While I am obviously biased: I highly encourage you to download the game, play a bit, give it a shot, decide if it is for you.
This is especially helpful to us now, while we’re fighting for critical mass.
I have been working on this game for over four years. From the beginning I knew we had something special. We assembled a true dream team and I think we knocked this out of the park.
Some of the game’s unique features: The modified turn order and letting players pre-play actions and traps ensures games flow smoothly while maintaining interaction. Selling cards on bonding curves via smart contracts on Tezos, and offering to buy them back, lets players buy exactly what they want without locking them in, and letting prices move will encourage card and deck variety. Cards will be able to go on their own quests to get chromatic upgrades. Brian David-Marshall gave the game the wildcard system, where you can attempt to play cards without paying for them, allowing us to have a unique resource system that preserves interesting decisions and keeping games unique without creating non-games. Oh, and he also brought along an entire fully formed comic book universe to build the game around.
When I play games today with decks full of only free cards, I still have a blast.
All right, announcement over. On to the miscellaneous news of the month.
The equilibrium is that we will auction off ADHD medication this way, see who is willing and able to make the calls. Which is a problem, both because this is a rather destructive auction, and also only the people able to focus enough to make all the phone calls will find their medication, which is supposed to treat ADHD.
I increasingly see reports that Amazon listings include more and more fake or terrible products. This is the latest one, which also proposes a solution.
My experience has been like River’s. If you go looking for the cheapest thing listed and don’t otherwise filter, you’re going to select for a lot of shoddy or fake goods. Don’t do that. It’s unfortunate you can no longer play in easy mode and save the maximum. To do that you need to do more work and pay attention. There is however an effective easy mode, which is to use Amazon’s Choice. That’s like buying insurance.
Generally Twitter is very good about not spoiling sports events with their ‘what’s happening.’ Monday the 13th, they broke that rule:
Very disappointing. I mean, yes, if you are on Twitter hoping to not know who won last night’s Super Bowl, good luck with that anyway. Still feels like the breaking of a sacred trust. I was going to tell Elon to fix this but I need to save my powder for (see AI Post #2) telling him to not doing the worst possible thing. So, never mind.
The question one must then ask is: Would informing the public about such dangers, or otherwise making them predictably less inclined to retaliate against Russian aggression, increase or decrease the long term risks of destruction from nuclear war? My model of Russian behavior, and of Putin’s behavior, and of the decision theory and game theory here, says that our willingness to retaliate is vital to preventing nuclear war. I don’t love the plan, the alternative still seems way worse. Sometimes it is better to lower awareness.
From March 2: China reports human case of H5N1. Only one case, on its own, still being news is good news. It means this is important enough to be news, and that it got reported rather than suppressed.
Remember you always, always have to pay for marketing, sales and the availability of the service, not only the marginal cost of the service provided. Also this is one of the reasons you probably aren’t charging enough.
Michelin Stars make restaurants more snobby, as basic theory would expect. Menu description length increases, references to cooking techniques go up, prices rise. If anything, given how hard it can be to get a reservation, such restaurants don’t charge enough. If something increases demand without increasing supply, price had better rise. The increased snobbery is unfortunate for those of us who want great food without the snobbery, but the incentives are obvious.
Michelin Guide’s 2005 entry into the New York City restaurant market serves as the setting for examining these issues empirically. Statistical analyses indicate that restaurants receiving Michelin stars have increased likelihood of failure. A mixed-method approach, including interviews with Michelin started restaurateurs, is used to investigate mechanisms at the upstream, downstream, and organizational levels. The evidence suggests that changes in attention and bargaining at each of these levels makes it more difficult for the restaurant to capture value after receiving a Michelin star.
The paper’s claim is that third party evaluators refocus attention in potentially damaging ways. If everyone is obsessed with star status, they are focused on something other than actually creating value, and it will show. This certainly matches what I know about such restaurants. You want to be a place that happens to have a star for doing what you would have done anyway. Beware Goodhart’s Law.
Public attention to a firm may provide valuable monitoring, but it may also have a dark side by constraining management’s decisions and distracting it. We use inclusion in the S&P 500 index as a positive shock to public attention. Media coverage, Google searches, SEC downloads, SEC comment letters, shareholder proposals, analyst coverage, and lawsuits increase following inclusion. Post-inclusion performance falls and is negatively related to the increase in attention. Included firms’ investment and payout policies become more similar to those of index peers and the increase in similarity is positively related to the size of the attention increase.
Although, wait a minute. Inclusion in the S&P 500 also means tons of index funds have to buy your stock. It is somewhat odd to use that as your triggering event and assume that public attention is the primary effect on future performance. One could also say that if a growing company has more attention then it also likely has more hype, so perhaps they are simply overvalued, whereas companies that stay under the radar are better buys because no one else is buying them, not because they are run better.
The effects described still make a lot of sense as a causal mechanism. Once you ‘hit the big time’ and need to satisfy a bunch of prying eyes, you stop doing what makes sense for you to do, and start doing what you want to be seen as doing by others. Which means standard procedures, whether or not they make sense in general or for you.
There is always a balance to be struck here. Too little scrutiny and attention means no discipline or constraints on behavior, and no alert when things are going wrong. Too much scrutiny, and your focus is on all the wrong things. Once again, beware Goodhart’s Law.
This seems related – report that announcements of layoffs do not predict changes in payrolls. So what are they for, exactly? Presumably, in large part, for outside observers.
This is a great example of the Bounded Distrust principles, where they are careful not to make false statements while advancing a narrative that does not fit the data and acting as if it was conclusively correct.
Compare these two graphs above versus below, and pay attention to state lines.
If their theories were right, we would see state lines show up on the credit score graph. We don’t. We would see the insurance graph and the credit graph match. They don’t.
In particular, if you look at Wisconsin, Louisiana, Kentucky and West Virginia on their own graphs, you see zero or almost zero impact when crossing over state lines where these policies change dramatically. This hypothesis seems clearly false.
Advice from Arnold Kling not to hire disgruntled employees. You want them to have things they disliked about their previous employer, passive is bad, you want them to want things to be better. Ideal is what Kling calls serene, with strong preferences for better choices and constructive ideas of how to improve. But you don’t want someone who has nothing positive to say, that will constantly find reasons to be unhappy without offering you actionable ways to fix things. This seems eminently reasonable.
Why do East Asian nations tend to be less happy than their income levels would predict? Hypothesis is that, while social connection is good for happiness, comparisons are toxic, especially when they involve money. An experiment is run comparing wheat areas versus rice areas in China, including places where farmers were randomly assigned a crop. Rice areas and farms, which are more like the rest of East Asia, have people making more comparisons as people are forced to work together and monitor each other, whereas wheat areas have less, and the wheat farmers and others in such areas are happier. Interesting. I do know that comparing yourself to others is, in terms of happiness, clearly a fool’s game.
