Is the Internet Different?(stratechery.com)
stratechery.com
Is the Internet Different?
https://stratechery.com/2020/is-the-internet-different/
41 comments
> Wal-Mart putting their own brand of toilet paper on more prominent end cap shelves is fundamentally different than Amazon promoting their own Amazon Basics toilet paper on a search results page on Amazon.com.
The other fundamental difference is scale. It can physically walk down the toilet paper aisle in Walmart in see all the available toilet papers in less than a minute. In fact, I can probably wander the entire Walmart store in about 30-45 minutes and see everything they are selling in that store. It is not entirely impractical.
On the other hand it is impractical for me to go through all the items Amazon is selling. Thus what they are promoting becomes much more important.
The other fundamental difference is scale. It can physically walk down the toilet paper aisle in Walmart in see all the available toilet papers in less than a minute. In fact, I can probably wander the entire Walmart store in about 30-45 minutes and see everything they are selling in that store. It is not entirely impractical.
On the other hand it is impractical for me to go through all the items Amazon is selling. Thus what they are promoting becomes much more important.
> In fact, I can probably wander the entire Walmart store in about 30-45 minutes and see everything they are selling in that store.
I'm pretty sure this can't be done. Consider how frequent it is to intend to purchase a particular item, which you know the store stocks, and yet be unable to find it even though you're looking for it in the correct aisle.
What if we ran the following experiment:
1. We give the subject a particular item to look for. They know what they're supposed to find.
2. The subject looks through the Walmart for up to 60 minutes for that item. They can't ask for help.
3. We then ask a single question, "is this item in stock at this Walmart?"
What percentage of subjects would answer correctly?
I'm pretty sure this can't be done. Consider how frequent it is to intend to purchase a particular item, which you know the store stocks, and yet be unable to find it even though you're looking for it in the correct aisle.
What if we ran the following experiment:
1. We give the subject a particular item to look for. They know what they're supposed to find.
2. The subject looks through the Walmart for up to 60 minutes for that item. They can't ask for help.
3. We then ask a single question, "is this item in stock at this Walmart?"
What percentage of subjects would answer correctly?
There is a difference between seeing everything (or at least the vast majority of things) and remembering everything. Assuming that for the most part like items are clumped together it doesn't matter if you can remember if Lay's southern barbecue was in stock it matters if, when strolling down the aisle, you were able to see the majority of the chips in one place.
It's a bit like sequential read vs random read. In a physical store it's largely possible to sequentially read the inventory even if you can't quickly search for individual items randomly quickly. On Amazon you're not going to go through all of the aisles sequentially, at best you go in knowing exactly what virtual aisles you want and then you can sequentially search that and even then the best it's going to work out is the worst case of walking through the entire store for an hour.
It's a bit like sequential read vs random read. In a physical store it's largely possible to sequentially read the inventory even if you can't quickly search for individual items randomly quickly. On Amazon you're not going to go through all of the aisles sequentially, at best you go in knowing exactly what virtual aisles you want and then you can sequentially search that and even then the best it's going to work out is the worst case of walking through the entire store for an hour.
You appear to be talking about something fairly different from what my comment addresses.
> it doesn't matter if you can remember if Lay's southern barbecue was in stock it matters if, when strolling down the aisle, you were able to see the majority of the chips in one place.
Perhaps, but I didn't ask about this. The challenge is that you are told in advance that you're looking for Lay's Southern Barbecue. Remembering whether it's there is not difficult, because it's the only thing you want. The difficult part is determining whether it's there.
> In a physical store it's largely possible to sequentially read the inventory even if you can't quickly search for individual items randomly quickly.
Again, this badly misunderstands what I was saying. It is not possible to sequentially read the inventory in a physical store; if this were possible, determining whether or not a particular product existed in the store would be trivial. Instead, this task is one that people routinely fail at.
> it doesn't matter if you can remember if Lay's southern barbecue was in stock it matters if, when strolling down the aisle, you were able to see the majority of the chips in one place.
Perhaps, but I didn't ask about this. The challenge is that you are told in advance that you're looking for Lay's Southern Barbecue. Remembering whether it's there is not difficult, because it's the only thing you want. The difficult part is determining whether it's there.
> In a physical store it's largely possible to sequentially read the inventory even if you can't quickly search for individual items randomly quickly.
