PillPack founder TJ Parker: drug pricing debate is distorted by list vs. net price confusion, EO's most promising element is direct consumer pricing reform

on are you getting some of that context directly on true social or do you wait till it hits uh other parts of the internet? Wait for the delay for sure. I mean I think maybe helpful to like level set like lay of the land for folks that are kind of less please familiar with the category.

Um, so, you know, I think probably the problem I've been most focused on and I've beat a, you know, beat a dead horse on on Twitter is this rebate issue. Mhm. Um, and I think in many ways that's the most interesting part of the executive order. It's the thing I'm most excited about.

Um, so today, if you're a normal consumer, you go to the pharmacy, you need to fill a branded prescription, like a higher cost prescription. Maybe you need an EpiPen and it's January and your deductible will just reset.

or you want to fill a GLP1 um at the counter and it's not covered by your insurance, uh you end up paying something like three to four times order of magnitude what that drug actually costs in the US. No, so not even talking about like what it costs in other countries, but what it costs net in the US.

So for like an Ozepic as an example, you're paying $1,000 or so, which is the list price, the gross price in the US. Um but the net price is actually $250.

just the price the PBM the pharmacy benefit managers negotiated and you've effectively paid $750 more than that drug costs and that money is getting captured by either the PBM the insurance company or your employer usually some combination of all of those things and everyone's always bemoaned pharmacy benefit managers obviously they're kind of everyone's favorite punching bag the middleman in healthcare and they're capturing all this margin and to some degree that's that's true um but no one has come up with a solution that could be an alternative alternative path that could work better, right?

That's the sort of fundamental issue is like you can hate how the mechanics work today, but without an alter like without some alternative, it doesn't matter.

Um, and so I think what's most interesting about the EO is that it there's a specific call out around requiring that pharma companies have comparable direct to consumer prices to list prices in other countries. Right? Right. So, you take that as a Zepic example. Mhm. In the UK, that's a $150 to $250 drug.

Actually, not that different from our net price, right? It's sort of similar. It's a little bit less, but it's like not that different. Um, they're basically requiring pharma to have that sort of pricing available if the consumer is just paying cash, kind of paying out of pocket.

And why I think this is so interesting is if you start having these things get priced in somewhat of a rational way for consumers, all of a sudden the need to negotiate these rebates and have all these kind of backdoor negotiations and these fees and all this stuff that creates the complexity effectively goes away because you have some version of rational pricing for consumers.

Um, so there's a bunch of other stuff in EO and I'm happy to get into some of the more nuanced complexities, but to me that's pretty exciting if you actually had like normal reasonable cash prices for the end customer.

So just staying on prices, um is is this conf is this confounded by the fact that in other countries they have uh larger government sponsored health care plans or so when we talk about Americans paying higher drug prices, are we saying American consumers versus European consumers or are we saying that the same pharmaceutical company actually makes more money from a single dose in America versus uh Europe?

Yeah. So both is the short answer. Um but I think what's happened in the press that has really been challenging is that everyone always talks about the list price as the price in the US, right?

And so the narrative if you listen to like the rhetoric over the last handful of years is like we're paying 10 times as much for a Zmpic or we're paying 10 times as much for insulin. We're not. The list price is 10 times as much as it as a list price in the UK. We're paying and it says this in the EO.

This is definitely the truth. We're paying two to three times the net price. Yeah. The Wall Street Journal here said that uh the list price for diabetes medication uh Jardians was $611 for a 30-day supply in the US versus $70 in Switzerland and $35 in in Japan.

That's what you're talking about being not entirely accurate. That drug is probably actually like 100 bucks net in the US. 150 bucks. Okay, got it. Yes. More expensive. Y but not 20x. That's interesting.

And so I think you're hearing and like you know I think you're hearing a lot of the dialogue be about like all the downstream implications of R&D and not going to be any investment if these prices come down in the US because they're the only ones will wanting to pay.

That's an interesting conversation people should have that conversation. It's it's pretty complicated and nuanced. But I think everyone can agree consumers shouldn't be paying 10 times as much when they get stuck with the full bill. Yeah.

And certainly shouldn't be paying three times as much as the PBM is paying or their employer is paying. Um, that to me is what's most interesting is you could come up with a there's a there's a real credible solve here potentially for that problem which I think is much more uh yes compelling and and interesting. Cool.

