A Nobel Economist's Plan To Tax Digital Ads
33:11 Share

When he was brain cog man, i'm a black, gay, non binary author, T, V, writer, actor, and i'm messy, but not in the way you think mess, as in i'm human and flaw. I'm on a mission to destroy shame around sex and the only way to do that is to to talk about sex. So that will do on my brand new podcast, tell me something messy, join me on tell me something messy with brand new episodes every thursday on the eyes, heart radio p apple podcast or wherever you listen to.

Causing media, hello, welcome to Better off line. I'm your host S.

Back in march of this year, IT economist, their own, a MIT sloan Simon Johnson, published paper called the urgent need to tax digital advertising, where digital advertisers will be levied with a flat tax, a fifty percent on all revenue above five hundred million dollars. And so I am joined by their own to walk through IT their own. Thank you for coming on.

Thank you. thanks. That is my pleasure.

So walk me through the idea. That seems pretty simple, but maybe there's more unions to IT.

What is very simple to implement is very simple to explain. The question is why, you know, why do something like this? You know, after all, who doesn't love something free? And we're getting a lot of three things because the entire online ecosystem is monetized via digital ads, right? The issue is I think this ecosystem creates a number of problems which are set to get worse with A I right.

The most obvious of those is something that's been discussed quite a bit over the last few years, which is dead. Monetising via digital ads means that you are going to encourage platforms to call leg more and more data about people. And this creates a much more intrusive, lower privacy and potentially distorted sort of system.

right?

The second, perhaps quality or separate thing, which has started reach receiving more attention, is that this also means you want to make sure that the eyeballs are glued to the screen, which are large screen, small screen. And that itself might generate a lot of other problems, including perhaps mental health problems, because you try to get people to stay on the platform by inducing strong emotions, uh, envy, jealousy, anger, outrage and so on. And sort of this creates a very different type of social environment than we are used to win widespread negative consequences.

But I think the one that i'm most worried about is that this reduces competition, right? IT is impossible today for any service by a newcomer to come in and say, okay, you're gonna get this from google, facebook, instagram for free, and i'm gonna provide higher quality content for which you have to pay a subscription fee or something else. Especially difficult because newcomers are not gonna have the network, there's going to be uncertainty about their quality.

So IT really cement sa system where everything's gona be monotoned via digital ad, and that's gonna discourage the entry of new products, new services and especially new technologies. For example, everybody worries about what social media and other online platforms, and A I implies will imply for democracy, well, something that actually creates more pro democratic conversations, higher quality content. That's gonna very difficile to get off the ground today.

right? So how does this tax actually change the incentive though?

Well, I think the main way in which changes the incentives is that IT makes IT possibly more likely that both existing platforms and new platforms will now say, let me experiment with new products that are higher quality for which I can build. I clean tell via subscription fees or other things because if I get hundred million dollars via digital ads, hopefully is gonna go away. If I can get that money for a subscribe fees that's good, is no longer so inferior to doing IT via digital ads.

But I think one of the things that concerns me with that, or maybe it's less of a concern, more just the after effect, will be is the people are used to not paying for social media.

That's why this seems to be high. You know, if this was a ten percent tax, IT wouldn't make so much of a difference. So in some sense, when I mentioned ecosystem, I was intending to employ by that both what the offering side, the platforms are doing and what the consumers are expected.

So is the syngas tic relationship between users, consumers and the firms. And you want to change that relationship again. You wouldn't want to do IT if the market system was working perfectly and everything was hanky dory. But I think there's a lot of evidence that's not the case at the moment.

So and the incentives would be that they just can't make that much money off of dita ads, so they will have to find new business lines. But could this not potentially kill off the idea of the free social network?

Well, first of all, I think that idea has become excessive. If we scale IT back, IT wouldn't be so bad. And in some sense, I think the purpose is to scale IT back because once everybody expects everything for free, that does create a race to the bottom in terms of quality, in terms of data protection, in terms of you know new technologies that you know will actually change the face of the kinds of offerings that we get.

right? And it's I remember in the past, we've had companies try paid social networks. And I just hasn't worked what?