Again, this badly misunderstands what I was saying. It is not possible to sequentially read the inventory in a physical store; if this were possible, determining whether or not a particular product existed in the store would be trivial. Instead, this task is one that people routinely fail at.
We're talking about the same thing unless your stance is 100% of people known 100% of the product SKUs they want when looking for things and they know these before they start looking. For those that don't go in knowing the want exactly "Lay's Southern Barbecue 9.5 oz Bag for $2.98" your scenario quickly breaks down into the points I discussed which is why I brought them up.
In more Amazon terms someone is going to search "tablet" not "New Apple iPad Air (10.9-inch, Wi-Fi, 64GB) - Rose Gold (Latest Model, 4th Generation) sold by Apple". Even if searching "Apple iPad" I get Amazon Fire tablets high in the results.
In more Amazon terms someone is going to search "tablet" not "New Apple iPad Air (10.9-inch, Wi-Fi, 64GB) - Rose Gold (Latest Model, 4th Generation) sold by Apple". Even if searching "Apple iPad" I get Amazon Fire tablets high in the results.
Are you suggesting that physically going to a Walmart and walking down the aisle to view all the toilet paper options is more "practical" than typing walmart.com, searching for toilet paper, and scrolling through the options?
Personally, I find it much easier to search for products at home. It's literally seconds versus at least 10 to 15 minutes to get to the nearest store and browse around. And what is the use case for browsing all of the products in a store? I can't imagine why I would need to walk around an entire Walmart.
Personally, I find it much easier to search for products at home. It's literally seconds versus at least 10 to 15 minutes to get to the nearest store and browse around. And what is the use case for browsing all of the products in a store? I can't imagine why I would need to walk around an entire Walmart.
> Are you suggesting that physically going to a Walmart and walking down the aisle to view all the toilet paper options is more "practical" than typing walmart.com, searching for toilet paper, and scrolling through the options?
No.
I am saying that once you are at Walmart, staring at the the end cap toilet paper, it is more practical to just walk down the aisle scanning the rest of the toilet paper available, than it it is to go through the many pages of results on either walmart.com or amazon.
No.
I am saying that once you are at Walmart, staring at the the end cap toilet paper, it is more practical to just walk down the aisle scanning the rest of the toilet paper available, than it it is to go through the many pages of results on either walmart.com or amazon.
I guess that's true, but I don't see the purpose of that comparison.
Also, if the website is designed well, such as Home Depot, it's actually better to use the Home Depot website on your phone to find what you need, and it tells you exactly where it is in the aisle. It would take me longer to search through all the products in the aisle to find exactly what I need.
Might not be the case for an item such as Toilet Paper, but websites that let me drill down into the attribute of the product I need are super useful.
Also, if the website is designed well, such as Home Depot, it's actually better to use the Home Depot website on your phone to find what you need, and it tells you exactly where it is in the aisle. It would take me longer to search through all the products in the aisle to find exactly what I need.
Might not be the case for an item such as Toilet Paper, but websites that let me drill down into the attribute of the product I need are super useful.
> I guess that's true, but I don't see the purpose of that comparison.
There is more noise online, so signal-boosting efforts are disproportionately effective there. That's all.
There is more noise online, so signal-boosting efforts are disproportionately effective there. That's all.
Based on accounts I’ve read of what vendors have to do in order to sell on Walmart shelves (most notably reduce their quality to let Walmart be the lowest price), I think that conflict of interest would actually make shopping on the internet better.
I reject Bork's novel premise that sole criteria is consumer welfare. Proper antitrust assessment is based on competition.
Rephrasing your point:
Both grocers and aggregators compete with their own partners.
At the very least, in house brands shouldn't have unfair advantage.
https://en.wikipedia.org/wiki/The_Antitrust_Paradox
Rephrasing your point:
Both grocers and aggregators compete with their own partners.
At the very least, in house brands shouldn't have unfair advantage.
https://en.wikipedia.org/wiki/The_Antitrust_Paradox
Can you elaborate? All of the examples I can think of where favoring an in house brand is "bad" ultimately end with the consumer being harmed by either paying a higher price or receiving a lower quality product. What are the other "bad" scenarios which don't reduce to either of these?
The Sherman Act doesn't say anything about consumer welfare. It doesn't matter if the price goes up, down, sideways. What matters is illegal anticompetitive actions.