How how have you uh what what are your thoughts on the market's reaction? I think everybody expected this massive selloff in pharma and biotech and maybe it's because there were there was other news announced this morning. We haven't seen that but but biotech markets went up by like 3%. Yeah.

Do you think people are processing this? We had what we heard earlier was that it was kind of like like the like a really really bad scenario was priced in and this was less bad so it popped. Do you buy that narrative or what what are you taking away from the kind of the market reaction? Yeah.

So the PBMs and pairs have gotten whacked which makes sense if you think about the the dynamic here. Um I think there's a lot of skepticism about how how much this can actually be implemented and the success that they'll have.

Um if you jump back to The last administration, they tried to repeal rebates, which is solving the problem I just described. It's it's sort of getting rid of this excess margin that's invisible to the consumer and they failed. They got it pretty far along.

It was like an admirable effort, but ultimately CBO shut it down because rebates do contribute to lowering premiums and there's things that made it difficult to get rid of rebates. They also had an MFN approach on Medicare Part B drugs, so like drugs that are dispensed if you're in office.

um uh things like that and that failed in the last administration.

So I think my sense is it's some combination of if the real punchline here is that effective this is a back door way to get rid of rebates it doesn't it's actually not that big of a deal for pharma like they're still getting their net economics it's just that for the middleman right so that's like oneation from the market and then there's probably also just some skepticism about whether this will actually get implemented and I think that's fair skepticism given the the historical track record here what does this mean for pharmaceutical R&D the way it's done we talked Zack Weinberg a little bit about NIH funding getting cut.

There's debates about how research gets done in, you know, the Ivy League universities that spin out a lot of drugs. Um, there's so many moving parts to actually get us what we want, which is probably just like a cure for cancer or just great new drugs, right?

Um, h how is the actual pipeline evolving through all these different changes when you stack them all up? Yeah. So if you jump to the other other components of this beyond the consumer implications, yeah, that's where I think these questions become more relevant, right?

So if there's really, you know, an an MFN for the net price that we pay compared to other similar situated companies or countries that that would have potentially some of the impacts folks are concerned about on R&D.

I think we're so far away from that and that's so difficult to imagine truly getting implemented on face value that I'm skeptical that that's really the the thrust of this EO. Um I really think it's a potentially a better way to to eliminate rebates than historically.

I do think if you kind of play out let's let's assume that all of a sudden OMIC does cost 200 bucks at the pharmacy counter which I do think is a credible path here and it would be reflective of what the true net cost.

Uh and all of a sudden consumers are sort of just paying for most of these products out of pocket versus like bothering to use their insurance and then you have pharma companies competing on price to win the to win the transaction.

That's actually I think how net prices come down ultimately not because there is going to end up being some true um kind of deadnet MFN that that they're able to execute against.

So, I think the the debate about R&D is sort of, you know, to me just from a practicality standpoint, um, kind of two clicks away from where we are today.

And I'm pretty focused on can we just get to a a place where consumers are choosing these products based on price instead of PBMs and other middlemen trying to determine price. Why why is the market for OEMIC so competitive on day one?

you hear this whole narrative around patents and um you know the whole idea is like yeah you you should be able to incentivize this like this heroic uh herculan research effort because you'll generate all this value over the long term and that certainly happened with Novo and their share price but at the same time it feels like the GLP ones came out there were like every major pharma company had one around the same time then there was compounding going on there was so it felt like they were already generic almost, but I know that they're not.

But, uh, why why did the market play out in this way? Is it just because it's such a dominant technology, everyone was thinking the same idea or something? What happened there?

Because it was originally approved for diabetes, not weight loss, and that weight loss was a byproduct that they figured out was a was a not a side effect, but like also an effect of the of the products.

And so, when it caught, you know, the public's eye, it was already much farther along in its life cycle than a traditional kind of new indication, new drug. Got it. That being said, like the the price dynamics with GOP ones are not that anomalous.

Like if there's a successful product and there's a known pathway, there's going to be other similar products that get developed. Mh. And the current ecosystem supply chain, current structure that does drive down net price. That's why these things don't cost $1,000 anymore.

That's why their net price is 200, 3000, $500 depending on the how new the GLP1 is. So like the the system actually kind of works. Okay. right now like the price does come down based on competition relatively quickly.

