Re 1, I think we are on the same page now. I'll consider his Iraq war views as basically not strong evidence either way.
Re 2, I don't think the second part is confused, but agree it is not relevant to Trump, just strategy selection for EAs.
Re 3, this is something I'd put like ~80% credence in, and I think it is more important than most points. The 20% comes from increased volatility/unpredictability.
Re 4, I believe C, and put a very low probability on B. I think it was rational for Trump, given his info state, to believe that he lost at least one state due to illegal voting. I think the vast majority of spammy claims and cases Republicans pressed were not credible at all, and I don't know how to feel about these overall in terms of norm decay, vs. attempting to get the legal system to check a lot of potential claims quickly when you don't yet have good evidence (I oppose anyone that was knowingly making false claims/cases). I do think their worries about mail-in ballots and vulnerability to illegal voting are justified, and that there is a lot gov could do to increase justified confidence in the elections. The states close to or below 1% margins went D with mail-in votes: its not surprising mail-in votes were more D heavy, but it's easy to see why they thought they got cheated. I'm pretty sure things like the Electronic Registration Information Center don't work super well given that I received mail in ballots and political calls for places I deregistered and no longer lived.
On the part about firing, see this NBER paper. It is necessary to firing people for political reasons to increase competency, and the left purges the bureaucracy more thoroughly that the right historically, granting the left cost advantages for programs they want. The problem is choosing good programs to do.
Re: Resisting expert pressure areas I think Trump made good decisions despite expert pressure: - High confidence: Energy policy (also in Europe) + strategy for getting allies and NATO to pay more, negotiating the Abraham accords. - Medium confidence: his version of the Afghan pullout strategy (keeping Bagram), striking Soleimani, using tariffs to renegotiate trade deals (though bad execution in some areas) - Mixed: COVID: bad cuts (justified citing CFHS on the U.S. being the most prepared), bad to initially downplay, good on travel relative to experts at the time, good to do Operation Warp Speed and push for earlier scaling, inconsistent on masks, good on re-opening earlier and schools.
On Harris' record: It's fair she didn't have much influence on CA policy and wasn't in a good position to influence much in Congress either. The bills she's proposed would have cost more than $20 trillion by now, but those didn't pass and may have just been to send signals.
I agree you have no reason to take anything I say on expert interviews at face value. I think your set of views is reasonable to have given your network.
Re 5: Due to greater economic policy rationality, explicit false beliefs on climate change that are typical of many Republicans are less costly in practice than Democrat implicit false beliefs on climate trade-offs. Texas is building more clean energy capacity than basically everywhere else in the U.S. combined. Environmental reviews, lawsuits, and over regulation of nuclear power are all issues that largely come from the left and make it hard to do any construction that would reduce emissions. Because climate is a virtue signaling topic for the left, typical proposals sacrifice more value than they could hope to save due to uneconomical spending proposals and bans (e.g. on pipelines with allies, fracking, etc.) To be fair, Harris has shifted to be pro-fracking now I think, but she did propose $10 Trillion in climate spending before. We could debate the merits of the Paris agreement pull out and I agree the U.S. should be more energy efficient per capita, but fundamentally it doesn't make sense to handicap the U.S. economy more than the Chinese economy and have allies free-ride on U.S. defense spending at the same time.
Agree that NIMBYism and car culture pose big problems and conservatives can be worse on both, though as Dems control the cities and the policies that drive cost growth in them, I think they are more to blame in the worst cases. As an example, the environmental review to even look at digging another metro tunnel under the bay was set to cost a billion dollars. In the bay most of the NIMBY arguments complain about gentrification, stopping greedy developers, and protecting the environment. For national policy, Trump's head of HUD claimed to be anti-NIMBY and aimed to condition HUD fundingon local zoning reform. That said, Walz is YIMBY too, the Biden admin does seem to be trying harder to increase housing supply now, and some of the permitting form looks potentially promising provided lots of the things they add on don't become veto points. Overall, I do think conservatives will be more NIMBY in the suburbs, but will open more areas to development and lower crime in a manner that facilitates relatively more urban density.
In terms of reducing wokeness there's both policy and attitudes. A Trump admin can continue repealing policies that incentivize and force people and companies to be more woke if they want to succeed or to defend rights that have little to do with discrimination. At the same time, people being mad about Trump will increase woke reactions, so that's fair and I am not sure how things net out on polarization. Causing the far right to go nuts doesn't sound great either when they have all the guns, but either way I don't want to be held hostage by extremist reactions.
On the praise for dictators thing, I misinterpreted your comma. Disregard.
Re: JD's statements around the assassination: I directionally agree, though if we consistently apply the standard that people who publicly jumps to conclusions about responsibility in response to violent events shouldn't be in office, then I'm not sure that many presidents/VPs reach the bar.
On immigration: my understanding was that the border proposal was unacceptable because it explicitly tolerates allowing just under 5,000 people in per day via illegal border crossings rather than via border control points. If this specific claim is not true, that would substantially change my view of how bad his opposition to the border compromise is.
On AI I share the same hopes as you. I don't want ideological capture by e/accs or EAs though, because both are too myopic. I want them counter balancing each other, and I want tech acceleration mostly focused on things other than AI and narrow/harder to abuse applications of AI. I think we need substantial growth to deal with the debt burden, and generate enough value to have more positive-sum politics and foreign policy. At the same time, I think it is hard to directly attack most of the EO as stated. One issue is largely on how the involvement of the government to assure that AI increases equity will lead to a lot of negative-sum behavior and censorship that has nothing to do with safety. Thiel sometimes articulates the more extreme version of the longer-term concern in terms of authoritarianism, but that seems further off.
Overall, I agree the biggest threats and opportunities aren't necessarily right or left wing. I feel now like I have a few points on foreign policy and immigration policy that could cause me to make large updates if I find more decisive counter-evidence to my current position. I think it may take me longer to sort through cruxes/points of info that would make me decisively more fearful of dictatorship risk.
I don't think all of the examples are reliable indicators, but I agree policy changes, declarations, and executive orders are clear demonstrations of intent. On specifics:
As far as I can tell, Trump doesn't support Fuentes and vice versa (at least for now?)
A lot of the quotes are from people in conflict with Trump that gain something from going to the press. The Hitler speeches story doesn't sound credible, and the latter part is clearly a joke. I worry more about John Kelly's claims.
There are a lot of 3 word matches you can get if you dig through every quote Hitler said.