The big problem is that Amazon has third party sellers and there is a strong incentive to use those third party sellers as a stepping stone to create their own brand. Since third party sellers are basically Amazon customers (they pay fees and use amazon services) Amazon ends up competing against its own customers. This is where the conflict of interest lies. Without third party sellers it's just another Walmart
House brands are not in competition with suppliers because house brands are often sourced from the same suppliers and labeling is actually a service that those suppliers offer.
House brands are not in competition with suppliers because house brands are often sourced from the same suppliers and labeling is actually a service that those suppliers offer.
> When it comes to understanding what really makes large tech firms dominant and predicting what judges will do, you aren’t getting the full picture. Maybe I’m just suggesting the site should be more modest and, in particular, be careful about treating aggregation theory as if it were almost like a brand that needs be elevated. The danger for readers is a false confidence, for what Thompson too rarely reminds the reader is what we don’t know.
This section off Wu's criticism I can agree with, Thompson has a way of shoehorning everything to fit his models which doesn't always work.
>> Antitrust should be dealing with the reality of anticompetitive behavior in markets, not ideals of how companies work
> Wu does not believe that Google is unique as far as scalability is concerned, he appears to assume that the company must be doing something nefarious to command such market share. And, by the same token, there must be some sort of unfair lock-in, because again, companies ought not be so dominant. This is a sure recipe for lazy arguments that end up criminalizing the basics of business.
This I can get behind Thompson on. Blindly enforcing laws which are not updated for the fundamental changes in technology will lead to poor outcomes. Laws have to evolve with what technology enables, and Thompson has at least thought about this in a depth that Wu doesn't even care to acknowledge.
This section off Wu's criticism I can agree with, Thompson has a way of shoehorning everything to fit his models which doesn't always work.
>> Antitrust should be dealing with the reality of anticompetitive behavior in markets, not ideals of how companies work
> Wu does not believe that Google is unique as far as scalability is concerned, he appears to assume that the company must be doing something nefarious to command such market share. And, by the same token, there must be some sort of unfair lock-in, because again, companies ought not be so dominant. This is a sure recipe for lazy arguments that end up criminalizing the basics of business.
This I can get behind Thompson on. Blindly enforcing laws which are not updated for the fundamental changes in technology will lead to poor outcomes. Laws have to evolve with what technology enables, and Thompson has at least thought about this in a depth that Wu doesn't even care to acknowledge.
Call me a bluff old tradionalist, but there are costs associated with "internet operations", tech firms: That is, they are or will abuse their enormous insight into the heartbeat of society. If you know for billions of people where they are, what they do, and all other thousands of attributes, what is your end goal? To sit on the money printer? No, tech just moves forward and what is possible will be implemented. I love tech, it's so cool to build stuff. But in 2030 the majority of the population will be flies in a web of tech, surrounding and controlling you in various ways, all day every day. Opt out will be the ultimate luxury item.
I read both articles and maybe I'm just a fanboi, but it really seems to me like Mr. Thompson has thought about this specific question a lot more than Mr. Wu, even if Mr. Wu has more experience in adjacent topics.
I stopped reading when he described users of search as Google's customers.
There didn't seem any point in reading further after such a basic misstatement.
The point about marginal costs is absolutely wrong too, but I don't have the patience to pull it apart in detail.
The reality is there is a customer acquisition cost, it will be spread across multiple domains (technical, legal, financial, and even political) and I would be very surprised if Google believes it's zero.
There didn't seem any point in reading further after such a basic misstatement.
The point about marginal costs is absolutely wrong too, but I don't have the patience to pull it apart in detail.
The reality is there is a customer acquisition cost, it will be spread across multiple domains (technical, legal, financial, and even political) and I would be very surprised if Google believes it's zero.
> [Google]’s scalability is effectively infinite, because serving additional customers is a function of fixed costs, not transaction costs; it really is not comparable to Amazon at all, in this regard, as the companies’ respective market shares demonstrates.
> The same reality applies to Google’s marginal costs (including distribution); while Google spends a tremendous amount of fixed costs on its data centers and networking, any one search is “free”, including Google accepting the search term, computing the result, and delivering it to the user.
Doesn’t google spend some insane amount on infra?
This argument could be made identically for the old retailer down on the street corner by my house. Once you’ve bought the building and fixed the roof and bought all your merch, handing over that t shirt in exchange for cash is free. It’s better than free, even, since people are literally just giving you money for stuff you don’t have to spend any more marginal dollars on!