I do want to notice an alternative hypothesis, which is that the other thing that is special about rice areas of East Asia is that they have quite a lot of people packed into remarkably little area, especially once you control for their level of wealth.
It is not a great leap to suggest this could be part of the (relatively modest) happiness shortfall we observe.
Pirate Wires notices ‘a staggering wave of grand theft auto.’ It turns out that it is very very easy to steal a manual transmission Kia made between 2011 and 2021, or a Hyundai manufactured between 2015 and 2021. Post suggests that, contra journalists, the fault lies not in TikTok for spreading the word, and not with the auto manufacturers, but with the people stealing the cars.
I am going to go ahead and primarily blame the car manufacturers. This was spectacularly bad design. If you make it trivial to steal your brand of car without triggering the alarm, once word gets out a lot of people are going to steal your brand of car. I would like to live in a world where that is not true, and I’m not not blaming the people stealing the cars. I’m also on board with cops arresting people who steal cars to take on joy rides, and throwing whoever did that into jail long enough to make everyone think twice about such behavior.
This is still happening for a very obvious reason. If I owned one of these cars, I would either find a way to fix the vulnerability or get a different car. If I was either manufacturer, I would do my best to figure out how to fix the vulnerability and offer this for free or at least at cost.
Manifold Markets got a scare that Stripe would do the thing where the card processor kills one’s business model entirely for no reason, but then they decided not to, so M$ will continue to have value. This kind of thing does seem to be increasing in frequency.
Food delivery in China is excellent and growing rapidly. I am also greatly impressed by the American version – I use the Caviar app, and as long as you spend $100 a month you get a 10% off unlimited-use code, plus Chase Sapphire card reduces fees, so basically you tip the dasher instead of tipping the waitress and you’re good.
At least one in six employees is lying: Bloomberg reports on the world’s largest four-day work week trial, finds few are going back and about one in six employees said no amount of money would convince them to return five days a week. Somehow I do not think that someone still willing to come in four days a week has no price on returning to their old work habits. I mean, I don’t want to go into an office or be fully ‘at work’ for five days each week, and it would likely be unwise to pay my price for either – but I assume you that the price exists and there are people who can afford it.
The pandemic caused a drop in cancer screening. People predicted a resulting increase in cancer. That didn’t happen. They are now falling back on ‘it could take years’ to see the impact. Have we considered the null hypothesis?
Earnings in Finland (and America) scale with cognitive ability and don’t plateau the way they were found to do so in Sweden. They keep going. I have seen a lot of smart sources use the Swedish finding of a plateau, which always seemed counter-intuitive and weird, as important load bearing in their models of work and intelligence and the economy exactly because it was so counter-intuitive and weird.
Could it be this simple?
The smartest people, perhaps, know better than to work for a salary.
Devil Parikh shares her method of using her calendar as a to-do list. Of course a Facebook AI researcher would have a method of time management and avoiding stress that is the exact opposite of what I imagine would ever work for a human, certainly for me, but hey. Works for her.
Theory of online dating as too efficient a market. If someone is ‘out of your league’ in real life, it’s a stretch, but you’ve got a shot because there are only so many options, and perhaps there is an especially good match here and fortune will favor the bold. There is a multi-dimensional less-populated space. If someone too good messages you on a dating app, that means there’s a catch, because the space has few dimensions and is well-populated. Maybe not the old OK Cupid, but any modern ‘swipe’ app works like this.
This cuts both ways, though. For everyone that ‘dates up’ someone else ‘dates down.’ Yes, you might be the one who wins the jackpot, but you also might be the jackpot someone wins. Expectation should be the same.
This also suggests the already-wise strategy of maximum differentiation. If everyone else is acting generically, and anyone who is a little above someone’s median stands out a lot, you can essentially offer to ‘date down’ a bit online and use that to let you aggressively filter for other things you want. Trying to get more of what others in the pool want is the wrong strategy here.
New York Post reports that an epilepsy drug appears to be able to ‘switch off’ autism symptoms in mice (study). I don’t have a terribly large amount of hope that this will work out to anything useful, still interesting.
Then, there are the large numbers of deals you’d like to do. You love the CEO, it’s just … the market isn’t really there. The numbers aren’t really there. The team isn’t really there. It’s Good, but not Great.
You want to believe.
It stacks up. You should pass, but you want to see just one more proof point, so you can believe. Maybe, by mulling it over, you’ll see that one more proof point.
Have 8-10 meetings like this, and without an amazing task management system (which clearly exist, but …) … you do get behind. If you meet with say 200-300 great entrepreneurs a year, you can be overwhelmed with the Almosts.
It’s not an excuse. But the reason is, for me at least, it’s the Almosts.
I’m sorry. I will try to do better.
The reality is … it’s annoying. I didn’t let it bother me as a founder. Because if a VC really wants to do a deal … you’ll know.
If one takes this logic seriously, getting a hard no that isn’t purely about some sort of clean mismatch is very bad news. Both for a job applicant and for a founder seeking funding. If you get strung along, you can both interpret that as a no that is likely a close no. It’s a no that could have been a yes, or that is at least a no for reasons they don’t have a good way to explain. The alternative is that they are simply doing this to everyone either way, which seems rather plausible to me, and it means nothing.
The essential reason that VCs and employers do this, stringing the almosts along, is that they have no incentive not to do it. They keep option value, and they don’t lose reputation. If we lived in a world where it got out who did this and who didn’t and it impacted who got the best applicants, hires and deals, they’d straighten up real quick. People simply don’t care enough to compile that information.
This leaves the question of when this strategy is still a mistake.
Suppose you have a clear almost. You can either:
Say nothing. String them along.
Tell them they are a no.
Tell them they are an almost.
Tell them they are an almost and what would turn that into a yes.
If you can do option #4, even if you have to raise the bar a bit due to having told them what you are looking for, that seems like the best option by far. It isn’t done. In my model this is because VCs are testing for ability to pass future VC tests, which also won’t give out this info, so they can’t either. My hunch is that this too is a mistake, and you can be a huge value add by providing this info.
The risk of telling people their almost status is that this can potentially cause a bunch of time wasting as they try to turn almost into yes, or that this causes adverse selection. It perhaps makes it harder to get into the deals that turn out to be popular and that you want, and makes it more likely you end up stuck with ones you didn’t want. This again is because your whole goal was to test the founders for their ability to pass VC tests – you want to be a bank and only fund them when they don’t need you.
Telling them no outright closes off your option. Which might actually be a win. You shouldn’t be funding the almost cases, you likely don’t want to waste time on their further pitches this round. Instead, you perhaps buy some goodwill for their next company or round and you avoid the distraction.