The "stigmatize opponents as Nazis" tactic wins short-term but undermines epistemics and amplifies long-term conflict by reducing info access and the viability of reasonable dissent.
People with stigmatized beliefs hide them to keep their friends, and others become more sensitive to detect them. False-positive and false-negative rates go up. False accusations polarize victims and mobs while true accusations lose credibility as people cry wolf. Some associate with stigmatized people when they don't intend to, others avoid associating with people that don't even hold stigmatized views for their correlated reasonable views.
As all want every vote they can get, each side competes to stigmatize the other, while dog whistling to extremists to gain support without spooking the center. It's a big defection trap we should step back from, and focusing directly on policy helps avoid distraction.
I think Trump would crack down on illegal immigration. I'd be unhappy if it goes beyond throwing out criminals with victims and illegal immigrants consuming more than they make. I doubt it will be as extreme as what he says, because few things ever are and it wasn't before, but I do worry about downside risk. As things are, I worry more about status quo and downside risks with Harris on immigration outcomes and policy.
I had thought I found a more credible source before and this is all I can find on search engines now. TL;DR: there isn't a public record of his opposition before the war started, and the person who claims he did oppose the war before it started primarily talks about right after it started and would have political incentives to lie or misremember. Will update the claim.
Re, coup risk: see my responses to Phib below. I agree anything coup related is extremely concerning. I understand why they did so much nonsense given their belief state, and I think it is extremely important to take action to credibly address their belief state. The right is exacerbating it's own belief state with spammy nonsense accusations, but the left is also amplifying the worst ones and ignoring their far more plausible complaints related to illegal immigration. In other words, the Dem's strategy intentionally (or via motivated laziness?) makes the risk of coup from the right worse to attract voters via fear and prevent the loss of any minor advantages derived from potential election vulnerabilities that on net favor the left. It is very much not a good situation overall, the election norms are currently in a very bad state on the left and the right, but I think there is more common knowledge of how bad things are to the right election manipulation wise than to the left. I do not think the risks are equal, and I don't want to play the both sides game.
On immigration, thanks for correcting me. Green cards are different from visas, I will correct my comment from earlier. That said, Trump walking back his promise to include a vetting process is ambiguous: on paper if you pass a background check and you get the degree, you'll still get a green card: so what matters in practice here is if he kills the process with bureaucracy (e.g. like the entrepreneur visa). On H-1Bs, increasing stringency doesn't matter much if the flows don't decrease: it just increases the expected value of those that get through at a given cap. In other words, for the cap, Trump made changes to select the best: the admin changed the lottery to prioritize those with the highest salary offers, which is both good for growth and for not driving down wages. The Trump admin basically stayed at or exceeded the Obama era H-1b maximums with the exception of COVID. The Biden admin went even lower for the second year of COVID, but I assume Biden will finally catch up to and exceed the Trump H-1b average this year via uncapped visas to Universities and NGOs?
On natalism, it is very unclear that illegal immigration will increase the domestic birthrate. Illegal immigrants can build houses but they can also occupy them. People who have their wages competed down will tend to have fewer kids, but being able to afford more may help. Ultimately I think the bigger issue is regulations more so than labor supply forcing minimum costs up, and cultural issues that leave basically all high income countries below replacement. If I recall correctly, Israel is the only high income country above replacement fertility (and now secular Israelis going below replacement too).
On competence: To steel man the Dems a bit, I do think they have a better talent base to select from (other than for some areas of business/industry) and an inherently lower alignment penalty which can reduce cost overruns in some programs. At the same time, this also generates a tremendous amount of groupthink in government relative to a Trump administration. In terms of witch hunts, you are missing the big picture: Dem admins don't need them because they can afford to thoroughly purge everyone they don't like at the start and do things more quietly. In the last Trump admin, something like 50% of the political appointees were Republican, below the 60% peak of the Bush admin. In the Obama and Clinton admins it was over 80% Democrats and supposedly around 90% for the Biden admin (though the NBER data I can find doesn't get that recent). When the ideology of the bureaucracy is so myopic and unopposed it can do worse than a few crazies that get counter balanced (Iraq War neocons may be a good counter example, but also the type of people Trump did seek to remove). Even when people with false beliefs get power, we have to consider how consequential their false beliefs are compared to the alternative. It doesn't matter if the director of EPA is biased enough to believe CO2's contribution to climate change are minimal, when overall energy and EPA policy went in a more desirable direction in the Trump admin than the Biden admin. From a long-term perspective we still need more de-regulation, to get nuclear going faster, to get environmental reviews out of the way of infrastructure, etc. Climate via warming is unlikely to be an x-risk, stricter regulations in the U.S. will often not be a net decrease in CO2 depending on how the economy shifts internationally, and reducing carbon emissions is often not going to be the most cost effective manner to mitigate harm. (I am still excited about the current permitting reform, though I think it stops very short of what is needed/further reforms that Republicans probably want more than Democrats). So perhaps my summary would be: Dems in theory should govern better due to talent advantages, but squander their talent advantage via choosing the wrong things to do in the first place due to ideology, groupthink, and bad incentives. If a Trump admin is more prepared with appointees this time and does enact reforms to fire more civil servants, A) performance incentives will improve within the bureaucracy, B) future Republican admins won't all keep facing as much of an alignment penalty as they currently have, and C) group think will likely be reduced at the margin (there is no way they can do without Democrats entirely given the talent base). Republicans nevertheless need more quality people to go into public service jobs that can tolerate living in/be able to raise families in the areas where such jobs are concentrated.
On Project 2025 and abortion, many of Heritage's recommendations are deeply misaligned with the direction Trump has taken the Republican platform. It's hard to tell from the outside what threats are real without networking in policy circles because Trump has incentive to distance himself, while Democrats have the incentive to hype up every bad thing on the Project 2025 agenda. Talking to Republicans, I don't see a national abortion ban happening, and Trump is explicitly against banning mifepristone. I almost wonder if JD is like an anti-assassination insurance policy.
I share this concern. I find the Georgia case very sketchy. There are other claims I haven't dug into as much yet.
My main observation is that he and his people really do think the election was stolen from them. In their media bubble you'll see stuff like surveys of non-U.S. citizens where double digit percentage say they vote: enough to sway some states that Trump lost if you extrapolate (but maybe this is complete disinfo?). Because they believe (whether their beliefs are accurate or not) that the democratic rules are already being hacked/broken against them, they want to get as close to cheating as possible (and some probably do want to just break rules fully). It is true that census resident counts (independent from citizenship) shape the electoral college numbers, and thus manipulating illegal immigration does directly contribute votes to whoever wins states with disproportionate shares of illegal immigrants.