But if I want to keep doing business, I’ll have to keep paying my mortgage, and paying my sales reps and if I want to grow, I’ll need more space (maybe another storefront per thousand transactions per month?), etc, just like if google wants to grow, they need more sales reps and more land in Mountain View, and for more customers and queries (maybe a new data center per gajillion queries per month, or per upgrade to costlier search processing expensive neural network layer?).
Without commenting on the rest, OP is throwing a lot into fixed costs. You can’t say it’s near zero when the revenue is also near zero. And you also can’t examine the costs in a way justifies zero marginal costs, pointing at the cost of someone who walks into and out of the retailer without buying anything, which is also basically zero (your typical user who never clicks an ad).
Maybe the most apt comparison is a nail making factory or even to AWS where, once you’ve bought a data center, renting it to your customers is free, but that’s not a useful viewpoint. The cost of one single marginal call to your AWS lambda is basically zero, but basically zero times basically infinity is a good old traditional number again.
> The same reality applies to Google’s marginal costs (including distribution); while Google spends a tremendous amount of fixed costs on its data centers and networking, any one search is “free”, including Google accepting the search term, computing the result, and delivering it to the user.
Doesn’t google spend some insane amount on infra?
This argument could be made identically for the old retailer down on the street corner by my house. Once you’ve bought the building and fixed the roof and bought all your merch, handing over that t shirt in exchange for cash is free. It’s better than free, even, since people are literally just giving you money for stuff you don’t have to spend any more marginal dollars on!
But if I want to keep doing business, I’ll have to keep paying my mortgage, and paying my sales reps and if I want to grow, I’ll need more space (maybe another storefront per thousand transactions per month?), etc, just like if google wants to grow, they need more sales reps and more land in Mountain View, and for more customers and queries (maybe a new data center per gajillion queries per month, or per upgrade to costlier search processing expensive neural network layer?).
Without commenting on the rest, OP is throwing a lot into fixed costs. You can’t say it’s near zero when the revenue is also near zero. And you also can’t examine the costs in a way justifies zero marginal costs, pointing at the cost of someone who walks into and out of the retailer without buying anything, which is also basically zero (your typical user who never clicks an ad).
Maybe the most apt comparison is a nail making factory or even to AWS where, once you’ve bought a data center, renting it to your customers is free, but that’s not a useful viewpoint. The cost of one single marginal call to your AWS lambda is basically zero, but basically zero times basically infinity is a good old traditional number again.
It is not free. Ongoing costs are there. They are small per item but still there. Do not confuse opex with capex. they are easy to muddle together especially at first. things like that leaky roof that needs to be fixed, someone to clean the floors, someone to run the register, AC costs, taxes (property and sales), modernization (you do not want your store to look like something from the 60s when everyone else is newer looking), etc Some of those items are operating expense, some are capital expense.
Well, I guess you could say that if the store was run by robots so you didn't have to pay employees and if the shirts were made out of thin air so that there are no COGS. I'm not trying to be facetious. I think the distinction between marginal and fixed costs is important here because it's all about the feedback loop for aggregators. And one can think of marginal costs as "friction" in this loop.
That being said, perhaps aggregation theory won't be complete until Ben considers fixed costs in addition to marginal costs, which I believe might be your argument?
That being said, perhaps aggregation theory won't be complete until Ben considers fixed costs in addition to marginal costs, which I believe might be your argument?
Aggregation theory is nonsense. Not all costs are technological, and similarly not all customer costs are direct. While aggregators exist, you can't understand the market-distorting effects of monopoly without appreciating the entire picture of capital flows within the system from surrounding businesses to the monopolist.
The second order effects are at the heart of antitrust action. Monopolists distort markets in ways that force everyone to pay extra even if they do not pay the monopolist directly.
For a trivial example, consider that dealing with SEO is a huge and expensive problem for any startup. If you don't appear in listings - preferably by buying ad spend - you may as well not exist.
Someone has to pay for that, and those costs will be passed on to customers in most Internet transactions. How much they pay is strongly influenced by Google's choices.
If search wasn't a monopoly those costs would be lower because there would be competition.
This is far from the only example of a market-distorting effect that generates extra costs for businesses and customers because of an online monopoly.