Another possibility is that people who say ‘I just want the truth’ are wrong. Perhaps they don’t want the no because it is demoralizing. It takes away hope. It restricts the stories they can tell other investors and their employees. It makes them mad at you for saying no and perhaps that is flat out worse than the anger of being strung along. In that case, the whole thing turns out to be efficient and correct.
Why can’t you trust the media? Because the media is and always has been fundamentally untrustworthy, says Matt Yglesias. The difference is that now the media is competitive enough and trustworthy enough, and we have good enough other means of sharing information and checking against reality, to notice it is unworthy of our trust. So essentially The Media has never been more trustworthy, yet we trust it less. This is part of a pattern of a bunch of things where:
Things used to be completely terrible in some way.
Things, in that way, are actually getting better, have been for a long time.
Our standards, however, have risen even faster.
Our willingness to compromise, to keep things running, has dropped.
We’re not fully wrong about all that, things aren’t exactly not terrible.
Making things fully not terrible as fast as possible is not a practical option.
It would still be good to notice things are rapidly getting better.
Another clean example is global poverty and global inequality. There are a lot of other such problems.
There are also problems that are in fact getting worse, often with power denying that there is a problem at all or that it is getting worse. It is important to notice the difference between these two groups of problems.
Forbes list of the largest 100 charities. Lots of food aid and disaster relief, lots of medicine. A bunch of religion. Huge emphasis on things one can feel good about supporting. This is not the most efficient list of places to give, yet mostly I am happy about the destination of the average tzedakah box. Some highlights of a sort: First cultural institution that isn’t religious is Smithsonian at #49. Largest orgs that represent a political perspective are Planned Parenthood at #30 and what remains of the ACLU at #45. Unless you count St. Jude’s Research Hospital (which is #3), the first research or scientific institution is the Michael J. Fox Foundation for Parkinson’s Research at #55, the closest thing here to an Effective Altruist cause area. Largest Environment/Animal charity by far is Nature Conservatory at #15, followed by World Wildlife Fund at #46, nothing I could see related to climate. PBS is #92, and if I had money to donate only to this list I’d likely put it there.
David Friedman provides suggestions for what we might do to improve law schools if the purpose of them was to teach students how best to practice law.
Biden Administration stays true to form, moves to require seeing a doctor in person before getting ADD medication or addictive painkillers. For addictive painkillers this seems appropriate, given how people interact with them in practice. For ADD medication, it depends on whether they are also rationing supply well below demand. If it was possible to go to a pharmacy with a prescription and reliably walk out with your medication, then requiring these in-person visits would effectively be rent extraction from the medical cartel. If instead you already need to call something like thirty pharmacies to find your medication, then we are currently rationing via time spent. It is a strict improvement to go to a hybrid of time spent and money paid over time spent.
It is highly plausible that these letters are great. This still seems like the wrong calculation. If you want to measure the costs and benefits, you need to include costs other than mailing the letter. Who do we think pays for the flu shot? Presumably the state of Denmark. The patient not paying does not mean that it is free. The time spent getting the shot, and the expenses of creating, distributing and administering the shot, are very real. Similarly, the social cost of mailing someone a letter is primarily that they have to spend time to process the letter, rather than the cost of mail.
My guess is this is still a great deal. Important to get the math right anyway.
Only a few days ago I said I expected Trump’s campaign, to the extent he had one, to be focused on the past. Instead, hello bold future? Including building new cities.
Trump says he would host a contest for the public to design and then build “Freedom Cities” on a small portion of federal land to “reopen the frontier, reignite American imagination, and give hundreds of thousands of young people and other people, all hardworking families, a new shot at home ownership and in fact, the American Dream.”
This is actually great. We should absolutely be building entirely new cities. It would be even better to build more housing in places people want to live. Thing is, even if we did that, we’d still want to be also building entirely new cities. It’s pretty great.
Some interesting details here. Petitions and targeting incidents both roughly doubled from 2019 to their peaks, but petitions peaked in 2020 while targeting, which is more individual, peaked in 2021. Sanctions peaked in 2021. The ‘success rate’ of targets and petitions seems to have been declining before the peak hit. Could be that some of this is a reporting issue – things that would have been ignored are now reported and recorded. Could also be that many more flimsy attempts got launched.
The decline could be that the fever is passing, or it could be that the low-hanging fruit of terrible people got picked or that people ‘cleaned up their act’ either by no longer being terrible or learning to behave and not violate the new rules, or that institutions found ways to disguise their actions.
Terminations in 2021 were already back near their 2019 levels and lower than 2017, although that could be noise.
Lots of room for this to mean different things and to be any combination of good and bad news.
I am willing to conclude that there was at least a local peak in lived experience of wokeness. That does not automatically mean a long term decline is on its way, or that the underlying dynamics have reversed course rather than the fading away of temporary effects from Trump and 2020. That could still go either way. I do think the number of identical-looking graphs do conclusively show that the local peak was real.
Lab Leak Hypothesis
I hope to never talk about this again, but, well, sigh. At least Covid-19 can now get stuck into monthly updates like everything else.
The behavior of those who did everything in their power to suppress the possibility of a lab leak, to drive everyone to treat one as impossible, have done immense damage to public faith and trust in science, public health, the media and authority in general by pursuing this path. And those institutions all deserve it. This is true whether or not Covid-19 came from a lab, which we likely will never know for certain.
It would be extremely helpful if these agencies disclosed their evidence, and we were able to have an accounting both of what likely happened, and the ways in which the anti-leak narrative was pushed. I do not expect any of that.
Instead, I expect more of the attitude Mehdi Hason takes here, that suppressing a valid hypothesis as ‘misinformation’ is the fault of the bad monkeys who associated that hypothesis with other bad things. How dare they.
Another tactic is to go full clown makeup, and say ‘yes maybe it came from a lab but does that’s not important now.’ Which means that you are the asshole, because you’re harping on this thing that doesn’t matter, long after we suppressed the truth long enough that the point was moot.
We need to find a way to reverse this protectionist trend, or damage will accelerate.
American-only requirements sound too good as political messages for politicians to pass up. If you know anything about economics or business they appear quite bad. They are so much worse than they first appear. It’s so much worse than a gigantic tax. This type of rule doesn’t have a buyout price attached. If there is some input or component that you can’t source here, or can’t get proper approvals to get made here, even some little $10 thing, there is no escape. If you accidentally use something with any component that does not qualify, you are forced to rip the thing out no matter how expensive.
Rick Scott continues to provide a target rich environment by suggesting default sunset of all federal legislation every five years. Then when people accuse him of doing what that would do under current circumstances, he gets mad.
