Secret History of America’s Involvement in the Ukraine War.” The New York Times, 31 Mar. 2025. www.nytimes.com/interactive/2025/03/29/world/europe/us-ukraine-military-war-wiesbaden.html

    Unlocked gift link:
    https://www.nytimes.com/interactive/2025/03/29/world/europe/us-ukraine-military-war-wiesbaden.html?unlocked_article_code=1.8E4.M1gJ.ynPXpFKOkCCX&smid=url-share

    Key takeaways:

    • A secret U.S.-Ukraine military partnership was launched in Wiesbaden in spring 2022 and became central to Ukraine’s battlefield strategy.
    • American intelligence, training, and weapons enabled Ukraine’s early successes against Russian forces.
    • Strategic divergences and political rivalries in Ukraine weakened the effectiveness of 2023’s counteroffensive.
    • Over time, the U.S. relaxed multiple red lines, eventually allowing missile and intelligence support for operations inside Russian territory.
    • Internal Ukrainian political dynamics, particularly the rivalry between Generals Zaluzhny and Syrsky, further complicated the war effort.
    • President Trump’s election in 2024 created uncertainty around continued U.S. support, while Biden’s administration sought to secure short-term gains.
    • Despite operational achievements, by early 2025 the war had reached a stalemate, with both sides trading small territorial gains.

    Notable quotations:

    • “They are part of the kill chain now.” – European intelligence chief on U.S. integration into Ukrainian military operations.
    • “We told them, ‘The Russians are coming – see ya.’” – U.S. military official, recalling the pre-invasion withdrawal.
    • “You can ‘Slava Ukraini’ all you want with other people… Look at the numbers.” – Gen. Donahue to Ukrainian counterparts.
    • “We should have walked away.” – Senior American official on the failed 2023 counteroffensive.
    • “From where?” – Gen. Cavoli in response to Ukrainian requests for massive supplies.
    • “It’s not an existential war if they won’t make their people fight.” – American official on Ukraine’s conscription hesitancy.
    • “What is going to happen if President Trump wins?” – Ukrainian leaders’ recurring concern.

    Word counts:
    Generated summary: 1,181 words
    Original article: Approx. 17,300 words

    Model version: GPT-4
    Custom GPT: Summarizer 2

  • For God’s Sake, Fellow Lawyers, Stand Up to Trump

    One-sentence summary: Three prominent attorneys call on the legal profession to defend the Constitution and the rule of law by resisting President Trump’s executive orders targeting law firms that have opposed him.

    In this opinion piece, attorneys John W. Keker, Robert A. Van Nest, and Elliot R. Peters condemn a recent executive order issued by President Trump that they argue is designed to punish the law firm Perkins Coie for its past representation of Hillary Clinton and other politically disfavored clients. They assert that the order is blatantly unconstitutional, violating the First, Fifth, and Sixth Amendments, and call attention to a federal judge’s swift action to temporarily block most of it.

    The authors explain that Trump’s order against Perkins Coie is part of a broader campaign, as he has issued similar executive orders targeting other prominent firms, such as WilmerHale and Jenner & Block, which have also taken legal action in response. These actions appear to be part of an attempt to intimidate the legal community and discourage representation of clients who challenge Trump.

    The authors sharply criticize the decision of Paul, Weiss — a traditionally justice-oriented firm also targeted by one of Trump’s orders — for capitulating rather than resisting. They claim that this response not only undermines the legal profession but also empowers Trump’s broader attack on the legal system. The authors express concern that other firms have hesitated to oppose Trump’s actions due to fear of political and financial retaliation.

    They urge the legal community to remember its professional obligations and moral duty to stand up for the rule of law, arguing that in the face of an increasingly autocratic executive, the courts and the lawyers who practice in them may be the last line of defense for American democracy. The authors emphasize that Trump’s attacks on lawyers, judges, and legal institutions reflect a belief in unchecked executive power — a threat they believe the legal profession must confront with unity and courage.

    Keker, John W., et al. “Opinion | For God’s Sake, Fellow Lawyers, Stand Up to Trump.” The New York Times, 30 Mar. 2025, www.nytimes.com/2025/03/30/opinion/perkins-coie-trump.html.