I think all these forms of defection are unacceptable. We can't have the Dems playing the Rajneeshi strategy and the Republicans playing voter suppression. Dems have been serious about election security in the past, but right now Republicans at least superficially seem more serious about fixing things, because concessions from the Dems would play into the Trump narrative. OTOH, it wouldn't surprise me if the reforms proposed by Republicans to make voting easier and more secure include poison pills?
But anyway the core point I am making here is that the most important thing to do for coup proofing/election hacking is to establish more broadly/justifiably credible election integrity. There is a path to do it that Republicans will agree to, but the Dems can't for narrative reasons + the view that stronger voter ID laws are voter suppression (which they could be if changed very close to an election).
Separate from narrative, many Dems dislike the electoral college in general because it does geographically bias against what they'd get with popular voting. This is a general problem for scaling democracy in a mutually beneficial manner: smaller states/coalitions need disproportionate power to have incentive to join, or they can expect their interests to be decisively vetoed all the time. It's why the UN and EU don't do anything by population popular vote either.
I agree with several things here, however I'm still net in favor of Trump (for now) in terms of making the future better. The following response is incredibly long and detailed and so there may be some confirmation bias. If it's not too costly, I'd like to know where and why I'm wrong on whatever you think my highest priority misconceptions are. Cheers!
___
1. I agree PEPFAR was wonderful, though it doesn't seem that likely to do a ton long-term, whereas the Iraq war and the fear it has inspired in other countries about U.S. self control/risk of going to war on the basis of false info has put us on footing for war with Russia, China, North Korea, Iran, etc. which is a much greater risk. Trump did oppose the Iraq War before it started [edit: not in public/this is contested], (although there was a period before it started where he was briefly supportive in an uncertain manner). To me this reflects someone that was unusually thoughtful/changed their mind early relative to public convo (and their political party) at the time.
2. Agree Trump will likely not be great on factory farming. I do think this matters if morality gets "locked in" before too long in the future. At the same time, animals do not right now have the agency to chose to shape the future in the ways that humans do, it makes sense to focus on getting human affairs sorted first unfortunately in the same way it made sense for governments to spend money winning WW2 faster over doing famine relief for those in Axis occupied territory. The fundamental contests that must be won and made more win-win are human affairs: success here enables greater success on AI and animals. Right now, to the degree there are any health trade-offs, the veganism focus tends to make the EA coalition intellectually weaker and more politically polarized. My long-term ideal nevertheless is nutritionally complete veganism and cultivated meat, or eating animals with healthy happy lives and painless deaths.
3. Trump's rhetoric on strongmen gets blown out of proportion by media and is close to optimal for manipulating foreign dictators into accepting deals that are better for the world at large. I've seen academics that hate Trump for tribal reasons admit this. Likewise when he's not making the blunder of being belligerent without being strategic, his personality combo greatly reduces the prospect of war via mixing the right kinds of fear and assurance. Right away, the other side has greater uncertainty and fear about how Trump might respond to provocation which creates hesitancy to do anything that might start a war. On longer timelines, Trump's aim for deal making, opposition to regime change wars, and lower degree of ideological subversion provide greater assurance that a Trump status quo will be less personally existentially threatening to such leaders, their families, and their ruling coalitions... so long as they don't do really bad stuff. It is true that we didn't get new wars under Trump, I do think the strike on Qasem Soleimani was smart, and I do think his Afghanistan pull-out strategy of keeping an airbase was good in expectation, and something that the Biden admin squandered. (Though if what people say about his orders after losing the election are true, Trump may have tried to squander the pull-out on purpose for Biden, though I am not aware of irrecoverable mistakes/actual changes before Biden was in office?).
4. I agree on character mattering. I do not think Trump's character is great on some dimensions, however the dimensions that are most critical I am pretty sure are better with Trump than with Biden and Harris. There's a variety of comparisons you can make, but in general you can expect we know more of the negative info about Trump than about Biden and Harris given the media environment, so unless you have deep personal experience the default should be some degree of discounting on Trump's character flaws relative to Harris.
On competence: I think Trump has more of a world model than Biden or Harris, does more reasoning for himself, and yet also delegates more decision making. This means he doesn't bottleneck decisions as much, but when he intervenes he has a higher level reason for doing so rather than micromanaging something that people have more relevant expertise in. Trump is also more willing to fire people whether for competence or personal loyalty, but the net result is more competence in decision making relative to the incentives provided by not firing people for anything other than scandals. His biggest area of incompetence before was in not being able to get aligned appointees, and having to sacrifice competence for loyalty in some cases. What has Harris lead that has gone well? The border situation isn't going well, California isn't doing well, I can't think of any high upside outcomes from a Harris admin, but would be excited to hear some.
On character, I think Trump is just braver, more independent, and has a more valuable form of honesty despite being a chronic exaggerator. He's willing to conflict with his own party and donors more than Biden and Harris, likely in part due to living much of his life with independent wealth and not needing to bend to social pressure as much. This enables greater off the cuff bluntness, though with the trade-off that speaking off the cuff tends to result in exaggeration, imprecise communication, and more (often unintentional) dishonesty. The difference between his words in writing vs. spoken are sometimes immense in terms of the resulting impression/belief state: IMO most deception that results from Trump's statements is the result of media reporting incentives and not what you'd naturally get from watching videos or chatting to him.