If you are proposing ‘either we pass a bill to do this every 5 years or it stops’ then you are putting that thing in danger. It is hard to pass laws, especially when one branch of government is controlled by people committed to things another branch cannot abide.
That doesn’t mean a universal sunset is a terrible idea. It does mean it has consequences. Also by default there would presumably be a ‘nothing sunsets act of 2028’ and then another one in 2033 and so on, all passed at the last minute, the same way we have a debt ceiling, so yeah it does sound like a terrible idea.
Their fact check was rather weird, starting with it being misleading to look at his jobs record, not because it was staring with the recovery from the pandemic, but because we don’t know what the next two years will bring? The nitpicks are strangely arbitrary all around, in a way I read as ‘politician being politician.’
Old claim from 2012 that calling for a Manhattan Project on some subject, AI or otherwise, may not be like you envision it. If you copied the original, it would mean investing massive resources, sufficiently massive that turning back will be politically impossible, and have it all overseen by a tiny cabal with no oversight that boxes out any hint of dissent, on something unlikely to work, and that it is unclear paid off. Also it would have to be actually secret, so a public call for one doesn’t make a lot of sense, and calling for an Apollo program is seen here as a decidedly mixed bag too, a propaganda exercise that fizzled out and was ‘complicated by a rupturing domestic scene.’ And after all that, he says, the project “worked” only in quotation marks, because as it happens we would have won the war anyway. And we don’t want a War on Cancer either, because sure it did a lot of good things, but it didn’t cure cancer.
My response to this is that it is very very hard to keep a long term secret where one spends this kind of money, especially given how our system works.
We did the Manhattan Project once. We did the Apollo program once. These were unique attempts. There is not a large array of failed such projects. We simply stopped using these kinds of systems and procedures to do things. Yes, there are a bunch of massive boondoggles out there but it seems easy to draw key distinctions in reference class. The record here is kind of spectacular.
So I’m not calling for such a Manhattan Project, because that would be a contradiction – if it exists, it very much wants to be secret. I am only saying that if one were to happen for AI, it would be a highly hopeful development.
Gamers Gonna Game Game Game Game Game
I finally put some time into Monster Train’s DLC, The Last Divinity. I’ve been having a lot of fun, Monster Train is great. The Last Divinity adds some new twists and options. It’s definitely worth picking up if you previously won Covenant 25 and want to keep going, or otherwise want more.
There are some balance issues.
It seems like the change makes Covenant 1-9 much easier than before, since you can take on a ton of threat without paying much of a price for it, then there is a huge difficulty spike when the bosses start scaling at the same time as the emberdrain on the top floor kicks in, and that kicks you in the nuts.
It also seems even more than before like the game is all about finding something super powerful. You are scaling the enemies more and faster and there is a new final boss to prepare for, so either you find something that can keep pace, or you don’t. You have to find a way to roll high.
There are also some very clear gift high rolls out there, even more so than before. In particular, getting the blacksmith event – which doesn’t even eat a Caverns slot – gives you a great giant unit and an insane spell you can then pact up with bonuses, in a way that essentially wins the run on its own with any reasonable support. It’s not a great sign when the game gives you two free cards and you suddenly want to duplicate both of them twice.
Then again, that’s kind of what the game is about, glorious imbalance.
I also don’t love that The Last Divinity has straight up purify, which feels like it invalidates some strategies, in a way that feels less fun and fair than what the Heart does in Slay the Spire. Game is still solidly Tier 2.
Some people have noted the DLC here seems a little pricey for what you get. I don’t care much about the whole DLC value proposition question, because once you know you love a game the marginal cost of new content is so much lower than the payoff. If you know you are going to sink dozens of hours into a game, compared to what other entertainment goes for or the fun you get to have with it, who cares what it costs?
I spent a bit of time with the old classic Mini Metro. It’s a nice, relaxed, fun time, and gives you interesting puzzles to solve. Tier 3. My main criticism is that the game encourages you to be moving trains and tracks around constantly in a super unrealistic and micromanagement-filled way, so you need to decide where to draw the line on that, but for a few bucks you can’t go wrong here.
I spent a few hours on Aces & Adventures. It’s not actually a rogue deckbuilder, you instead grind collection and level abilities as you play adventures and then build your deck out of all the packs you’ve opened. Some very cool things here, also some large design flaws. In particular, you get rewarded for a lot of very grind-focused, battle-extending behavior, there are strong incentives to pick one character and stick to it, and different runs and adventures do not feel different, and some strategies seem way way way better than others the game clearly wants you to try. Still, I did enjoy seeing what they’d done with the concept. Tier 4, missed opportunity.
The main game I’ve been playing is Elden Ring, which everyone knows is great. I agree, although given how frustrating it will be to many people I can only give it Tier 2 – I wouldn’t tell everyone to play. I’m not so great at it, so I’m struggling, and running into a wall where I don’t see where to go other than fighting bosses, and I’ve run out of ones I know how to beat while levels are starting to give diminishing returns. Git gud, I guess.
Before I begin, I have an announcement to make: My game, the Emergents Trading Card Game, has launched!
You can download the game here for free for Windows, Mac and Android. On Android, you’ll have to side load it for now.
While I am obviously biased: I highly encourage you to download the game, play a bit, give it a shot, decide if it is for you.
This is especially helpful to us now, while we’re fighting for critical mass.
I have been working on this game for over four years. From the beginning I knew we had something special. We assembled a true dream team and I think we knocked this out of the park.
Some of the game’s unique features: The modified turn order and letting players pre-play actions and traps ensures games flow smoothly while maintaining interaction. Selling cards on bonding curves via smart contracts on Tezos, and offering to buy them back, lets players buy exactly what they want without locking them in, and letting prices move will encourage card and deck variety. Cards will be able to go on their own quests to get chromatic upgrades. Brian David-Marshall gave the game the wildcard system, where you can attempt to play cards without paying for them, allowing us to have a unique resource system that preserves interesting decisions and keeping games unique without creating non-games. Oh, and he also brought along an entire fully formed comic book universe to build the game around.
When I play games today with decks full of only free cards, I still have a blast.
All right, announcement over. On to the miscellaneous news of the month.
Bad News
ADHD medication situation keeps getting worse.
The equilibrium is that we will auction off ADHD medication this way, see who is willing and able to make the calls. Which is a problem, both because this is a rather destructive auction, and also only the people able to focus enough to make all the phone calls will find their medication, which is supposed to treat ADHD.
Via Jorbs, a warning not to use the truly atrocious Softgiving.
IRS is so slow to make determination of whether it is taxable income when the state gives you money (which, as I read it, includes giving you your own money back) that major tax preparers H&R Block and TurboTax had to made the determination for them?
I increasingly see reports that Amazon listings include more and more fake or terrible products. This is the latest one, which also proposes a solution.