I didn't work because they were up against free social networks digitalize. H, so I monitise ed by a digital ad, especially at a time when people didn't understand what the costs of these things were. You know, I think even today. There are consumers who do not fully recognize the amount of data as being collected about the so if you go back ten years ago, I think both the addictive or quiz I addictive nature of social media, the extensive data collection and how that data can be used both, you know, to guide you toward certain products, perhaps to behavior, if you like you, or perhaps to charge your higher Prices. All of these things were not completely understood.

And I think I I get what you mean. It's like the the latest stage model of instagram and facebook is so different because when everything's free, you don't really have a mechanism to control the customer other than just tricking them just continually.

The inking is probably the wrong word.

There is tricking.

There is tricking. It's not everything. So you when you get a mad, it's useful for finding out about products. But what is the trade off between how much of your private data you want other people to have verses getting some of those products? And second, once people once platforms have access to that data, what is there to stop them from offering some of the manipulative products as well as some of the useful products?

At your job, do you ever have to deal with a nose roller about a snub pulling of you? Installing a new convarted t system dealing with the different components can sound like you're speaking a foreign language. Luckily, you've got a team ready to help rangers.

Technical product specialists are fluent and maintenance, repair and Operations. So whenever you wanted talk shop, just reach out call, click anger outcome, or just not buying granger for the ones who get IT done. My name is brain cog man.

I'm a black, gay, non binary author, TV, writer, actor. And i'm messy, but not in the way you think messy, as in i'm human and flaw. I'm on a mission to destroy shame around sex.

And the only way to do that is you to talk about sex. So that will do on my brand new podcast, tell me something messy round of smash or okay here, this measure, pass, spit play. I don't know.

I don't know how I feel about bottling flewed ds, be an on unless because.

We're doing the .

pull were living on the age, my god.

I was not a baby. Like I always say, if you know how to work, that body, that sexual is and that heart, you're unstoppable. Impress your power. That's really what we gona do on the show. Join me on telling me something messy with brand new episodes every thursday on the I heart radio apple podcast or wherever you listen to podcast.

So perhaps the biggest thing because I just while prepping for this episode, I went and did some man and what you are suggesting would kill matter. Now i'm not saying you you're suggesting you should kill meter and just saying the matters for the last four quarters, the net income was fifty one point three five billion of a hundred and thirty billion dollars worth of revenue. So this would make their business model. And tinder is that.

But first of all, no, I mean, you know instead of fifty five billion, if they make know thirty billion, it's not the end of the world. I would be happy to have a company that is s thirty .

thirty billion of .

as as would I but the hope is that they would also then offer products and services that are higher quality, that are subscription based. You know we know people still you know pay for certain things, although the less like netflix, netflix is also now being sort of being forced to move more in that direction.

You know is there a world in which higher quality netflix, for which some people pay is viable? Well, well, as you know, some lower quality one finance, a mode via ads also coexist. I think those are questions and I think opens up the market system to different a different sort of forces. So if there is a paid whats up that has Better security features and you don't get these completely unexpected and knowing things, uh, that have started popping up and what's up perhaps that attractive vist, some people will be happy to pay ten dollars a year .

for that kind of feels like the advertising model is the lazy man's business as well because IT doesn't incentivise you to make a Better product to incentivises a monopoly.

You know, how do they arrive to idea my stick for the last, you know, two decades at least? And especially with regards to A I and dividual technologies, is these technologies are extremely valuable. We can develop them in different ways.

We can have higher quality technologies that increase worker productivity, or we can have rot automation, lazy automation, ripe and have addictive social media that makes you go into your cocoon and you become uninterested in forming bridges and engaging in democratic citizens. Or we can have platforms that actually encourage different types of communications and the active citizens. Real democracy requires. Which ones of those will be? And I think since we have gone more and more in the automation way and in the in the sort of toxic environment of social media, IT change requires new products, new technologies, and that means a more open system.