With Biden or Harris and rehearsed speech from speech writers, you get a lot of deception via factually true statements with selective context and cases of confident, convincing, and precisely calibrated lies where neither Biden nor Harris realize the talking points were false because of their relative lack of world model. These sorts of things fall apart in real time conversation, which is why we don't get to see many such real-time conversations and interviews. This can cause more deception in terms of people's impressions/belief formation: taking them seriously results in worse decision making on my end and in the federal bureaucracy. They are more vulnerable to groupthink pressure from their supporters in multiple areas of policy, whereas Trump isn't as bound by narrow alliances, elites, or even by stupid populist pressures despite being fairly populist. I think that's all upside, with the potential downside of personalist dictatorship risk: though I think that risk would be much higher yet with a younger more popular candidate and the candidates of the 2028 election if Trump is not elected. (will return to this later)
On being able to deal with people honestly: When talking with people from other countries who focus on foreign policy, most experts I engaged with from ally countries thought Trump and his admin were easier to deal with than Biden. Most people I engaged with thought Trump has better deal making prospects with Russia, China, and North Korea and that it was his lack of alignment with more extreme Republican appointees/heeding ally advice that undermined negotiations. People I spoke with thought he will do better with more aligned political appointees that he is likely to have this time if he wins. The performance of the admin/Jared Kushner in forging the Abraham Accords is good evidence for this line of reasoning, and I highly encourage you to listen to Kushner's podcast with Lex if you get time. It gives a sense of how dysfunctional a lot of the diplomatic services are by default and accordingly how much value is created when new people come in looking for mutually beneficial deals with fresh eyes. You can give examples with Trump not being trustworthy toward traditional conservatives, but: A) that means he's less corruptible, and B) ultimately we aren't traditional conservatives either, and both U.S. interests and our values are further advanced by him not giving those much further to the right in his party everything they want. Similar arguments may apply for JD Vance, but I won't vouch for his honesty: he has clearly flipped around a bunch. At the same time, some of the things JD is right about are deeply unpopular (e.g. some natalism related views) and there are trade-offs to staying elected with such views. Overall I just get this really weird impression all the time from media and group conversations that Trump is awful and Biden/Harris are at least okay, but as soon as I go looking for details or chat in person with otherwise liberal individuals the impressions flip.
5. I am also worried about democratic concerns, especially in terms of freedom. That said:
What matters most to me here is reducing misaligned coercive power. Democracy where the majority merely beats down the minority is not good. When I look back at the forms of coercion I have faced and seen many dogooders face, they have almost all been from the left: social coercion, crime, higher taxes (not ones calibrated to harms), and regulations that curtail many opportunities to do good without forcing negative externalities on others. The worse things get in these ways, the more likely we are to get deeper authoritarian backlash in the future, and I'd rather face an overhyped risk now, rather than the real thing in the election after this one as people get more unhinged yet.
In terms of social coercion, I think I enjoy quite high personal wellbeing due to resisting relatively "woke" social pressures, but I know multiple people who have committed suicide either due to being more isolated due to avoiding/facing such pressure or via being driven to extremes of mental unwellness and cognitive dissonance via giving into it. Weaponized empathy and not being able to challenge potentially false claims are big problems for any partisan ideology, but in practice are more extreme on the left ATM than on the right once you select for elite popular culture and the subset of people that actually influence policy. It doesn't matter if QAnon dude on the internet says unhinged stuff when Republican appointees and their staff aren't actually as crazy or coerced into non-sensical beliefs. I'm tired of having to tread on egg shells in policy work and day to day life, and often only having freedom to speak honestly anonymously or in private face to face conversation. Despite having many left wing beliefs, it is far easier for me to speak and reason with Republicans because of where our discourse norms are. In the areas where conservatives do enforce norms, these tend on average to be more traditional which means on average they will still at least be conducive to survival in prior environments.
On rule of law: a Trump admin that keeps rolling back the regulatory state enables rule of law much more than the status quo, because the status quo involves politicized and disproportionate enforcement of a massive and inherently contradictory set of rules. The rules need to be at least somewhat gutted before people more strategically apply AI to the federal code of regulations and either fine or lock-up their political opponents for breaking all sorts of rules they could not have known about. Some of the most obvious cases of bias of this sort have been against Elon, whether it was a Delaware judge basically seizing $50B of his wealth, unreasonable inclusions and delays in SpaceX environmental reviews, or the Department of Justice bringing court cases against him for *complying with ITAR regulations against hiring non-U.S. persons that he is required to follow. I expect more freedom to do good things in a Trump admin overall, and while I don't think the change in abortion rights was great, Sam is right that it is a sunk cost.
On media and having a democracy: I am especially concerned about coercion in media, internet, and access to information. We designed the federal gov for balance of power, but there has been no such thing in the media ecosystem where on the one hand we have very harmful attention seeking races to the bottom, and at the same time relative monopolies in some domains like search engines that can mess with people's views at scale. I think Elon buying Twitter was a mess, but community notes is one of the best improvements to the info ecosystem I have seen [edit, the initiative started prior to Musk, but scaled under Musk]. I think it makes sense that he made Grok given what he saw with other language models. While conservatives can have nutty views, they are less likely to coerce you or censor you for not having the same views. People generally consume too much news from sources that are too fast, too unknowledgeable, and too pressured on certain topics to be honest in public.
(minor note) Where in Trump's Jan 6th speech does he praise dictators? I tried a bunch of searches referring to different world leaders, countries, etc. but I may have just used the wrong terms. I'm not saying your claim has no merit, but in general I consider most claims like this to be overly narrativized and not really trying to theorize much about alternative explanations of his intentions.
Re JD: I'm not surprised JD emotionally reacted in the way he did to the Trump assassination attempt. I think it's hard for the right to not have a bias toward blaming the left given the years of unusually strong rhetoric against him and claims that Trump poses authoritarian risk. Will be interesting if we ever find out a more definitive motives, but either way this behavior is so unsurprising in the current context that I don't consider it authoritarian evidence. I'm more worried about JD's interest in neoreactionary types, and while he seems to be more focused on promoting good incentives that's not a guarantee he wouldn't promote crazier ideas than what he's already willing to say.
In terms of Jan 6th itself, there's huge plausible deniability problems. Trump did send the protestors toward the Capitol, but also told them to be peaceful, follow the law, listen to police, and recorded a video telling them to stop (idk why the last wasn't released?). Correct me if I am wrong, but in the days before, he did tell the Acting Sec Def & his chief of staff that they should deploy ~10,000 National Guard, but got rebuffed that the Capitol would need to request them and opposed by the Mayor of DC in writing, so he didn't make the order until the request came (which came quite late). He could have made the order in contradiction to this pressure, but consider what would that looked like politically? (It would get branded as a coup attempt itself and basically everyone was pressuring him to avoid an overly militarized response.) Overall, motivated incompetence enabled the protestors to go farther and many on the left wanted to enhance the coup narrative. Motivated/selective incompetence to enable politically favorable violence, needs to be punished with defeat if we want long-term dictator risk to trend down. We need to stop selectively restraining police efforts to restrain political violence in general.
On Sam's other arguments you responded to:
I think the "Pivotal Act" here is just a pun and referencing how different gov may be in a Trump admin assuming that a lot of people do get fired and a lot of regulations do get pulled back. It would put us on a permanently different course, for better or worse.
I am confused on medical innovation. Trump didn't go with Thiel's guy for the FDA, but did support the "right to try" which is great.