My experience has been like River’s. If you go looking for the cheapest thing listed and don’t otherwise filter, you’re going to select for a lot of shoddy or fake goods. Don’t do that. It’s unfortunate you can no longer play in easy mode and save the maximum. To do that you need to do more work and pay attention. There is however an effective easy mode, which is to use Amazon’s Choice. That’s like buying insurance.
We only pay Fed Chair Jerome Powell $190k a year. He says in a video that that’s fair because that’s what you have to say. The comments know better and point out it’s way way too low.
Generally Twitter is very good about not spoiling sports events with their ‘what’s happening.’ Monday the 13th, they broke that rule:
Very disappointing. I mean, yes, if you are on Twitter hoping to not know who won last night’s Super Bowl, good luck with that anyway. Still feels like the breaking of a sacred trust. I was going to tell Elon to fix this but I need to save my powder for (see AI Post #2) telling him to not doing the worst possible thing. So, never mind.
USA and UK publics mostly do not know about the concept of nuclear winter, those shown infographics on it were much less inclined to favor nuclear retaliation, although I would worry there was partly a priming effect involved here that shouldn’t alter the result.
The question one must then ask is: Would informing the public about such dangers, or otherwise making them predictably less inclined to retaliate against Russian aggression, increase or decrease the long term risks of destruction from nuclear war? My model of Russian behavior, and of Putin’s behavior, and of the decision theory and game theory here, says that our willingness to retaliate is vital to preventing nuclear war. I don’t love the plan, the alternative still seems way worse. Sometimes it is better to lower awareness.
From February 23: 11-year-old girl dies of H5N1. Of 12 contacts of the index case 4 had flu-like symptoms. It is probably nothing to worry about, and we now have a solid week of not seeing scary follow-ups. This does not indicate human-to-human transmission and we don’t have further evidence pointing in that direction. It is still a clear case, to me, of ‘if you did not respond by instituting a very wide quarantine zone while you gathered the necessary information, your civilization is inadequate.’
From March 2: China reports human case of H5N1. Only one case, on its own, still being news is good news. It means this is important enough to be news, and that it got reported rather than suppressed.
PhD applications becoming more like pure signaling credentials earned via servitude, since anyone able to get accepted did not need the training.
Remember you always, always have to pay for marketing, sales and the availability of the service, not only the marginal cost of the service provided. Also this is one of the reasons you probably aren’t charging enough.
We need to continue the drug war, you see, because, well (28 second video)…
Michelin Stars make restaurants more snobby, as basic theory would expect. Menu description length increases, references to cooking techniques go up, prices rise. If anything, given how hard it can be to get a reservation, such restaurants don’t charge enough. If something increases demand without increasing supply, price had better rise. The increased snobbery is unfortunate for those of us who want great food without the snobbery, but the incentives are obvious.
Alas (direct link, paper is gated):
The paper’s claim is that third party evaluators refocus attention in potentially damaging ways. If everyone is obsessed with star status, they are focused on something other than actually creating value, and it will show. This certainly matches what I know about such restaurants. You want to be a place that happens to have a star for doing what you would have done anyway. Beware Goodhart’s Law.
That lines up with this paper’s finding that public scrutiny hurts firm performance.
Although, wait a minute. Inclusion in the S&P 500 also means tons of index funds have to buy your stock. It is somewhat odd to use that as your triggering event and assume that public attention is the primary effect on future performance. One could also say that if a growing company has more attention then it also likely has more hype, so perhaps they are simply overvalued, whereas companies that stay under the radar are better buys because no one else is buying them, not because they are run better.
The effects described still make a lot of sense as a causal mechanism. Once you ‘hit the big time’ and need to satisfy a bunch of prying eyes, you stop doing what makes sense for you to do, and start doing what you want to be seen as doing by others. Which means standard procedures, whether or not they make sense in general or for you.
There is always a balance to be struck here. Too little scrutiny and attention means no discipline or constraints on behavior, and no alert when things are going wrong. Too much scrutiny, and your focus is on all the wrong things. Once again, beware Goodhart’s Law.
This seems related – report that announcements of layoffs do not predict changes in payrolls. So what are they for, exactly? Presumably, in large part, for outside observers.
Why does the South have such low credit scores? The Washington Post attempts to lay the blame on medical debt, and in particular lack of Medicaid expansion and lack of health insurance.
This is a great example of the Bounded Distrust principles, where they are careful not to make false statements while advancing a narrative that does not fit the data and acting as if it was conclusively correct.
Compare these two graphs above versus below, and pay attention to state lines.
If their theories were right, we would see state lines show up on the credit score graph. We don’t. We would see the insurance graph and the credit graph match. They don’t.
In particular, if you look at Wisconsin, Louisiana, Kentucky and West Virginia on their own graphs, you see zero or almost zero impact when crossing over state lines where these policies change dramatically. This hypothesis seems clearly false.
Advice from Arnold Kling not to hire disgruntled employees. You want them to have things they disliked about their previous employer, passive is bad, you want them to want things to be better. Ideal is what Kling calls serene, with strong preferences for better choices and constructive ideas of how to improve. But you don’t want someone who has nothing positive to say, that will constantly find reasons to be unhappy without offering you actionable ways to fix things. This seems eminently reasonable.
Why do East Asian nations tend to be less happy than their income levels would predict? Hypothesis is that, while social connection is good for happiness, comparisons are toxic, especially when they involve money. An experiment is run comparing wheat areas versus rice areas in China, including places where farmers were randomly assigned a crop. Rice areas and farms, which are more like the rest of East Asia, have people making more comparisons as people are forced to work together and monitor each other, whereas wheat areas have less, and the wheat farmers and others in such areas are happier. Interesting. I do know that comparing yourself to others is, in terms of happiness, clearly a fool’s game.
I do want to notice an alternative hypothesis, which is that the other thing that is special about rice areas of East Asia is that they have quite a lot of people packed into remarkably little area, especially once you control for their level of wealth.
It is not a great leap to suggest this could be part of the (relatively modest) happiness shortfall we observe.
Pirate Wires notices ‘a staggering wave of grand theft auto.’ It turns out that it is very very easy to steal a manual transmission Kia made between 2011 and 2021, or a Hyundai manufactured between 2015 and 2021. Post suggests that, contra journalists, the fault lies not in TikTok for spreading the word, and not with the auto manufacturers, but with the people stealing the cars.
I am going to go ahead and primarily blame the car manufacturers. This was spectacularly bad design. If you make it trivial to steal your brand of car without triggering the alarm, once word gets out a lot of people are going to steal your brand of car. I would like to live in a world where that is not true, and I’m not not blaming the people stealing the cars. I’m also on board with cops arresting people who steal cars to take on joy rides, and throwing whoever did that into jail long enough to make everyone think twice about such behavior.