The second order effects are at the heart of antitrust action. Monopolists distort markets in ways that force everyone to pay extra even if they do not pay the monopolist directly.
For a trivial example, consider that dealing with SEO is a huge and expensive problem for any startup. If you don't appear in listings - preferably by buying ad spend - you may as well not exist.
Someone has to pay for that, and those costs will be passed on to customers in most Internet transactions. How much they pay is strongly influenced by Google's choices.
If search wasn't a monopoly those costs would be lower because there would be competition.
This is far from the only example of a market-distorting effect that generates extra costs for businesses and customers because of an online monopoly.
My point is that he’s calling certain costs fixed that I would call marginal. For COGS or data centers you have to choose something to amortize over. For a traditional good, it’s the lifecycle of the physical product. For a data center, you could amortize it relative to its capacity, maybe? Or cost per ad dollar sold, rather than per query? It really depends on the question you’re asking, as fixed or marginal is really fixed or marginal with respect to some factor. It’s unclear what the correct way to do it is, but it’s clear to me that calling it fixed is wrong, relative to the specific points he’s making.
I’d look at how AWS looks at its data center costs and apply that model. Google Search may not be renting infra to searchers, but that’s not the important distinction.
I’d look at how AWS looks at its data center costs and apply that model. Google Search may not be renting infra to searchers, but that’s not the important distinction.
somehow saying
>falling into the novelty trap by asserting things like “the internet has made transaction costs zero” — a sentence that would make any serious economist howl with laughter.
gets translated into
>the idea that the Internet has not had a massive impact on transactions costs.
>falling into the novelty trap by asserting things like “the internet has made transaction costs zero” — a sentence that would make any serious economist howl with laughter.
gets translated into
>the idea that the Internet has not had a massive impact on transactions costs.
> a sentence that would make any serious economist howl with laughter.
A serious economist like Krugman who said the internet was going to have an impact on society no more than the fax machine? Or the serious economists who helped create the asian financial crisis? Or the serious economists who created the housing bubble?
It almost seemed like you think society should care what "serious economists" think.
A serious economist like Krugman who said the internet was going to have an impact on society no more than the fax machine? Or the serious economists who helped create the asian financial crisis? Or the serious economists who created the housing bubble?
It almost seemed like you think society should care what "serious economists" think.
Economists make mistakes. Like any other science. Their uncertainty is larger than something like physics, but not large enough to be considered pseudoscience.
In newspapers they’ll shoot their mouths, but in their journals, they’re quite scientific.
Your criticisms kinda remind me of data science work, where you say “the effect of doing this will probably be between -3 and +7, assuming X, Y, and Z, so it’s worth looking into.” Which PMs (like politicians) take as 2 (+/- 5) and promise the rest of the org a gain of 2. Then we implement it and it comes out as -1 and you need a post mortem about how data scientists (economists) got it so wrong.
In newspapers they’ll shoot their mouths, but in their journals, they’re quite scientific.
Your criticisms kinda remind me of data science work, where you say “the effect of doing this will probably be between -3 and +7, assuming X, Y, and Z, so it’s worth looking into.” Which PMs (like politicians) take as 2 (+/- 5) and promise the rest of the org a gain of 2. Then we implement it and it comes out as -1 and you need a post mortem about how data scientists (economists) got it so wrong.
Much of macro economics is unlike physics not only in uncertainty, but more critically in lack of experimentation or falsifiability.
https://www.thecrimson.com/article/2013/12/13/economics-scie...
http://skepticalphilosopher.blogspot.com/2009/02/falsifiabil...
(I'm aware there are branches of economics which are based on experimentation but they seem to not be the norm)
https://www.thecrimson.com/article/2013/12/13/economics-scie...
http://skepticalphilosopher.blogspot.com/2009/02/falsifiabil...
(I'm aware there are branches of economics which are based on experimentation but they seem to not be the norm)
Well, what actually matters in reality is how they shoot their mouth off in newspapers.
In this respect, they are mostly quacks. Peddling psuedo science on the basis of pretty weak while rigorous science.
Economists should stop shooting their mouth in neespapers as if they know something, that's being misleading. And hiding ideology behind _math_.
Economists should stop shooting their mouth in neespapers as if they know something, that's being misleading. And hiding ideology behind _math_.