It just doesn't work particularly well for consumers in these instances where they're paying out of pocket. So everyone else is wearing a hat. So I wanted to put on uh how how uh can you comment at all on Europe's existing solution?

They have some complex system of price controls where they tell US drug makers what they can charge is what we heard from an earlier guest. But yeah, somebody else was saying that this EO is like Bernie Sanders dream. Is that true at all? Are these like are we in like leftwing or right-wing territory?

I can't even tell anymore. Yeah, I mean that's going to depend a lot on exactly what's being contemplated here, right?

I think if the extreme interpretation is these are literally price controls that the government setting like yes, I think you're kind of in the Bernie Sanders extreme version of uh managing these prices.

I think if you take the other extreme and it's really focused on consumers and they say well look either you can set reasonable prices for consumers that are somewhat comparable to list prices in other countries or consumers can just buy these drugs from other countries and import them here.

Yeah, that's not particularly that's not a price control. That's just actually opening up the market to more competition. So really I think it depends on the exact uh the exact implementation and what's being contemplated. Yeah.

uh where are you seeing opportunity in startups in healthcare to kind of carve out value or deliver value in healthcare broadly right now? I mean I'm as you can probably tell from my my uh tweeting I'm I'm quite interested in how you pull forward shopping and pricing into healthcare.

Like to me that is the lynch pin to make the category work better. That's a lot of the work we did at Amazon, as far as I know, was the first time you could actually see insurance price for something before you bought it. Healthcare, which is um pretty insane.

Um and so, you know, I think being able as a consumer to understand what your options are, how much they cost, what's available, and ultimately like transact in a very seamless way, I think is the most interesting thing. Now, that could be like literally what feels like a marketplace, right?

I think that's a very interesting opportunity in different categories. It could be utilities that make that possible. So like super clear like APIs for pricing. There's a bunch of permutations of things that enable that, but that's what I'm still How are we talking about APIs in 2025?

Like it's the age of AGI, ASI, like h how do the insurance companies not have APIs? This is insane. Yeah. I mean, what's going to end up happening because you know, similar to faxes, we still send faxes, but really it's just two digital interfaces on both ends.

you're with all these like phone agents that are talking to each other, but it's basically I swear to God, I don't this isn't hyperbole. Like this is crazy. Think like the insurance companies be like, "Okay, cool. Like you want clear pricing and you're bombarding us with phone calls with your AI bots.

We'll just give you like a dedicated AI bot and you guys can prosecute this and figure it out. " Like bullish for 11 labs. Yeah, this is going to be like voice agents win everything instead of just building an API. Are there any places where AI in healthcare makes sense or is exciting for you?

Uh there's like the super frontier research going on at Deep Mind. Uh there's protein folding is solved now. There's stuff like that. But then at the same time, I can imagine like the the use cases you described being very real business cases.

Yeah, the voice agent specifically feels like an area that's experiencing somewhat rapid growth, but the nature of that also means that it's probably, you know, it is hyper competitive. So, I'm curious if you think categories like that are are even investable. It's tricky, right?

I think I so I buy that there's going to be a bunch of administrative efficiency in healthcare, right? things like phone agents and prior authorizations and other kind of very laborious work stream like workflows. It's much less clear to me where that value occurs. I think that's the the tricky bit.

Like there's an argument that it actually occurs to the service businesses. Like that's historically where the value occurs in healthcare. And so maybe the most um effective service companies can acrew that value, which is being good at implementing AI.

The place where I'm pretty convicted it will be valuable and will drive real venture returns is companies that are are good at compounding those as differentiation for consumers.

So as an example of before it would have taken a human $40 to book you an appointment to see a specialist if you can use an AI bot to do that for two bucks that unlocks a new consumer experience too right it's not just like an administrative efficiency and there's lots of these examples that's probably what I'm the the most excited You mentioned compounding.

What's your take on drug compounding? Oh god. Uh I'm I'm quite vocally anti-drug compounding. Uh why is it a quality issue or is it just a market dynamic issue? It feels like he's an IP respector. I'm an respctor above the law and below whatever the right nomenclature is. Uh it's twofold.

I mean I do think there are there are real safety concerns. Mhm. Last time there was a prolific compounder compounding a commercially available product at scale was doing a compounding center and they killed 100 people with menitis like happened. Yeah.