And so the way that I sort of started sort of coming towards this idea was, well, what's a barrier to this? Well, the fact that there are a few big tech companies that can acquire competition is a barrier, but so is the fact that even if they don't acquire you, they pigeon hold you into the same ecosystem in which they exist. And I think that's a big problem.

How do you mean they put on hold you even if they haven't acquired as user.

to compete against them, you need to also offer your products for free, which means the only thing you can do is for digits advertising. If you quality data you can't afford IT, you're not going to have an incentive to get IT. If you need for high quality niche products, you know, nobody, he's going to buy them because every boy's used to this free free reason, everything.

right? So he creates a kind .

of I yeah yeah raised the bomb.

but also just there's no way to there's no reason they would have to compete like you couldn't compete as they we haven't really seen any new social networks in years other than blue sky and threads and threats is a part of a part of the facebook machine.

And we have seen tiktok. But tiktok is just a amlie ation of the same business model, same motivation system, same sort of weakness as being exploited as instagram, facebook and so.

And they burned, and they burned billions and billions of dollars to get there. IT wasn't like the people join. I don't know why people possibly thin this, but they frame, bite, dance, some lucky upstart. Vers is just a massively funded chinese jog. Ont, actually.

why I actually use the word ecosystem, because this model itself is highly syngas. Tic wood companies growing very fast. And why do they grow very fast? Because they want market.

I want to dominate the the market, but I also want data. And how is that possible? How can you grow so fast? Well, you get venture capital or microsoft to bank rollers in the case of open a ee. So you burn through a lot of cash early on in order to get data dominance and market dominance. And again, ran not a pro competitive picture that's emerging here.

I feel like generate a vi is something separate because with that, they but they don't have a business model yet.

They do not have business model. But you know what that does? IT makes IT even more likely that we're gonna just repeat the same sense of earlier social media.

Look at what is OpenAI doing is burning through a lot of cash in order to acquire market share and data. It's not monetizing through anything. And and I think the most likely cash car ideas that are coming up, we're gona use generate A I for internet surge and we're gonna over the add revenues. Well, that's more of the .

same and and I don't different conversation. I think that would work, but it's nevertheless less. I can see the idea. It's funny. These incentives really have i've never really thought about that before this conversation, but IT feels like digital vertigo really did harm the tech industry in that they found something very profitable that got more profitable without making the users life Better at all.

Well, Larry parents are gay. Brin, when they first form their company, said, we don't want ads. That's not like a good model.

Yeah, it's that to problem the .

moment somebody came and gave them venture capital money and said, but we do IT so that you can actually tizer these laws.

Yes, what if you are super rich? But IT doesn't sound like this would kill that this lads industry.

Because if I don't, I don't think we should kill IT either. I mean, I think again, if we created A A monoculture that won't be good if everything was. On online, everything was based on subscription. That wouldn't a good model lighter.

So you want a variety, but so well is more Operational level. I'm guessing that this would also have to have some way of cutting through things like if people set up subsidiaries to try and pass off the tax as well, you would have to make sure that there was a way around.

you know, that's why this is not a thing of graduate attacks that is easier to game. But if you know five hundred million, you know it's like a very small amount of money. So you just set a minimum like that, perhaps a little bit more so that the really small companies are not burden by IT.

We know one problem, for example, with the european and GDP r is that once you put a regulation like that, the burden is heavier or on smaller companies. And so I actually is a competitive advantage for the large platform. You don't want to do that.

But I think five hundred million word actually, that would make a very healthy market for smaller companies.

which is great. We want more smaller companies as well.

But wouldn't I I think that this would help with the problem of a great deal. But don't we also need something to do with the algorithms themselves? Because I feel we could do this and I would solve some problems that would begin incentivising them in the right direction. Theoretically, yes, but you still have this problem of they wouldn't stop algorithm. In fact, this might encourage them to be more algorithm ally.

And that's why I don't think this is a silver bullet. You know, if you need a uh a range of policies, but the hope is that at least such a policy would create some push towards some platforms and products to have Better quality algorithms that don't trap you like that. But for example, if we're talking about social media, we should also be talking about repeal.