This is still happening for a very obvious reason. If I owned one of these cars, I would either find a way to fix the vulnerability or get a different car. If I was either manufacturer, I would do my best to figure out how to fix the vulnerability and offer this for free or at least at cost.
FTC morale problems continue to get worse. Presumably there is good reason.
Good News, Everyone
LessWrong 2021 Review offers overview of their best posts, awards prizes.
Manifold Markets got a scare that Stripe would do the thing where the card processor kills one’s business model entirely for no reason, but then they decided not to, so M$ will continue to have value. This kind of thing does seem to be increasing in frequency.
Food delivery in China is excellent and growing rapidly. I am also greatly impressed by the American version – I use the Caviar app, and as long as you spend $100 a month you get a 10% off unlimited-use code, plus Chase Sapphire card reduces fees, so basically you tip the dasher instead of tipping the waitress and you’re good.
Super Bowl ads were actually pretty good. My favorite was the Tubi one. Also, to respond to those for Google Pixel, yes you can edit people and things out of photos, but you mostly shouldn’t, that’s missing the point and kind of dystopian and wrong.
At least one in six employees is lying: Bloomberg reports on the world’s largest four-day work week trial, finds few are going back and about one in six employees said no amount of money would convince them to return five days a week. Somehow I do not think that someone still willing to come in four days a week has no price on returning to their old work habits. I mean, I don’t want to go into an office or be fully ‘at work’ for five days each week, and it would likely be unwise to pay my price for either – but I assume you that the price exists and there are people who can afford it.
Bill Watterson, creator of Calvin & Hobbes, has a new book called The Mysteries, out October 10, Amazon pre-order link here. Has anything ever been more self-recommending? This is the way.
TikTok (reminder: Chinese spyware and general wasteland, Do Not Use) limits teens to 60 minutes to prevent binges. Which is a real limitation, since the average user somehow spends 95 minutes a day looking at the short videos. Except you can get around it with a password. And the setting can be modified. I do presume a reminder at the 60 minute mark would have some effect, but even that can be removed.
The pandemic caused a drop in cancer screening. People predicted a resulting increase in cancer. That didn’t happen. They are now falling back on ‘it could take years’ to see the impact. Have we considered the null hypothesis?
Eagles take responsibility for their loss in the Super Bowl. Class act. This is the way.
This is also the way, Biden proposes letting SSI recipients receive meals and groceries from friends without cutting their benefits. Or more accurately, lifting the technical requirement to do this that no sane person would ever comply with in full. I mean, all of this is obviously crazy to think people should live like this:
Earnings in Finland (and America) scale with cognitive ability and don’t plateau the way they were found to do so in Sweden. They keep going. I have seen a lot of smart sources use the Swedish finding of a plateau, which always seemed counter-intuitive and weird, as important load bearing in their models of work and intelligence and the economy exactly because it was so counter-intuitive and weird.
Could it be this simple?
The smartest people, perhaps, know better than to work for a salary.
Devil Parikh shares her method of using her calendar as a to-do list. Of course a Facebook AI researcher would have a method of time management and avoiding stress that is the exact opposite of what I imagine would ever work for a human, certainly for me, but hey. Works for her.
Theory of online dating as too efficient a market. If someone is ‘out of your league’ in real life, it’s a stretch, but you’ve got a shot because there are only so many options, and perhaps there is an especially good match here and fortune will favor the bold. There is a multi-dimensional less-populated space. If someone too good messages you on a dating app, that means there’s a catch, because the space has few dimensions and is well-populated. Maybe not the old OK Cupid, but any modern ‘swipe’ app works like this.
This cuts both ways, though. For everyone that ‘dates up’ someone else ‘dates down.’ Yes, you might be the one who wins the jackpot, but you also might be the jackpot someone wins. Expectation should be the same.
This also suggests the already-wise strategy of maximum differentiation. If everyone else is acting generically, and anyone who is a little above someone’s median stands out a lot, you can essentially offer to ‘date down’ a bit online and use that to let you aggressively filter for other things you want. Trying to get more of what others in the pool want is the wrong strategy here.
New York Post reports that an epilepsy drug appears to be able to ‘switch off’ autism symptoms in mice (study). I don’t have a terribly large amount of hope that this will work out to anything useful, still interesting.
Words to live by.
(Link to a brief post suggesting being sure you are at least getting either good feelings or utility from an interaction.)
This exists and I didn’t know until now so sharing in case you didn’t either.
While I Cannot Condone This
VC explains why he and all other VCs are so bad at getting back to founders.
If one takes this logic seriously, getting a hard no that isn’t purely about some sort of clean mismatch is very bad news. Both for a job applicant and for a founder seeking funding. If you get strung along, you can both interpret that as a no that is likely a close no. It’s a no that could have been a yes, or that is at least a no for reasons they don’t have a good way to explain. The alternative is that they are simply doing this to everyone either way, which seems rather plausible to me, and it means nothing.
The essential reason that VCs and employers do this, stringing the almosts along, is that they have no incentive not to do it. They keep option value, and they don’t lose reputation. If we lived in a world where it got out who did this and who didn’t and it impacted who got the best applicants, hires and deals, they’d straighten up real quick. People simply don’t care enough to compile that information.
This leaves the question of when this strategy is still a mistake.
Suppose you have a clear almost. You can either:
If you can do option #4, even if you have to raise the bar a bit due to having told them what you are looking for, that seems like the best option by far. It isn’t done. In my model this is because VCs are testing for ability to pass future VC tests, which also won’t give out this info, so they can’t either. My hunch is that this too is a mistake, and you can be a huge value add by providing this info.
The risk of telling people their almost status is that this can potentially cause a bunch of time wasting as they try to turn almost into yes, or that this causes adverse selection. It perhaps makes it harder to get into the deals that turn out to be popular and that you want, and makes it more likely you end up stuck with ones you didn’t want. This again is because your whole goal was to test the founders for their ability to pass VC tests – you want to be a bank and only fund them when they don’t need you.
Telling them no outright closes off your option. Which might actually be a win. You shouldn’t be funding the almost cases, you likely don’t want to waste time on their further pitches this round. Instead, you perhaps buy some goodwill for their next company or round and you avoid the distraction.
Another possibility is that people who say ‘I just want the truth’ are wrong. Perhaps they don’t want the no because it is demoralizing. It takes away hope. It restricts the stories they can tell other investors and their employees. It makes them mad at you for saying no and perhaps that is flat out worse than the anger of being strung along. In that case, the whole thing turns out to be efficient and correct.