Real like not not claiming that's going to happen here but that is like the overhang from compounding. It also just it it sort of outside of these temporary shortages very much violates IP law. And to the conversations before about whether there's any incentive to do R&D.

Well, if someone can immediately just copy product, there's certainly no incentive to do R&D. It's much worse than some kind of price.

Is it kind of is it kind of a beneficiary like is is there a boom in compounding because of the internet or like is there a is there a market force that's driving the increase in compounding or is it just like no one ever thought to do the loophole before? Well, the funny the funny dynamic. What was that, Jordy?

I was just going to say it's not like people are compounding and then you know, hey, we didn't have to do R&D. We're just sort of like going to compound in this and then we're going to pass the savings on to you.

That's been like a big part of like the messaging online and it's like no, like you're going to pay the same and like you're going to like it and just like be happy that you're getting the drug has been seemingly I think the market dynamic is rebates, right?

If WGOI or Empic was the net price for consumers because these are now out of pocket drugs and they're 200 bucks, 300 bucks. Is there like much of a market for compounding which are also 200 bucks and 300 bucks?

Like the reason that there was so much demand for it is because of this disconnect between list prices and and net price, right? And so if that dynamic goes away, all of a sudden the opportunity for compounders to really gain share is is is minimized, right?

Because like I can understand why someone would really not be willing to or could afford paying $1,000 a month and they can get the compounded product for 200, but if we're talking 200 versus 100 or 150 for a compounded product versus the real product is a very different calculus for the consumer. Yeah.

What what percentage of doctors do you think are using chat GPT on a weekly basis? And should there is there a big rapper to be built there just giving them a more HIPPA compliant version of of chat GPT?

I can't imagine shouldn't be I mean I bet it's like on a like using it at all on a weekly basis like most doctors I would imagine at this point uh using it like as part of their workflow all day. Probably still a smaller subset but definitely in the double digits. Is that good? Is is chat GBT HIPPA compliant?

He's just asking the most obvious. Well, you don't want to be dropping like specific patient information into chat GPT, but like you can certainly do generalized. You can do generalized.

There's other ways to do that, but certainly generalized like workups and notes and like we're using it a bunch and one of the companies I'm involved with to help, you know, automate the questionnaires or the the notes or things that happen downstream for efficiency.

Like makes tons of sense should it absolutely happens and will continue to happen. Yeah. some of the biggest, you know, I think everybody has this idea in their head that you're going to be able to go to like AI doctor. I don't know what that looks like, right? I don't know if it's embodied.

I don't know if it's a screen. I don't know if it's a doctor is chat GPT roughly. Yeah. Yeah. But I don't know. I don't necessarily know what the form factor is. But as an investor, are you is that a category? How do you think about that opportunity, right?

you know, at a high level, is that one company that's just gonna Yeah, we've seen for lawyers. Is there going to be Yeah. Or is it more of like a co-pilot approach and it's just less exciting than it sounds? It's just more doctors can see more patients and deliver better service. But how how do you see that category?

Because I I it's just one of those areas that people always bring it up in the context of, oh yeah, everybody's going to have like an AI doctor. Yeah. And it's like, okay, well, what does that actually look like? Yeah.

I mean, I, you know, I think for sure the first step when you're trying to deal with symptoms and figure out what's going on is going to be going to chat GBT or another similar CX, like pretty confident that's going to be the first step.

It's not that dissimilar from the first step today, which is people were googling symptoms and trying to figure out like what's going on, and it was just much less efficient, but ultimately like a very similar kind of flow. So I think that's for sure going to happen.

I think the interesting question is are the vertical players actually going to have a better differentiated product than the foundation models? That's a lot less clear to me.

Like I still think the found like the core foundation models in my personal experience are better at diagnosis and better at figuring these things out still even though there are dedicated other models. Well, and that's the that's a big issue, right? You could have a rapper that's specific for diagnostics, right?

That you can message and say, "Hey, I have this like, you know, skin thing. Like, what's going on? " But that's like so infrequent. You're not paying $20 a month for something like that. If Chat GBT can do it 80% as well and, you know, it can kind of get you on the right path. And so I don't think CHBT eats WebMD.

Yeah, for sure. I think the the the more interesting problem to solve, right? is like at some point you get through most of that conversation, you're pretty sure you figured out what's going on, you still need an intervention of some sort if there's anything moderately serious going on.