A relaxation of section to thirty of the digital uh uh uh act so that you know, companies that algorithm ally boost content cannot then say, well, you know, this is not our speech, this is somebody else is we have nothing to do with IT. So so there are some details there that we have to think about. And and I think section to thirty, which was written a different age in the one thousand nine ninety, is definitely not up for dealing with issues of algoma boosting and debating.

Section to thirty aside because that is also a separate conversation. IT does feel like we have just kind of not given tech companies is much responsibility for what .

they're doing. I just just none. That's the main reason why I wanna have a conversation of a section to thirty because I can be optimal that the that corporations that are arguably the largest and the most powerful humanity has ever seen can then wash off their hands and say we're not responsible for anything that happens on our platforms.

Yeah IT feels with with two thirty and I am not as well red as I should be about IT. But IT feels as if that responsibility side is the real niggling issue, but also the most stories, because on some level that makes sense with social media platforms, that they should be able to say we are not responsible for all the posts. And if they had to be.

they and look, I think free speeches is a major issue. I am worried about the road of free speech, but you know, i'm also not like a one hundred percent free speech. absolutely.

I think we have to baLance IT and the way that I would suggest as is again. So now we're changing topic. IT is fine, thanks to thirty is you know subject to whatever legal requirements there are you know abusive content at setup.

If you post something on instagram or facebook and the company doesn't algorithm promoted, then it's completely right. It's there is my speech. If my friends find out about IT, they can go and look at IT.

If somebody stumbles on IT, that's fine. But already make boosting is like new york times putting you on the front page. We say new york time is putting on your front page and you can see all sorts of lies and crazy things.

And this just edom of, well, yeah, okay, yeah, yeah. Now I know what you may know. If you have a more, more preferred newspaper, i'll use that example.

No, no, no, I know what you mean. IT almost. And IT seems like the problem is that you're saying is IT isn't two thirty itself, is the fact that the algorithms boost these particular .

things was fine for the agent was written where nobody could could have understood the onslaught of alarm, boosting, promotion and manipulation that would come.

And he feels like it's an instance problem that we just it's almost that platform should either move away from the algorithm model because if you because the amendment section to thirty that just says I should go away kind of doesn't sit right with me. I don't think you should IT would make platforms like wordpress impossible to years. Or like logging platforms, the idea of what they promote being more intentional, al, that may .

actually really like, I ginst lately protect wordpress. World press doesn't promote.

right?

So this sort of reformed to thirty is completely finding that respect.

So as far as this fifty percent tax, do we have any kind of historical precedents for something .

like this being done? That's why, you know, am I shore, fifty percent is the right? That will absolutely not. There have forty percent. There have thirty percent.

but I think fifty. great.

Thank you. Well.

because the thing is, the way I look at IT is these companies are so adept to avoiding responsibility and they're so unwilling to change their ways and so unwilling to be responsible with what they're doing that IT needs to be like this. I'm sure that if this actually got anywhere nearer government, they would have the world's biggest tantrum. But I feel like we need stuff.

needs to be a shock, the absolution.

I think that what actually maybe that's the question. How practical would this be to actually implement?

Well, that depends on what you mean by that.

I mean, how how hard would this be to get this to get this actually into existence with the government able to level that .

tax IT would be extremely hard because of the lobbying, as he pointed out. But if there was agreement, if the chinese government wanted to do IT, they could do IT overnight. It's it's all out there.

The usual add revenue is there is measured, uh, you know, all and and the flat tax is very easy to implement. You could do IT IT source. When no advertisers companies pay for advertising.

You could do and invent many different ways. So very easy. And we and is a version of variety of V A T like evolution taxes.

We have them in under much more complicated situations when there is much lower quality data about what's going on. You know, many middle income countries have a very complex sales tax or A V A. T. tax.

yeah. And IT almost feels that you could have a little fun with that as well. Maybe you could do this fifty percent tax and free IT directly into some sort of national venture fund that would end up funding the view IT IT feels as if the as go, like on a larger scale.