I agree scope sensitivity matters. I don't think housing policy is a rounding error as it is a central factor in the sustainability of both the U.S. and Chinese systems. Bad zoning and housing policy makes agglomeration of talent in the U.S. far more costly than necessary, and rent-seeking competition on zoning probably worsens political ideology which in turn is a major constraint on our ability to do AI policy. We are leaving percentage points of GDP growth on the table, and that's huge both in terms of competition and trying to reduce zero-sum pressures to compete harder. Housing prices are up almost 40% since the start of the Biden admin, Trump wasn't as bad though inflation and COVID probably have a lot of the explanatory power. Either way, it's not good to have the U.S. on a low-growth, unsustainably indebted trajectory where people's political incentives grow increasingly zero-sum and less able to sustainably allocate resources toward global public goods.
On science & labor: R&D productivity has been dropping not just because of low hanging fruit being exhausted but also due to progressively worse allocation of science funding and escalating dishonesty in academia. I highly recommend reading Stuart Ritchie's book "Science Fictions" or the writing of people like Richard Hamming and others who used to work at places like Bell Labs or early day RAND. A lot of the most promising research can't be funded because the grant makers don't have deep expertise themselves or have to use superficial and hacked metrics to determine who they give money to. The scientists basically end up in a giant competition to dishonestly boost their metrics and to focus on marginal progress over larger progress. New R&D funding, using different methods of allocation, and going around the uni ideology/capture bubble offers the potential for much higher R&D productivity which is massive for welfare long term. I have some skepticism of open sourcing, but I can see how it fits with the logic here, especially if you have sufficient state capacity to do something about people trying to do bad stuff with open sourcing and want to max out positive combinatorial innovation.
On natalism: humans evolved and are not happiness machines in a vacuum: we should expect people will have higher wellbeing when they have healthy kids than when they don't. More happy productive people is good, and sustainable fertility lowers the pressure for AI as a deus ex machina to save/sustain everyone in the future.
On immigration, Trump has said he will do STEM degree [edit] greencards, and has put forward point systems before for increasing high-skill immigration. The people I know who worked on high-skill immigration in the Biden admin got knee capped by their own side (politically and via incompetence) more than by the right (though I do think Trump messed up the entrepreneur visa when he got into office the first time). Fundamentally the compromise that the population will accept for more legal immigration is with a more secure border and less crime, and Trump pushes for that, not against it. In terms of how positive illegal immigration is, it is more positive in states with decent law enforcement that creates extremely good incentives and selection pressures on illegal immigrants, whereas in sanctuary cities illegal immigrants are far more likely to consume state resources, jack up housing prices, and increase crime in a manner that exceeds their contribution to productivity. The long-term sustainable rate of immigration, both legal and illegal, is much higher with the direction the right is pushing than the direction the left is pushing at this time.
Agree there is a lot of good stuff in the Biden AI executive order. Unfortunately it is far too long and got married to a lot of nonsense for political compromise. Ultimately I think Elon is right that some of the woke nonsense would be quite likely to force AI development in an untruthful and misaligned direction. I don't think we should double down on the nonsense and polarize AI policy more than necessary. I think a lot of the good stuff is likely to come back in a Trump admin even if the whole order were cut at the start.
On doomsday machines: I agree we shouldn't do doomsday machine competition, we should aim to build fairly low risk AI systems to generate a ton of value in a manner that enables more room for compromise and win-win outcomes in the future. Agree that the tactics (open sourcing, public convo, etc.) are questionable though plausible and I find myself frustrated that people haven't seemed to sort through the overall arguments decisively. I think more honest public conversation is required to not create deeply messed-up and perversely captured AI policies, and that the regulation will need to be fairly narrow in scope or to have much better and less biased regulator human capital. Independent from AI alignment risk concern, I have reservations on too "all-in" AI Manhattan project strategies just based on economic bankruptcy if it fails and the threat of what leaders may do with such a project even if alignment is achieved.
On a final, less related note: Trump is ironically more politically centrist than Harris. His rhetoric allows him to appease the Republicans without going as socially conservative or staffing the government with as many Republican appointees. Democratic admins are increasingly one party states with extreme groupthink, whereas a Trump admin in practice will likely be far more politically balanced even if he purges a ton of people and brings in a few nuts. You can look at party affiliations of appointees over time if you are skeptical of that claim.
I very much agree that a larger area burning all at once will loft soot higher on expectation than a thinner moving flame front, because the hot gasses at the center of the burning mass essentially have nowhere to diffuse but up. That specific argument isn't really a matter of crying wolf, it was just very silly for people to claim that oil well fires could have such big effects in the first place.
That said, this it doesn't really matter in terms of the estimates here because the soot loft estimate is based on the Los Alamos model, which is also where I'd place my bet for accuracy. Also, the initial Los Alamos model was based on Atlanta's suburbs which means their outputs are highly relevant for burning in U.S. cities, and cover some of Dave's objections as well.
I think it can be productive here to taboo the word "capitalism" and to just think about trade-offs between incentive alignment and skill.
A lot of regulations prevent governments from hiring in an efficient manner, and this results in a comparative lack of talent in many areas that consultancies excel... or seem to excel (e.g. short-term metric hacking). Many of these regulations make sense from the standpoint of reducing conflicts of interest *within* the government, however these conflicts of interest and vulnerabilities are sometimes immediately recreated when outsourcing. If the regulations posed less burden to efficient hiring and work within government, it would be easier to internally develop aligned talent. Regardless of if advice comes from inside or outside of government, you want there to be competitive pressure for the advice to be good (one of the reasons gov outsources) and long-term incentive alignment on good outcomes.
One should be concerned with extractive models generally, not just in the context of capitalism. People can form extractive coalitions in business and they can do the same within government agencies as well. Policies that require more transparency from consulting firms when it is critical to avoid conflict of interest may be helpful, but there really needs to be lower cost ways to enact and enforce such regulations to prevent more morally dubious workarounds.
Thanks Linch!
Re 1, I think we are on the same page now. I'll consider his Iraq war views as basically not strong evidence either way.
Re 2, I don't think the second part is confused, but agree it is not relevant to Trump, just strategy selection for EAs.
Re 3, this is something I'd put like ~80% credence in, and I think it is more important than most points. The 20% comes from increased volatility/unpredictability.