Why can’t you trust the media? Because the media is and always has been fundamentally untrustworthy, says Matt Yglesias. The difference is that now the media is competitive enough and trustworthy enough, and we have good enough other means of sharing information and checking against reality, to notice it is unworthy of our trust. So essentially The Media has never been more trustworthy, yet we trust it less. This is part of a pattern of a bunch of things where:
Another clean example is global poverty and global inequality. There are a lot of other such problems.
There are also problems that are in fact getting worse, often with power denying that there is a problem at all or that it is getting worse. It is important to notice the difference between these two groups of problems.
NYC’s three legal weed shops somehow all very close to the NYU campus – 750 Broadway, 144 Bleecker St and 62 East 13th. For everyone else there are still infinite new (technically) illegal weed shops without licenses. If you don’t enforce the law, there is no law.
Forbes list of the largest 100 charities. Lots of food aid and disaster relief, lots of medicine. A bunch of religion. Huge emphasis on things one can feel good about supporting. This is not the most efficient list of places to give, yet mostly I am happy about the destination of the average tzedakah box. Some highlights of a sort: First cultural institution that isn’t religious is Smithsonian at #49. Largest orgs that represent a political perspective are Planned Parenthood at #30 and what remains of the ACLU at #45. Unless you count St. Jude’s Research Hospital (which is #3), the first research or scientific institution is the Michael J. Fox Foundation for Parkinson’s Research at #55, the closest thing here to an Effective Altruist cause area. Largest Environment/Animal charity by far is Nature Conservatory at #15, followed by World Wildlife Fund at #46, nothing I could see related to climate. PBS is #92, and if I had money to donate only to this list I’d likely put it there.
David Friedman provides suggestions for what we might do to improve law schools if the purpose of them was to teach students how best to practice law.
Biden Administration stays true to form, moves to require seeing a doctor in person before getting ADD medication or addictive painkillers. For addictive painkillers this seems appropriate, given how people interact with them in practice. For ADD medication, it depends on whether they are also rationing supply well below demand. If it was possible to go to a pharmacy with a prescription and reliably walk out with your medication, then requiring these in-person visits would effectively be rent extraction from the medical cartel. If instead you already need to call something like thirty pharmacies to find your medication, then we are currently rationing via time spent. It is a strict improvement to go to a hybrid of time spent and money paid over time spent.
Reminding people to get flu shots causes more flu shots.
It is highly plausible that these letters are great. This still seems like the wrong calculation. If you want to measure the costs and benefits, you need to include costs other than mailing the letter. Who do we think pays for the flu shot? Presumably the state of Denmark. The patient not paying does not mean that it is free. The time spent getting the shot, and the expenses of creating, distributing and administering the shot, are very real. Similarly, the social cost of mailing someone a letter is primarily that they have to spend time to process the letter, rather than the cost of mail.
My guess is this is still a great deal. Important to get the math right anyway.
Without comment:
Presidential Candidate Has Bold New Ideas
Only a few days ago I said I expected Trump’s campaign, to the extent he had one, to be focused on the past. Instead, hello bold future? Including building new cities.
This is actually great. We should absolutely be building entirely new cities. It would be even better to build more housing in places people want to live. Thing is, even if we did that, we’d still want to be also building entirely new cities. It’s pretty great.
Has Wokeness Peaked?
Cancellation attempts at universities seem to have peaked.
Some interesting details here. Petitions and targeting incidents both roughly doubled from 2019 to their peaks, but petitions peaked in 2020 while targeting, which is more individual, peaked in 2021. Sanctions peaked in 2021. The ‘success rate’ of targets and petitions seems to have been declining before the peak hit. Could be that some of this is a reporting issue – things that would have been ignored are now reported and recorded. Could also be that many more flimsy attempts got launched.
The decline could be that the fever is passing, or it could be that the low-hanging fruit of terrible people got picked or that people ‘cleaned up their act’ either by no longer being terrible or learning to behave and not violate the new rules, or that institutions found ways to disguise their actions.
Terminations in 2021 were already back near their 2019 levels and lower than 2017, although that could be noise.
Lots of room for this to mean different things and to be any combination of good and bad news.
Summary thread discussing various data on whether wokeness has peaked. I’m skipping the decline in attempted cancellations of professors at universities since I covered that previously.
There was a decline in TV programming discussing “diversity,” “equality” and “racism.”
There are less incidents on campus.
The classic ‘overcome prejudice’ question answer has crested in all groups.
Academic research on bias declined, although modestly. My guess is lead times are longer here, it takes forever to actually get things published.
Support for reparations declined among Democrats, although it increased (from very low levels) among Republicans. Non-empty-talk levels of support, I am guessing, are far lower.
We can also add the word frequency charts for The New York Times.
I am willing to conclude that there was at least a local peak in lived experience of wokeness. That does not automatically mean a long term decline is on its way, or that the underlying dynamics have reversed course rather than the fading away of temporary effects from Trump and 2020. That could still go either way. I do think the number of identical-looking graphs do conclusively show that the local peak was real.
Lab Leak Hypothesis
I hope to never talk about this again, but, well, sigh. At least Covid-19 can now get stuck into monthly updates like everything else.
The Energy Department of the United States now believes that the most likely origin of Covid-19 is from a lab leak, joining the FBI.
This is not obviously that big an inherent update, it is 1 of 8 agency assesments has been updated to ‘weakly confident’ from unknown. It claims to be based on new information, which increases its significance somewhat. It also updates the Overton Window on the question, and reflects how far we have come.
The behavior of those who did everything in their power to suppress the possibility of a lab leak, to drive everyone to treat one as impossible, have done immense damage to public faith and trust in science, public health, the media and authority in general by pursuing this path. And those institutions all deserve it. This is true whether or not Covid-19 came from a lab, which we likely will never know for certain.
It would be extremely helpful if these agencies disclosed their evidence, and we were able to have an accounting both of what likely happened, and the ways in which the anti-leak narrative was pushed. I do not expect any of that.
Instead, I expect more of the attitude Mehdi Hason takes here, that suppressing a valid hypothesis as ‘misinformation’ is the fault of the bad monkeys who associated that hypothesis with other bad things. How dare they.
Another tactic is to go full clown makeup, and say ‘yes maybe it came from a lab but does that’s not important now.’ Which means that you are the asshole, because you’re harping on this thing that doesn’t matter, long after we suppressed the truth long enough that the point was moot.
Zeynep Tufecki has a thread with her thoughts, essentially that we do not have enough information to know what happened and we should have said so from the start.
Various people are taking victory laps, or expressing anger at not being able to take proper victory laps. I do not begrudge.
The State of Our Infrastructure
Our President reports our union is strong, if you were wondering.
In case you were instead wondering if we would get much of anything out of our infrastructure spending or were capable of building things, the answer is essentially no. We have chosen a different path instead, as Joe Biden has done consistently.