    Key takeaways:

    • President Trump issued an executive order targeting Perkins Coie, allegedly for its past legal work against him and his interests.
    • The order is widely viewed as unconstitutional and has been partially blocked by a federal judge.
    • Trump has issued similar orders against other law firms such as WilmerHale and Jenner & Block.
    • Some law firms, like Paul, Weiss and reportedly Skadden, have chosen not to resist and instead made concessions.
    • The authors call on lawyers and firms to defend the legal profession, constitutional principles, and the rule of law, even at potential personal or financial cost.
    • The authors see the legal community as a crucial bulwark against rising autocratic tendencies in the Trump administration.

    Most important quotations:

    • “It could not have been more blatantly unconstitutional than if a legal scholar had been asked to draft a template for an unlawful executive order.”
    • “If lawyers and law firms won’t stand up for the rule of law, who will?”
    • “Paul, Weiss’s choice was particularly disappointing because it further empowered Mr. Trump’s attack on our profession.”
    • “The common denominator among the president’s recent spate of actions is that he appears to believe he has absolute authority to govern by fiat.”
    • “Lawyers and big firms: For God’s sake, stand up for the legal profession, and for the Constitution.”

    Word count of summary: 603
    Word count of original input: 1,119

    Model version: gpt-4
    Custom GPT name: Summarizer 2

  • Federally Dependent States

    Federally Dependent States

    Examining state political sentiment and dependence on federal government.

    How dependent are states on federal government for funding their governments economies? Wallethub has collected data, which it explains as follows:

    States receive federal aid for many reasons, from providing relief during natural disasters and health crises to funding improvements in education, transportation, infrastructure, healthcare and more. Some states receive much larger aid packages than others, but it’s not just the dollar amount that matters. It’s important to contextualize the money flowing in by comparing it to things like what percentage of the state’s revenue it makes up and how much the federal government gets back through its taxes on the state’s residents.

    In order to find out exactly how big the difference in federal dependence is from state to state, WalletHub compared the 50 states in terms of three key metrics: the return on taxes paid to the federal government, the share of federal jobs, and federal funding as a share of state revenue. (1)https://wallethub.com/edu/states-most-least-dependent-on-the-federal-government/2700

    I created a chart, and asked ChatGPT to explain. (Click chart for larger)

    This chart plots U.S. states by two main variables:

    1. Federal Dependency Score (vertical axis):

    – A higher “total score” on this axis means the state is more dependent on federal funding (as measured by the study).

    1. Trump Vote in 2024 (horizontal axis):

    – This is the estimated/actual percentage of the vote for Donald Trump in the 2024 election. Farther right indicates a higher Trump share.

    Additionally, circle size corresponds to each state’s population — larger states are shown with bigger bubbles.

    In short, the chart visually represents a relationship in which many states that vote more heavily for Trump also score higher on measures of federal financial dependence. Meanwhile, states that lean more Democratic in 2024 (to the left side) tend to have lower federal dependency scores, although there are exceptions in both directions.

    General Pattern

    1. Overall correlation
      There is a visible upward-sloping trend line (roughly from bottom-left to top-right).

    – This suggests that, as a state’s share of the vote for Trump (2024) increases, so does its likelihood of having a higher federal dependency score.

    1. Clusters of states

    Lower federal-dependency / lower Trump vote (bottom-left): States such as Massachusetts, California, and New Jersey appear here with both low federal-dependency scores and lower Trump support.
    Higher federal-dependency / higher Trump vote (top-right): States such as Mississippi, Kentucky, West Virginia, and Alaska appear with both high federal-dependency scores and a high percentage of Trump votes.

    1. Mid-range

    – Around the center, you find states like Michigan, North Carolina, Pennsylvania, Florida, and Texas—moderate-to-high Trump vote shares but with moderately high federal-dependency scores.

    Notable Outliers

    • Hawaii and Vermont appear relatively high on the dependency scale (above the mid-40s or 50s on the vertical axis) but still have lower Trump vote percentages.
    • Alaska is an extreme case in the top-left corner (high dependency but not the absolute highest Trump vote).
    • Utah and the Dakotas are further right on the x-axis (indicating higher Trump vote), but their vertical position isn’t as high as some other strongly Republican states.