Re 4, I believe C, and put a very low probability on B. I think it was rational for Trump, given his info state, to believe that he lost at least one state due to illegal voting. I think the vast majority of spammy claims and cases Republicans pressed were not credible at all, and I don't know how to feel about these overall in terms of norm decay, vs. attempting to get the legal system to check a lot of potential claims quickly when you don't yet have good evidence (I oppose anyone that was knowingly making false claims/cases). I do think their worries about mail-in ballots and vulnerability to illegal voting are justified, and that there is a lot gov could do to increase justified confidence in the elections. The states close to or below 1% margins went D with mail-in votes: its not surprising mail-in votes were more D heavy, but it's easy to see why they thought they got cheated. I'm pretty sure things like the Electronic Registration Information Center don't work super well given that I received mail in ballots and political calls for places I deregistered and no longer lived.
On the part about firing, see this NBER paper. It is necessary to firing people for political reasons to increase competency, and the left purges the bureaucracy more thoroughly that the right historically, granting the left cost advantages for programs they want. The problem is choosing good programs to do.
Re: Resisting expert pressure areas I think Trump made good decisions despite expert pressure:
- High confidence: Energy policy (also in Europe) + strategy for getting allies and NATO to pay more, negotiating the Abraham accords.
- Medium confidence: his version of the Afghan pullout strategy (keeping Bagram), striking Soleimani, using tariffs to renegotiate trade deals (though bad execution in some areas)
- Mixed: COVID: bad cuts (justified citing CFHS on the U.S. being the most prepared), bad to initially downplay, good on travel relative to experts at the time, good to do Operation Warp Speed and push for earlier scaling, inconsistent on masks, good on re-opening earlier and schools.
On Harris' record: It's fair she didn't have much influence on CA policy and wasn't in a good position to influence much in Congress either. The bills she's proposed would have cost more than $20 trillion by now, but those didn't pass and may have just been to send signals.
I agree you have no reason to take anything I say on expert interviews at face value. I think your set of views is reasonable to have given your network.
Re 5: Due to greater economic policy rationality, explicit false beliefs on climate change that are typical of many Republicans are less costly in practice than Democrat implicit false beliefs on climate trade-offs. Texas is building more clean energy capacity than basically everywhere else in the U.S. combined. Environmental reviews, lawsuits, and over regulation of nuclear power are all issues that largely come from the left and make it hard to do any construction that would reduce emissions. Because climate is a virtue signaling topic for the left, typical proposals sacrifice more value than they could hope to save due to uneconomical spending proposals and bans (e.g. on pipelines with allies, fracking, etc.) To be fair, Harris has shifted to be pro-fracking now I think, but she did propose $10 Trillion in climate spending before. We could debate the merits of the Paris agreement pull out and I agree the U.S. should be more energy efficient per capita, but fundamentally it doesn't make sense to handicap the U.S. economy more than the Chinese economy and have allies free-ride on U.S. defense spending at the same time.
Agree that NIMBYism and car culture pose big problems and conservatives can be worse on both, though as Dems control the cities and the policies that drive cost growth in them, I think they are more to blame in the worst cases. As an example, the environmental review to even look at digging another metro tunnel under the bay was set to cost a billion dollars. In the bay most of the NIMBY arguments complain about gentrification, stopping greedy developers, and protecting the environment. For national policy, Trump's head of HUD claimed to be anti-NIMBY and aimed to condition HUD funding on local zoning reform. That said, Walz is YIMBY too, the Biden admin does seem to be trying harder to increase housing supply now, and some of the permitting form looks potentially promising provided lots of the things they add on don't become veto points. Overall, I do think conservatives will be more NIMBY in the suburbs, but will open more areas to development and lower crime in a manner that facilitates relatively more urban density.
In terms of reducing wokeness there's both policy and attitudes. A Trump admin can continue repealing policies that incentivize and force people and companies to be more woke if they want to succeed or to defend rights that have little to do with discrimination. At the same time, people being mad about Trump will increase woke reactions, so that's fair and I am not sure how things net out on polarization. Causing the far right to go nuts doesn't sound great either when they have all the guns, but either way I don't want to be held hostage by extremist reactions.
On the praise for dictators thing, I misinterpreted your comma. Disregard.
Re: JD's statements around the assassination: I directionally agree, though if we consistently apply the standard that people who publicly jumps to conclusions about responsibility in response to violent events shouldn't be in office, then I'm not sure that many presidents/VPs reach the bar.
On immigration: my understanding was that the border proposal was unacceptable because it explicitly tolerates allowing just under 5,000 people in per day via illegal border crossings rather than via border control points. If this specific claim is not true, that would substantially change my view of how bad his opposition to the border compromise is.
On AI I share the same hopes as you. I don't want ideological capture by e/accs or EAs though, because both are too myopic. I want them counter balancing each other, and I want tech acceleration mostly focused on things other than AI and narrow/harder to abuse applications of AI. I think we need substantial growth to deal with the debt burden, and generate enough value to have more positive-sum politics and foreign policy. At the same time, I think it is hard to directly attack most of the EO as stated. One issue is largely on how the involvement of the government to assure that AI increases equity will lead to a lot of negative-sum behavior and censorship that has nothing to do with safety. Thiel sometimes articulates the more extreme version of the longer-term concern in terms of authoritarianism, but that seems further off.
Overall, I agree the biggest threats and opportunities aren't necessarily right or left wing. I feel now like I have a few points on foreign policy and immigration policy that could cause me to make large updates if I find more decisive counter-evidence to my current position. I think it may take me longer to sort through cruxes/points of info that would make me decisively more fearful of dictatorship risk.
I don't think all of the examples are reliable indicators, but I agree policy changes, declarations, and executive orders are clear demonstrations of intent. On specifics:
https://www.snopes.com/fact-check/trump-very-fine-people/
The "stigmatize opponents as Nazis" tactic wins short-term but undermines epistemics and amplifies long-term conflict by reducing info access and the viability of reasonable dissent.
People with stigmatized beliefs hide them to keep their friends, and others become more sensitive to detect them. False-positive and false-negative rates go up. False accusations polarize victims and mobs while true accusations lose credibility as people cry wolf. Some associate with stigmatized people when they don't intend to, others avoid associating with people that don't even hold stigmatized views for their correlated reasonable views.
As all want every vote they can get, each side competes to stigmatize the other, while dog whistling to extremists to gain support without spooking the center. It's a big defection trap we should step back from, and focusing directly on policy helps avoid distraction.