IT feels like governments artist, twenty years behind technology, not even putting a side of in section to thirty. IT feels like we should have an E, P, A or an F, D, A for data. IT feels like we should have ways of actually, we don't know how these algorithms work, and he just feels a little crazy to me.

absolutely. But even the way I would say that is i'm not sure that I would say they are twenty years behind in terms of knowledge. I actually oh.

not knowledge, just .

legislation, legislation. yeah. So like a completely different but related topic is we actually need and infrastructure for data markets.

right?

So you know, I think everybody says data is gna be one of the most important factors of production for the future, more important than land. Imagine that I told you that today, land that is still pretty relevant for many businesses. It's up for grams. You can just go and get whatever piece of land you want and you don't have to pay for. You know, that tragedy of the comments is well understood from history, will be disastrous yeah .

we just it's just a wild's rest.

Yeah wild's west. Nobody produces high quality data. Nobody gets compensated for the data that they have encourages. You know, the monetization, that model that we talked about where you actually sweep people's data without their permission or without their understanding and you try to monitise IT what I did to.

My name is brain cog man. I'm a black, gay, non binary author, T, V, writer, actor. And i'm messy, but not in the way you think messy, as in i'm human and flaw.

I'm on a mission to destroy shame around sex and the only way to do that is you to talk about sex. So that will do on my brand new podcast, tell me something out of smash or okay here. This measure, pass, spit, play.

I don't know. I don't know how I feel about bodily fluid being an onion less. It's because.

Are doing the pool living on the age.

Md, I was not .

expecting. Like I always say, if you know how to work, that body, that sexual and that heart, you're unable impress your power. That's really what we will do on the show. Join me on tell me something messy with brand new episodes every thursday on the eye art radio APP apple podcast or whatever you listen to podcast.

So a complimentary thing is we we think about how how is that, that we can have a system where data producers are encouraged to produce higher quality data so that perhaps, you know, next version of a large language model learns not from redit, but learns from high quality din domain. Relevant expertise.

I think in that case, I think in that case, i've not sure I agreed. Just because large language models getting high quality data is APP like how you build them? Would like IT just highlight ality date, would just mean they still need more.

But I get we mean that if these companies were incentivised to actually have good data and provide good services with the data, that would be Better because that's the thing. I'm not I don't love digital targeting. I don't love any of IT. But man, if I have all this data, while all their services so impersonal, they don't feel like there for us.

why does microsoft word always crash?

Yes, I know the that that ones IT doesn't matter. They have the dog and dog x model. They have a tiny little kind of monopoly. Now though they were very sense. Just actions have change.

Let me actually push back a little bit on generate A I okay, because I think your statement makes sense. If you buy in that generate AI is most productively developed in the form of general purpose human like checkbox.

okay.

they will need like huge amount of data, because next word prediction is very, very inefficient. And you need to timitch humans in and variety of circumstances. But imagine that we use generated A I in a very domain specific way. You know, I wanna know what drug creates, what side effects in conjunction with other drugs, right? For that I don't need my generate a eye tool to communicate with me and you know human like fashion or right experience, so that I just need some very specific domain c expertise, but that high quality data .

is actually not out there. And I fully agree with that. And smaller language models, focus language models, those make sense. But I think we are actually agreeing because this is an incentive problem. The reason OpenAI isn't really focused on that is because doesn't make .

that much money. But OpenAI itself .

is not making a lot of money.

I think I think they are not focused on that because a if you want to get big very, very fast and collect a lot of data, domain specific models are not gna work. So creating hype around something that sounds very intelligent is a much Better tool for doing that.

And second, I think in the industry is still in an unhealthy way, in my opinion, preoccupied with artificial general intelligence, right? Human like intelligent, even if that's not what we need, even if that's not what's feasible, even if that's got a lot of outsides. So that's why I don't, in my opinion page, sufficient attention to these domain specific .

expertise m models. So I feel I actually agree, but also that that is economics problem. That is a these companies are incentivised to grow at all costs to get as big as possible is what you're suggesting .

that IT is IT is not just an economic problem, but IT has an important economic leg. If we did not have venture capital be so important, this model would not have got ten off the ground.