    Possible Explanation of Trends

    • Economic Composition: States with larger proportions of federal land, defense spending, or social welfare needs (for instance, states with higher poverty or smaller tax bases) may rely more on federal support. These states (e.g., Alaska, West Virginia, Mississippi) often have strong Republican leanings but also benefit from federal dollars.
    • Political History: Some states have historically voted Republican and have industries (mining, agriculture, defense installations) that draw significant federal funding—leading to higher dependency scores.
    • Population Size: Larger states (California, Texas, Florida, New York) have large bubbles. Even if their “per capita” or “per dollar” federal dependency is moderate, the total funds are substantial. But in this chart, total state dependency is measured by the Wallethub scoring system—which typically weighs factors like grants, return on taxes paid, etc.

    References

    References
    1https://wallethub.com/edu/states-most-least-dependent-on-the-federal-government/2700
  • State Tax Collections

    State Tax Collections

    Using data from Tax Foundation, I examined state political sentiment and state tax collections.

    One way to measure the political sentiment of a state is the proportion voting for Donald J. Trump. The Tax Foundation provides data on the total value of state government tax collections, and also per capita (state residents). This data is for fiscal year 2023.

    I plotted the data, using Trump vote as an independent variable shown on the horizontal axis, and tax collections per resident on the vertical axis. The size of the marks is proportional to the state population. (click for larger)

    I asked ChatGPT to perform a regression analysis and explain the results. I also included the state population as an independent variable. Following is a short summary of findings:

    This analysis looked at whether two factors — how strongly a state voted for Donald Trump in 2020 and the state’s total population — help explain differences in how much tax the state collects per person. We found that states with higher Trump vote shares tend to collect less in taxes per resident on average. The population of the state didn’t show a clear or consistent effect on tax collections per resident. Overall, these two factors together explain roughly one-quarter of the differences in how much states collect in taxes per person, meaning many other factors also matter.

    In more detail:

    Residuals:
         Min       1Q   Median       3Q      Max 
    -2681.42  -985.63    42.02  1016.91  2686.93 
    
    Coefficients:
                   Estimate Std. Error t value Pr(>|t|)    
    (Intercept)     7857.18    924.98   8.49  1.06e-11 ***
    x             -6139.42   1657.40  -3.70  0.00061 ***
    size_values       -0.00      0.00  -1.65  0.10600    
    ---
    Signif. codes:  0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
    
    Residual standard error: 1398 on 47 degrees of freedom
    Multiple R-squared: 0.236,  Adjusted R-squared: 0.203 
    F-statistic: 7.258 on 2 and 47 DF,  p-value: 0.00179

    Key Points:

    1. Intercept (7857.18)
      • Interpreted as the expected tax collections per resident when both the Trump vote share (x) is 0 and the population (size_values) is 0.
      • Although literal interpretation of “population = 0 and Trump vote share = 0” doesn’t make practical sense for real states, the intercept is still the baseline from which changes in x or size_values start.
    2. Coefficient for x (-6139.42)
      • This is a strong negative relationship with the dependent variable (y), statistically significant at well below the 1% level (p < 0.01).
      • Because x ranges between 0 and 1 (representing 0% to 100% of the Trump vote share in a state), a difference of 0.10 (i.e., 10 percentage points in Trump vote share) would, all else equal, correspond to a decrease of about 614 dollars of tax collections per resident (-6139.42 * 0.10).
      • In other words, states with higher Trump vote shares tend to have lower state tax collections per resident, all else being equal.
    3. Coefficient for size_values (-0.0000369)
      • The effect of population is negative but not statistically significant at the 5% level (p ? 0.106).
      • That means we do not have strong evidence that population size on its own (once we also include x in the model) clearly predicts higher or lower per-resident tax collections. It might be mildly negative, but the data aren’t conclusive enough to confirm.
    4. Model Fit (R-squared = 0.236)
      • About 23.6% of the variation in state tax collections per resident is explained by these two variables combined.
      • This is a modest amount, indicating that while there is some relationship, other factors also play a significant role in determining state tax collections per resident.
    5. Overall Model Significance (p = 0.00179)
      • Jointly, the two predictors (x and size_values) are statistically significant in explaining variation in y.
      • Within the model, the Trump vote share is strongly significant, while population is not as definitive.