I think Trump would crack down on illegal immigration. I'd be unhappy if it goes beyond throwing out criminals with victims and illegal immigrants consuming more than they make. I doubt it will be as extreme as what he says, because few things ever are and it wasn't before, but I do worry about downside risk. As things are, I worry more about status quo and downside risks with Harris on immigration outcomes and policy.
That's fair. I would like to see better analysis of the responsibility chain here.
I had thought I found a more credible source before and this is all I can find on search engines now. TL;DR: there isn't a public record of his opposition before the war started, and the person who claims he did oppose the war before it started primarily talks about right after it started and would have political incentives to lie or misremember. Will update the claim.
Agree that's a strong signal!
Re, coup risk: see my responses to Phib below. I agree anything coup related is extremely concerning. I understand why they did so much nonsense given their belief state, and I think it is extremely important to take action to credibly address their belief state. The right is exacerbating it's own belief state with spammy nonsense accusations, but the left is also amplifying the worst ones and ignoring their far more plausible complaints related to illegal immigration. In other words, the Dem's strategy intentionally (or via motivated laziness?) makes the risk of coup from the right worse to attract voters via fear and prevent the loss of any minor advantages derived from potential election vulnerabilities that on net favor the left. It is very much not a good situation overall, the election norms are currently in a very bad state on the left and the right, but I think there is more common knowledge of how bad things are to the right election manipulation wise than to the left. I do not think the risks are equal, and I don't want to play the both sides game.
On immigration, thanks for correcting me. Green cards are different from visas, I will correct my comment from earlier. That said, Trump walking back his promise to include a vetting process is ambiguous: on paper if you pass a background check and you get the degree, you'll still get a green card: so what matters in practice here is if he kills the process with bureaucracy (e.g. like the entrepreneur visa). On H-1Bs, increasing stringency doesn't matter much if the flows don't decrease: it just increases the expected value of those that get through at a given cap. In other words, for the cap, Trump made changes to select the best: the admin changed the lottery to prioritize those with the highest salary offers, which is both good for growth and for not driving down wages. The Trump admin basically stayed at or exceeded the Obama era H-1b maximums with the exception of COVID. The Biden admin went even lower for the second year of COVID, but I assume Biden will finally catch up to and exceed the Trump H-1b average this year via uncapped visas to Universities and NGOs?
On natalism, it is very unclear that illegal immigration will increase the domestic birthrate. Illegal immigrants can build houses but they can also occupy them. People who have their wages competed down will tend to have fewer kids, but being able to afford more may help. Ultimately I think the bigger issue is regulations more so than labor supply forcing minimum costs up, and cultural issues that leave basically all high income countries below replacement. If I recall correctly, Israel is the only high income country above replacement fertility (and now secular Israelis going below replacement too).
On competence: To steel man the Dems a bit, I do think they have a better talent base to select from (other than for some areas of business/industry) and an inherently lower alignment penalty which can reduce cost overruns in some programs. At the same time, this also generates a tremendous amount of groupthink in government relative to a Trump administration. In terms of witch hunts, you are missing the big picture: Dem admins don't need them because they can afford to thoroughly purge everyone they don't like at the start and do things more quietly. In the last Trump admin, something like 50% of the political appointees were Republican, below the 60% peak of the Bush admin. In the Obama and Clinton admins it was over 80% Democrats and supposedly around 90% for the Biden admin (though the NBER data I can find doesn't get that recent). When the ideology of the bureaucracy is so myopic and unopposed it can do worse than a few crazies that get counter balanced (Iraq War neocons may be a good counter example, but also the type of people Trump did seek to remove). Even when people with false beliefs get power, we have to consider how consequential their false beliefs are compared to the alternative. It doesn't matter if the director of EPA is biased enough to believe CO2's contribution to climate change are minimal, when overall energy and EPA policy went in a more desirable direction in the Trump admin than the Biden admin. From a long-term perspective we still need more de-regulation, to get nuclear going faster, to get environmental reviews out of the way of infrastructure, etc. Climate via warming is unlikely to be an x-risk, stricter regulations in the U.S. will often not be a net decrease in CO2 depending on how the economy shifts internationally, and reducing carbon emissions is often not going to be the most cost effective manner to mitigate harm. (I am still excited about the current permitting reform, though I think it stops very short of what is needed/further reforms that Republicans probably want more than Democrats). So perhaps my summary would be: Dems in theory should govern better due to talent advantages, but squander their talent advantage via choosing the wrong things to do in the first place due to ideology, groupthink, and bad incentives. If a Trump admin is more prepared with appointees this time and does enact reforms to fire more civil servants, A) performance incentives will improve within the bureaucracy, B) future Republican admins won't all keep facing as much of an alignment penalty as they currently have, and C) group think will likely be reduced at the margin (there is no way they can do without Democrats entirely given the talent base). Republicans nevertheless need more quality people to go into public service jobs that can tolerate living in/be able to raise families in the areas where such jobs are concentrated.
On Project 2025 and abortion, many of Heritage's recommendations are deeply misaligned with the direction Trump has taken the Republican platform. It's hard to tell from the outside what threats are real without networking in policy circles because Trump has incentive to distance himself, while Democrats have the incentive to hype up every bad thing on the Project 2025 agenda. Talking to Republicans, I don't see a national abortion ban happening, and Trump is explicitly against banning mifepristone. I almost wonder if JD is like an anti-assassination insurance policy.
I share this concern. I find the Georgia case very sketchy. There are other claims I haven't dug into as much yet.
My main observation is that he and his people really do think the election was stolen from them. In their media bubble you'll see stuff like surveys of non-U.S. citizens where double digit percentage say they vote: enough to sway some states that Trump lost if you extrapolate (but maybe this is complete disinfo?). Because they believe (whether their beliefs are accurate or not) that the democratic rules are already being hacked/broken against them, they want to get as close to cheating as possible (and some probably do want to just break rules fully). It is true that census resident counts (independent from citizenship) shape the electoral college numbers, and thus manipulating illegal immigration does directly contribute votes to whoever wins states with disproportionate shares of illegal immigrants.
I think all these forms of defection are unacceptable. We can't have the Dems playing the Rajneeshi strategy and the Republicans playing voter suppression. Dems have been serious about election security in the past, but right now Republicans at least superficially seem more serious about fixing things, because concessions from the Dems would play into the Trump narrative. OTOH, it wouldn't surprise me if the reforms proposed by Republicans to make voting easier and more secure include poison pills?