So how do we push back on the VC model? How how do we actually make technology work without that model? Because IT, I agree, it's the growth or costs build as big as possible than IPO, and everyone gets rich other than the user. Well, what are the alternatives?

I I don't know. I mean, I not so much of an intervention is that i'm going to say, you know, you should tax V C as well. But I think one thing that encourages the V C grow at all costs model is that we don't have any entire trust, right? There was a very strong entire trust legislation and implementation, then becoming so big wouldn't be so attractive and you wouldn't be able to acquire all your competitors in the process as well, which is a very important part of this, get very big, very quick. So I think our big failure of you know, upholding existing entire trust laws and introducing new entire trust laws appropriate for the digital age, I think, has contributed to this problem.

Is there a way of innovating VS? So actually, this is how we this is the final question. You might actually have an answer here.

How do we incentivize venture capital and the start up industry to start investing at the earlier stage? Because a big fund just gave back a chunk of their their fund because they were not finding as many opportunities in the late stage. And IT, like most of the money goes into that late stage, how do we incentivize .

that is a much harder problem because even well qualified VC are not gonna have an easy time recognizing a very promising product when is in the garage stage. But and this brings another you know set of issues, the fact that we tax capital so lightly contributed to this. Again, we're subsidizing vcs because if you make your money via v as as sort of investment on your capital, you pay very little tax. You know, if you are, you know, a tech billionaire, I want name names.

You don't .

even pay yourself a salary. You keep on borrowing money from venture capital or other other specialized financial vehicles, and you pay yourself out of that. Everything is capable of income. You pay minimal taxes. So our tax system, already a big contributor to inequality, actually also distorts the digital landscape. So one simple thing again, which is probably even less likely to be implemented in our current polarized environment, is that tax capital and labor at a flat rate, uh, the same or you can act what program you want but you don't not distinguish capital and labor income if you going to tax labour income at thirty percent, tax covered at thirty percent.

I agree on that one. IT just feels like IT would be great if there was a way of just almost in because you're talking a lot about taxation such an I agree that we need those controls there. Is there way to incentivize them to put more money earlier, make more risky bets? Are not saying a deduction because that would be crazy, but IT feels like that would also solve in part in concert with these taxes, a way of getting that money into the area ecosystem.

There maybe, but I don't know exactly. We should think more about IT, but there is an alternative solution, yet another policy proposal, as we fun a federal agency on A I, which is task both by communicating and developing best standards on things like privacy data, you know, a ice standard, security, safety, but also has deep enough pockets that I can play that incubator role for, especially for technologies that are deemed to be socially beneficial.

So if we have technologies that actually protect in user of privacy and enable users to make Better decisions that the kind of thing that the government should put money in early on, some of IT will go to waste, some of IT will go back up. But if a few of them are successful, great. And if the alternative is that the VC are going on a develop the technologies that are most manipulative, there is even more reason for putting this money.

So thank you so much for joining me again. It's been such a pleasure.

of course, that this was my pleasure. Thank you for a being interested in these issues and having such .

a great conversation.

Thanks so much.

Thank you for listening to Better roof line the editorial um composer of the battle offline theme song is matter sales cape.

You can check out more of his music and audio projects, a meta sale, sky dot com M A T T O S O W S K I dot com, you can email me are easy at Better off line 点 com or visit Better off line 点 com to find more podcast links。 And of course, my newsletter, I also really recommend you go to chat, that was your ed, or at to visit the discord and go to ask lash betroth line to check out our redit. Thank you so much for listening.

Better off line is a production of cool zone media or more from coal zone media. This is our website. Clues on we get a dot com or check us out on the I heart radio up apple podcast or whatever you get your podcast.

My name is a cow, good man. I'm a black, gay, non binary author. T V, wider actor end, i'm messy, but not in the way you think messy, as in i'm human and flaw. I'm on a mission to destroy shame around sex and the only way to do that is to talk about sex. So that will do on my brand new podcast, tell me something messy, join me on tell me something messy with brand new episodes every thursday on the I heart radio APP apple podcast, or wherever you the test.