Yes. Price controls for Insulin. Also import controls for Insulin.
Even he previously gave in at the last minute and granted waivers for these buy American provisions that he touted. Presumably that is done now.
In general the theme has been economic nationalism, a negative sum game that by default and with notably rare exceptions hurts everyone, including America. It reliably fails to differentiate between allies and enemies, and this is unlikely to be fixed.
It seems that Biden is going to make ‘prevent people from doing things, piss off allies and raise prices via American-only restrictions’ the cornerstone of his 2024 campaign. The central policy of the incumbent President is going to be active sabotage of the American economy.
No, it is not about ‘American jobs,’ except that it is because people think it is.
Nor is this about real security of our supply chains, since it excludes our reliable allies like Japan and Germany. Any scenario where we can’t trade with Europe is one in which we have much bigger problems that none of these policies would solve.
Here is a mostly gated full primer from Scott Lincicome at The Dispatch.
We need to find a way to reverse this protectionist trend, or damage will accelerate.
American-only requirements sound too good as political messages for politicians to pass up. If you know anything about economics or business they appear quite bad. They are so much worse than they first appear. It’s so much worse than a gigantic tax. This type of rule doesn’t have a buyout price attached. If there is some input or component that you can’t source here, or can’t get proper approvals to get made here, even some little $10 thing, there is no escape. If you accidentally use something with any component that does not qualify, you are forced to rip the thing out no matter how expensive.
Rick Scott continues to provide a target rich environment by suggesting default sunset of all federal legislation every five years. Then when people accuse him of doing what that would do under current circumstances, he gets mad.
If you are proposing ‘either we pass a bill to do this every 5 years or it stops’ then you are putting that thing in danger. It is hard to pass laws, especially when one branch of government is controlled by people committed to things another branch cannot abide.
That doesn’t mean a universal sunset is a terrible idea. It does mean it has consequences. Also by default there would presumably be a ‘nothing sunsets act of 2028’ and then another one in 2033 and so on, all passed at the last minute, the same way we have a debt ceiling, so yeah it does sound like a terrible idea.
Washington Post highlights of the speech here were entirely content free, nothing but petty political optics and who jeered when, so overall you likely didn’t miss much.
Their fact check was rather weird, starting with it being misleading to look at his jobs record, not because it was staring with the recovery from the pandemic, but because we don’t know what the next two years will bring? The nitpicks are strangely arbitrary all around, in a way I read as ‘politician being politician.’
Old claim from 2012 that calling for a Manhattan Project on some subject, AI or otherwise, may not be like you envision it. If you copied the original, it would mean investing massive resources, sufficiently massive that turning back will be politically impossible, and have it all overseen by a tiny cabal with no oversight that boxes out any hint of dissent, on something unlikely to work, and that it is unclear paid off. Also it would have to be actually secret, so a public call for one doesn’t make a lot of sense, and calling for an Apollo program is seen here as a decidedly mixed bag too, a propaganda exercise that fizzled out and was ‘complicated by a rupturing domestic scene.’ And after all that, he says, the project “worked” only in quotation marks, because as it happens we would have won the war anyway. And we don’t want a War on Cancer either, because sure it did a lot of good things, but it didn’t cure cancer.
My response to this is that it is very very hard to keep a long term secret where one spends this kind of money, especially given how our system works.
We did the Manhattan Project once. We did the Apollo program once. These were unique attempts. There is not a large array of failed such projects. We simply stopped using these kinds of systems and procedures to do things. Yes, there are a bunch of massive boondoggles out there but it seems easy to draw key distinctions in reference class. The record here is kind of spectacular.
So I’m not calling for such a Manhattan Project, because that would be a contradiction – if it exists, it very much wants to be secret. I am only saying that if one were to happen for AI, it would be a highly hopeful development.
Gamers Gonna Game Game Game Game Game
I finally put some time into Monster Train’s DLC, The Last Divinity. I’ve been having a lot of fun, Monster Train is great. The Last Divinity adds some new twists and options. It’s definitely worth picking up if you previously won Covenant 25 and want to keep going, or otherwise want more.
There are some balance issues.
It seems like the change makes Covenant 1-9 much easier than before, since you can take on a ton of threat without paying much of a price for it, then there is a huge difficulty spike when the bosses start scaling at the same time as the emberdrain on the top floor kicks in, and that kicks you in the nuts.
It also seems even more than before like the game is all about finding something super powerful. You are scaling the enemies more and faster and there is a new final boss to prepare for, so either you find something that can keep pace, or you don’t. You have to find a way to roll high.
There are also some very clear gift high rolls out there, even more so than before. In particular, getting the blacksmith event – which doesn’t even eat a Caverns slot – gives you a great giant unit and an insane spell you can then pact up with bonuses, in a way that essentially wins the run on its own with any reasonable support. It’s not a great sign when the game gives you two free cards and you suddenly want to duplicate both of them twice.
Then again, that’s kind of what the game is about, glorious imbalance.
I also don’t love that The Last Divinity has straight up purify, which feels like it invalidates some strategies, in a way that feels less fun and fair than what the Heart does in Slay the Spire. Game is still solidly Tier 2.
Some people have noted the DLC here seems a little pricey for what you get. I don’t care much about the whole DLC value proposition question, because once you know you love a game the marginal cost of new content is so much lower than the payoff. If you know you are going to sink dozens of hours into a game, compared to what other entertainment goes for or the fun you get to have with it, who cares what it costs?
I spent a bit of time with the old classic Mini Metro. It’s a nice, relaxed, fun time, and gives you interesting puzzles to solve. Tier 3. My main criticism is that the game encourages you to be moving trains and tracks around constantly in a super unrealistic and micromanagement-filled way, so you need to decide where to draw the line on that, but for a few bucks you can’t go wrong here.
I spent a few hours on Aces & Adventures. It’s not actually a rogue deckbuilder, you instead grind collection and level abilities as you play adventures and then build your deck out of all the packs you’ve opened. Some very cool things here, also some large design flaws. In particular, you get rewarded for a lot of very grind-focused, battle-extending behavior, there are strong incentives to pick one character and stick to it, and different runs and adventures do not feel different, and some strategies seem way way way better than others the game clearly wants you to try. Still, I did enjoy seeing what they’d done with the concept. Tier 4, missed opportunity.
The main game I’ve been playing is Elden Ring, which everyone knows is great. I agree, although given how frustrating it will be to many people I can only give it Tier 2 – I wouldn’t tell everyone to play. I’m not so great at it, so I’m struggling, and running into a wall where I don’t see where to go other than fighting bosses, and I’ve run out of ones I know how to beat while levels are starting to give diminishing returns. Git gud, I guess.