Like you either need a prescription, you need labs, you might need some physical intervention, like you need something. And that's where nothing in healthcare has actually been digitized, right? Mhm.

Like if you think about this compared to other retail categories which I do think is the right mental model like chat GBT is now implementing you know Shopify checkout right in chat GPT right so you can go from that what should I buy straight through to a checkout flow no one's built that checkout flow in healthcare like you don't have a product catalog you don't know how much stuff costs you don't know like what's available to you based on your insurance like I think that's actually a more interesting problem for startups to go wrestle through because chat TV is going to be pretty damn good at the things you used Google for before and it's not going to deal with any of this stuff.

And so the stuff I'm working on and spending a lot of time on is much more in like how do you build the equivalent of product catalog and healthcare? How do you ultimately make it easy to transact?

How do you make it easy to shop because that infrastructure still has to get built regardless of where the kind of front door is into that experience. Shopify for health something like that. Uh can we do overrated underrated uh tellaalth broadly? How are you feeling about tellahalth?

Uh I am bullish on the sort of usage of tellahalth still. I'm probably bearish on most of the existing business models and monetization. It's probably the short bit. What about overrated underrated existing business models? Ha have you been uh surprised to see like him's almost second act in the public markets?

Do you think they would even have had a second act if it wasn't for GLP1s or is there some underlying unlock that they have? Uh I mean it's sort of like commenting on any meme stock, right? It's a tricky thing to to have a strong opinion on. Well, I think that's enough commentary. That's enough commentary.

Uh overrated, underrated GLP1s. Uh underrated underrated. What about uh some of the foundational AI applications coming out of Deep Mind protein folding? Any of that stuff moving the healthcare markets or uh how you think about uh healthcare? Not intelligent enough to have an opinion on that. Yeah. Too complicated.

Too complicated. get out of it. Anything else?

Uh, did you when you were a pharmacist, did you was there a world did you imagine a a world in which you retired uh as a pharmacist uh you know at 65 or did you like cars too much from a young age that you knew that you'd need to figure out another way to build a collection?

I think it was pretty clear that the working behind the counter was going to be a short-lived endeavor for me if that's the question. Are you asking is that what you're asking? Yeah, basically. Yeah, I think that was pretty clear from the jump. How much do you love being a venture capitalist now? Is it just the best?

I love it. I sort of like the second I took the job, I became an expert at everything. It was so convenient. Yeah. Yeah. Maybe it's almost instant. It really is almost take us on a little world tour of uh all the geopolitical conflicts, how we would resolve them really quickly.

I'm just really happy that my first conversation on here was political in nature. That was what I was looking for. Yeah, it's fantastic. Yeah.

Um the last question I have uh just generally curious given um uh you guys are are I'm sure do multi-stage but I imagine given like are are you like trying to pick like are are you investing in a couple companies a year? What what's the goal for you?

And and I'm assuming you meet a lot of companies all the time that you like but don't end up investing because you're not like you just more of from a conflict standpoint. If I do this deal and you know somebody a better company comes around and in six months or a year I'll be conflicted out. Is that accurate? Yeah.

I mean certainly the first bit's accurate. We're we're you know doing concentrated early stage investing kind of classic venture. So, I'll do two to three deals a year in any given year is probably a pretty standard uh standard year for for me and for the other partners of Matrix.

Um, I mean, I certainly think about the conflict thing. I think that reassumes that I'm constantly liking things and wanting to do all of them. That's probably not the the dynamic.

It's really like finding like really interesting things that I want to spend a bunch of energy and a bunch of time on versus seeing like a lot of different things that I'd love to kind of be less involved in. So, it's certainly a dynamic, but we're, you know, we're doing a couple deals a year.

As you do deals, when do you start using the term we to describe how the company works? It happened pretty quick, too. It was like the same as my expertise. Just immediately write some small check and it's we we are building the future. Yep. Dropping a Wii as a venture capitalist strong.

That's you know, hey, I'm on the board. Yeah, I'm part of the team. Former operators otherwise you're not allowed. What uh what the last thing uh before uh you take off, what should people be paying attention to in the days and weeks from now around the actual implementation of the of the EO?

What are you looking out for? I I think it's obvious that, you know, you're not going to walk up to a pharmacy after work today and and see lower prices. Uh but but what are you looking for?