But anyway the core point I am making here is that the most important thing to do for coup proofing/election hacking is to establish more broadly/justifiably credible election integrity. There is a path to do it that Republicans will agree to, but the Dems can't for narrative reasons + the view that stronger voter ID laws are voter suppression (which they could be if changed very close to an election).
Separate from narrative, many Dems dislike the electoral college in general because it does geographically bias against what they'd get with popular voting. This is a general problem for scaling democracy in a mutually beneficial manner: smaller states/coalitions need disproportionate power to have incentive to join, or they can expect their interests to be decisively vetoed all the time. It's why the UN and EU don't do anything by population popular vote either.
I agree with several things here, however I'm still net in favor of Trump (for now) in terms of making the future better. The following response is incredibly long and detailed and so there may be some confirmation bias. If it's not too costly, I'd like to know where and why I'm wrong on whatever you think my highest priority misconceptions are. Cheers!
___
1. I agree PEPFAR was wonderful, though it doesn't seem that likely to do a ton long-term, whereas the Iraq war and the fear it has inspired in other countries about U.S. self control/risk of going to war on the basis of false info has put us on footing for war with Russia, China, North Korea, Iran, etc. which is a much greater risk. Trump did oppose the Iraq War before it started [edit: not in public/this is contested], (although there was a period before it started where he was briefly supportive in an uncertain manner). To me this reflects someone that was unusually thoughtful/changed their mind early relative to public convo (and their political party) at the time.
2. Agree Trump will likely not be great on factory farming. I do think this matters if morality gets "locked in" before too long in the future. At the same time, animals do not right now have the agency to chose to shape the future in the ways that humans do, it makes sense to focus on getting human affairs sorted first unfortunately in the same way it made sense for governments to spend money winning WW2 faster over doing famine relief for those in Axis occupied territory. The fundamental contests that must be won and made more win-win are human affairs: success here enables greater success on AI and animals. Right now, to the degree there are any health trade-offs, the veganism focus tends to make the EA coalition intellectually weaker and more politically polarized. My long-term ideal nevertheless is nutritionally complete veganism and cultivated meat, or eating animals with healthy happy lives and painless deaths.
3. Trump's rhetoric on strongmen gets blown out of proportion by media and is close to optimal for manipulating foreign dictators into accepting deals that are better for the world at large. I've seen academics that hate Trump for tribal reasons admit this. Likewise when he's not making the blunder of being belligerent without being strategic, his personality combo greatly reduces the prospect of war via mixing the right kinds of fear and assurance. Right away, the other side has greater uncertainty and fear about how Trump might respond to provocation which creates hesitancy to do anything that might start a war. On longer timelines, Trump's aim for deal making, opposition to regime change wars, and lower degree of ideological subversion provide greater assurance that a Trump status quo will be less personally existentially threatening to such leaders, their families, and their ruling coalitions... so long as they don't do really bad stuff. It is true that we didn't get new wars under Trump, I do think the strike on Qasem Soleimani was smart, and I do think his Afghanistan pull-out strategy of keeping an airbase was good in expectation, and something that the Biden admin squandered. (Though if what people say about his orders after losing the election are true, Trump may have tried to squander the pull-out on purpose for Biden, though I am not aware of irrecoverable mistakes/actual changes before Biden was in office?).
4. I agree on character mattering. I do not think Trump's character is great on some dimensions, however the dimensions that are most critical I am pretty sure are better with Trump than with Biden and Harris. There's a variety of comparisons you can make, but in general you can expect we know more of the negative info about Trump than about Biden and Harris given the media environment, so unless you have deep personal experience the default should be some degree of discounting on Trump's character flaws relative to Harris.
With Biden or Harris and rehearsed speech from speech writers, you get a lot of deception via factually true statements with selective context and cases of confident, convincing, and precisely calibrated lies where neither Biden nor Harris realize the talking points were false because of their relative lack of world model. These sorts of things fall apart in real time conversation, which is why we don't get to see many such real-time conversations and interviews. This can cause more deception in terms of people's impressions/belief formation: taking them seriously results in worse decision making on my end and in the federal bureaucracy. They are more vulnerable to groupthink pressure from their supporters in multiple areas of policy, whereas Trump isn't as bound by narrow alliances, elites, or even by stupid populist pressures despite being fairly populist. I think that's all upside, with the potential downside of personalist dictatorship risk: though I think that risk would be much higher yet with a younger more popular candidate and the candidates of the 2028 election if Trump is not elected. (will return to this later)
5. I am also worried about democratic concerns, especially in terms of freedom. That said:
On Sam's other arguments you responded to:
On a final, less related note: Trump is ironically more politically centrist than Harris. His rhetoric allows him to appease the Republicans without going as socially conservative or staffing the government with as many Republican appointees. Democratic admins are increasingly one party states with extreme groupthink, whereas a Trump admin in practice will likely be far more politically balanced even if he purges a ton of people and brings in a few nuts. You can look at party affiliations of appointees over time if you are skeptical of that claim.
I very much agree that a larger area burning all at once will loft soot higher on expectation than a thinner moving flame front, because the hot gasses at the center of the burning mass essentially have nowhere to diffuse but up. That specific argument isn't really a matter of crying wolf, it was just very silly for people to claim that oil well fires could have such big effects in the first place.
That said, this it doesn't really matter in terms of the estimates here because the soot loft estimate is based on the Los Alamos model, which is also where I'd place my bet for accuracy. Also, the initial Los Alamos model was based on Atlanta's suburbs which means their outputs are highly relevant for burning in U.S. cities, and cover some of Dave's objections as well.
I think it can be productive here to taboo the word "capitalism" and to just think about trade-offs between incentive alignment and skill.
A lot of regulations prevent governments from hiring in an efficient manner, and this results in a comparative lack of talent in many areas that consultancies excel... or seem to excel (e.g. short-term metric hacking). Many of these regulations make sense from the standpoint of reducing conflicts of interest *within* the government, however these conflicts of interest and vulnerabilities are sometimes immediately recreated when outsourcing. If the regulations posed less burden to efficient hiring and work within government, it would be easier to internally develop aligned talent. Regardless of if advice comes from inside or outside of government, you want there to be competitive pressure for the advice to be good (one of the reasons gov outsources) and long-term incentive alignment on good outcomes.
One should be concerned with extractive models generally, not just in the context of capitalism. People can form extractive coalitions in business and they can do the same within government agencies as well. Policies that require more transparency from consulting firms when it is critical to avoid conflict of interest may be helpful, but there really needs to be lower cost ways to enact and enforce such regulations to prevent more morally dubious workarounds.