I personally would be focused most on how much this is really about the difference between gross and net prices, right, which we talked a lot about, or whether it's really about true competitive net prices with other countries because those are pretty different paths, right?

um and has all sorts of different implications. So I think the more this feels like it's about solving the consumer problem and less about solving the the kind of payer problem or the fairness problem, the better for me personally.

Um and the more it feels like we're trying to do some true price setting kind of net MFN against other countries, the more I'd be a little bit nervous. So that's where it pivots for me is is this about consumer? Is this about net MFN? Great. Do you think the Mission E makes it to production?

Do you think Porsche goes a different direction with this next supercar? You know, I I know so little about like new modern Porsches. I'm like the the furthest from an expert, but good luck. No idea. It's the short answer. Ask me about something old and I I'll talk about it for a long but yeah.

Well, we'll uh we'll be coming out your way. We're working on putting together the Aman rally. So, uh more news to come there. you'll be one of the first to know. This is fantastic. Awesome. Thank you. Thank you for coming on, TJ. Yeah. Thanks so much. This was great. We'll talk to you later. Appreciate the insight.

Cheers. Uh did were you able to clock his watch? Uh something on Aquinaut. Well, if you're looking for an Aquinaut like TJ, head over to bezel. Your bezel concier is now available to source any watch on the planet. Seriously, any watch. I think I already did that actually.

And then of course, we got to tell you about Wander. Find your happy place. Book a wanderer with inspiring views, hotel grade amenities, dreamy beds, top tier cleaning, 247 concier service. It's a vacation home but better. Jordan, hit me with some soundboard. What do you want, John? I got options.

Uh Ashton Hall, you know, it's my favorite. Oh, the the movie trailer. We also have some poly markets up that are tearing up the news. Uh the odds of Donald Trump eliminating capital capital gains tax this year are surging. It's now looking more likely than not that they'll be eliminated. 57% chance.

Unclear if that means they'll be taxed as income or not taxed at all. Uh but they will be eliminated. We eliminated capital gains tax. It's now 50% 50% or 0%. Who knows? Uh also the odds of Larry Ellison buying Tik Tok have doubled this morning. One in five chance now that he buys it before July. He's up at 18%.

Someone knows something. Someone's going in with size. Um, but of course this it's phrased as who will acquire Tik Tok. Big question about whether that's a whole co-acquisition or a minority investment from an American company. Uh, could also be the data is hosted on Oracle's uh, you know, cloud their servers.

Could be that the algorithm is run on Oracle servers, but that doesn't necessarily mean that Oracle knows what all the weights do. involved in the training of the algorithm.

And so that's a big question is that hey, yeah, you're inferencing Tik Tok's algorithm, but the weights have been trained in China to, you know, be favorable to China. That still doesn't really resolve the issue. So, I'm sure we'll be tracking that. Um, anyway, let's run through some timeline and then get out of here.

We're doing four hour shows now. Have you noticed that it's just grown? It's crazy, John. We're just doing longer and longer show. We're too addicted to podcasting. The Pope was chosen in two days. Your your remote hire needs doesn't need five rounds of interviews. I said what I said.

Uh so funny that this got 5,000 likes on threads and then 350,000 likes on X as a screenshot. That's what's going on here in this screenshot. Do you see this? 350,000 likes. I haven't seen that on anything but an Elon post.

I I and and you would think remote hire about interviewing like that's not that big of a that's very it seems niche but five million views here 31,000 repost also squared is just banger god yeah I mean the the I should put that in banger archive seriously now the the funny thing is the counterpoint here is that if you're meeting somebody for the first time it's not like the the cardinals were meeting the pope for the first time and they were just like, you know, oh, like we just decided in two days, right?

There's a lot of past history there. Uh, if you're talking to a remote hire, hiring the wrong person is just so painful. It really is the worst for everybody involved, right?

The person who got the job, who ended up not being the right fit, company who hired the person and has to invest all this time trying to ramp them. Yep. And then, um, unramp, dramp, uh, let them go. ramped up. Um, switch your business. Um, no, it's just really painful.

And so if if you're if you're, you know, four interviews in and you're not Yeah. confident, do the fifth one. You can avoid a lot of pain. So, what's the right number of interviews for a remote hire? 50. 50.

Have each of your 50 interns, as long as you get at least a few in-person workouts in that pro in that interview process, I think you're going to be good. Uh, anyway, I wanted to highlight Jesse Pelton. Uh interesting new account. Hadn't seen this guy before.

Elon Musk's been amplifying him a bunch and he's been posting about uh solar energy and energy a lot and just has some great posts and seems like a newer account. I don't know, he might be on there forever, but uh seems to be blowing up most recently.

He said, "If God wanted to build us a type one, if God wanted us to build a type one civilization, a Cardartesev type one civilization, he would have one put a giant fusion reactor in the sky emitting black body radiation around 5800 Kelvin. Uh, which is exactly what we have.

Uh, two, made 28% of Earth's crust out of a semiconductor with a matching band gap 1. 1 EV or so. Uh, which is exactly what happened. And then three, filled the oceans with an alkaline metal. we could use to store limitless quantities of energy, sodium or something similar.

Uh, and he says that would be a crazy coincidence. And he's basically uh super long EV battery energy production. And I just think there's something about this account where I just saw a few of the posts.

A lot of them were uh lightly technical, but for coming from this very optimistic techno perspective that I thought was I thought was interesting. Hadn't seen the account before, so I wanted to highlight it here.

thought uh we I mean we should honestly have him on the show at some point because it seems like he knows his stuff but I want to dig in deeper there.

Uh the other interesting post that I wanted to highlight was from Mike Nuke who's been on the show uh founder of Zapier uh and uh and Arc AGI and he said we are targeting at least three orders of magnitude instead of 3x with ARC AGI 3 but directionally a lot of useful insight on AI benchmark development in this thread and so of Press posts I have a post where I talk about how to build good uh language model benchmarks.

I've had to edit the part where I talk about how you should think about making your benchmark hard multiple times now. And so the first time he says you have to make your benchmark challenging to evaluate language models.

Uh the accurate if you if you drop your benchmark and the accuracy at launch with the best language model is 80% people will see your benchmark is already being solved and they won't even try to build models to improve performance on it.

So you should think about launching a benchmark at between 1 to 35% accuracy for the best model in the game. So take Gemini or take OpenAI's best model and if you if you can get your benchmark to 1 to 35% you're good to launch. Then he edits it in January of 2025.

He says uh due to the extremely fast development I now say that you should do 0. 1 to 9%. Take that way down. Anything higher means that the benchmark is too easy. And then he says, uh, May of 2025. Sorry, I have to make another edit.

Due to the speed of development, I'm now asking people to think, uh, you you you can't just think your benchmark has 0% at launch. You have to imagine that your models would achieve 200% human. The smartest human ever. Yeah. Should get a negative score on the test. Negative score. Find benchmarks that are so hard.

Find questions that are so hard that even if the models improved 3x, they still get zero. Just building a benchmark where models get zero% today might not even be enough anymore. You have to look at how the models have been improving over the past 3 to six months.

Try and predict where they'll be in the future and build benchmarks that would not only make the current models fail, but benchmarks that would make the models of next year fail. Anything easier than that might get saturated much more quickly than you expect. So, I just thought there was a funny uh funny interpretation.

We've been talking about benchmark saturation and the death of the of the benchmark. Um but uh but you know it's interesting to hear someone in the industry kind of put it put put it in such hard terms there.

Uh anyway uh Palmer Lucky is advocating that the United States should immediately expand our naval base on Guantanamo Bay into Liberty City at American Singapore of the Caribbean. Ship building, arms manufacturing, near equator space launch, energy refining and more in beautiful paradise.

Any Cuban who can escape there by running, swimming, boating, or biking gets asylum and a work permit to help build this mega project. By robbing the communist of their worker, the communist party of their workers in critical areas like power generation and military might. We accelerate regime change.

Liberty City will be the first domino. I like how he's thinking outside the box. Found Liberty City. Liberty City GTA reference. Yeah, it's great. Uh anyway, we talked about uh Apple's new product launches. It's going to be a big 2027 that they're lining up for. Um, and a bunch of new iPhones coming out.

Um, what else is interesting to talk about in the news? Orb, but what else? Yeah, this is interesting. The foldables thing is so funny. They got to do it.

But I mean, foldables, it was one of those things where, you know, like it was a interesting idea, but the first iterations even from the top tech companies, not Apple, not named Apple, uh, were bad and everyone kind of knew that they were bad.

And so funny that it used to be it really used to be much more of a thing to text like this horizontal. So I was thinking the foldable phone you bring it back. But I think people just got pretty fast at at typing. Yeah. This one. Swap it on. New hat. You got to do all the different hats. This one's a little tight.

Uh anyway, WorldCoin or launched the Orb Mini designed to make human verification more accessible. It's transportable and seamlessly accommodates your lifestyle. The Orb Mini, it goes. And I guess there's a bunch of uh or uh world stores, showrooms throughout the US.

Now, we're going to hopefully have the founder of World on tomorrow, Alex. And so, it's interesting because so my take on this is like I I understand the main pitch, which uh they they put out this infographic that said like 80% of gamers online say that their online gaming experiences are being ruined by bots.

And that makes sense, like there's a cheater, there's a bot, there's AI. Like no one really wants to go and play chess against the hardest AI possible. That's just not fun. They want to play against humans, right? And so if you play against a cheater, it's just less fun, right? You want to play against real people.

Uh so like verification makes sense there. But what's odd is like if you play out the AGI ASI thing really really far, you would imagine that at some point the AGI figures out how to either steal or grow eyeballs, right? Yeah. like or assemble them. That's really dark. It's really dark. But like stealing the eyeballs.

Ah, I've hit a I've I've hit a worldcoin human verification checkpoint. I should just get a human. Yeah, the future is the a the term keep us in cages and make us do arc ai puzzles and then scan for the world coin. And it's just like there's arc AGI puzzles all day long because we're just like being scanned. Um, brutal.

But the final job UBI is just offering. If you believe in the nanobots and you believe in the the the paper clipping, if you could turn me into a paper clip, you can turn a paper clip into an eyeball, right? And if you can turn a paper clip into an eyeball because it's just matter that you're reconstituting, right?

You're just assembling atoms and molecules. Matter reconstitution. Yeah. So, I mean, I guess that that's farther away. The like the bot crisis is here almost today, right? It is in the sense that like there's a lot of spam out there. We need a better version of capture. RKGI is maybe like six months after being cooked.

Uh and so uh the orb kind of makes sense in that case, but it is it is an interesting like intermediate step. But I don't know. Uh it's it's an interesting story to to to follow and interesting to see how it rolls out. And uh I'll be I mean I I'm I'm excited to see you know there there will be gamers who sign up.

they will test it and they will let us know if their COD lobbies are cleaner, you know, if they're on if they're on Rust and they're having more fun. Isn't the bot problem something you can solve at the software layer though?

Like like obviously if you could get, you know, a device you can get a device that does that that I know exactly what you're saying. Yes, you can get a robotic arm to move a physical mouse and press buttons on boards. Yes. And use computer vision just to look at the screen and you can cheat that way. It is possible.

It's not perfect there. It's obviously better to just go into the code at a lower level and then and then see, okay, I'm not even getting pixels. I'm getting more like screen view. Exactly. Yeah. But but screen view is a is a viable cheating strategy form of botting. Yes.

And so obviously it's much better to just interface with the the net code that's coming across so you can see through walls and you can understand the full map and there's no fog of war, all that type of stuff.

Um, but you can cheat just by camera at the screen and like that's that's almost impossible to detect until the until the the the front-facing camera turns on and it's constantly scanning your eyes to make sure that there's a human on the other end which I mean Face ID is pretty secure.

It does does kind of make sense that there's be something here. Uh, you got to build it in the next Xbox I guess or something. I don't know. Uh, well before we go we should talk about Figma. Figma is uh our latest partner. We talked about this a little bit last week.

Uh it is the design tool that we've used to build the TBPN brand. Every single asset that we create for the most part comes out of Figma and it's a tool that I've used for my entire career. So very uh excited to partner with them and and be on the extended team. Uh we were at uh config last week.

They launched a bunch of new products. The take our takeaway is the Figma for Figma is Figma. Uh I've said it before. Uh but the new product they launch is Figma sites which is sounds simple but is really exciting. Basically you can turn whatever you make in Figma into a website.

Uh so if you're not using Figma already go check them out and uh thank you to the whole Figma team for uh supporting the show. Fantastic. And thanks to you for watching and listening. We'll see you tomorrow and making it this far. Enjoy the bull market day.