To me it seems like economics is just above numerology. Operationally it's a state pseudo-religion. The priests consult their tea leaves or goat guts or whatever and advise the kings, and no one ever talks about how full of shit they are, because that's their whole purpose.
A scientific economics would be grounded in ecology and anthropology. It would talk about calories and niches. It would make reliable predictions.
A scientific economics would be grounded in ecology and anthropology. It would talk about calories and niches. It would make reliable predictions.
Can't trust doctors because one time a doctor made a mistake!
Flawless logic!
Flawless logic!
I hope we can both agree the standards of measurement for doctors and economists are not exactly comparable, given that one is just "is the patient better" and the other can be judged in any way imaginable. Is the economist macro, micro, do their solutions add to income inequality, or reduce it? The field of economics is clearly much more mercenary than the field of medicine.
One time... Okay.
Also, my dig wasn't at economists as individuals, but the pseudoscience called economics as a whole.
There is a difference between a "mistake" like oops gave you the wrong medicine by accident and a pseudoscience like chinese medicine.
And the internet prediction, asian financial crisis, housing bubble, etc aren't "mistakes"? It's just shows "serious economists" are just charlatans. The priests of modern era.
Also, my dig wasn't at economists as individuals, but the pseudoscience called economics as a whole.
There is a difference between a "mistake" like oops gave you the wrong medicine by accident and a pseudoscience like chinese medicine.
And the internet prediction, asian financial crisis, housing bubble, etc aren't "mistakes"? It's just shows "serious economists" are just charlatans. The priests of modern era.
You are being to harsh on chinese medicine.
I like the way Richard Wollf put it: There is a reason economics and business school are separate buildings.
I like the way Richard Wollf put it: There is a reason economics and business school are separate buildings.
I don't understand what you are trying to say here? Zero and near zero are effectively the same.
No, they aren't. "Zero" is zero, it means no cost. "Near zero" means some cost. Some cost, in aggregate, can easily become a major cost. Ever wonder why manufacturers of physical goods choose the component that's 1 penny less? Because that one penny can translate into millions in savings over the lifetime of a product. "Near zero" could translate into millions in costs over the lifetime of a product versus "zero".
The discussion is around financial and market dynamics for different types of companies. Quibbling over the definition of zero vs near-zero is completely side-stepping the argument.
The point is, for Google search variable costs are immaterial. This is not the case for any business that needs deals in physical inventory or human-provided services (which was the majority of businesses pre-Internet).
Does Google incur some marginal infra costs from increased traffic or growth of the web? Yes. If you project out infinite growth does it become a "major cost"? No, the limit of profit over variable cost approaches zero.
The point is, for Google search variable costs are immaterial. This is not the case for any business that needs deals in physical inventory or human-provided services (which was the majority of businesses pre-Internet).
Does Google incur some marginal infra costs from increased traffic or growth of the web? Yes. If you project out infinite growth does it become a "major cost"? No, the limit of profit over variable cost approaches zero.
>The point is, for Google search variable costs are immaterial.....
>No, the limit of profit over variable cost approaches zero
Ok well Google is an example, but the argument is about Internet companies not just Google, but ok Let's take Google and use that to go back to the more general level.
aside from any other arguments -
is it possible that variable cost approaches zero because
Google is a monopoly with a high profit margin. If the profit were to decrease a lot from each transaction then they would be at a position where variable cost becomes more relevant.
That is to say Google's special place has nothing to do with it being a magical internet company but rather its being in a seller's market that they dominate.
If it were suddenly to no longer be in a seller's market or to have to do stuff like save data longer to get value from it, or to earn much less per search because advertisers didn't want to pay, or to find that there are side effects from having too much traffic that ends up decreasing ability to serve relevant data because of signal to noise ratio thus causing a loss of profit etc. etc. we would end up with the cost to Google of a transaction being much closer to what they earn from that transaction.
In the last scenario I put up there, which I believe could actually be a problem if you project out infinite growth (search growth) does it become a "major cost", yes, in such a scenario too much traffic decreases the value of all traffic.
In fact too much traffic decreasing the value of all traffic is a well known feature of Internet companies, because internet companies actually use physical devices (just wanted to point that out, since there seems to be this special thing about Internet companies not having physical inventory) to serve their customers or users (depending on which we're talking about here) and this is why DDOS is a thing and why even natural spikes in usage can be problematic. So in fact, I would actually say about just about any online company there is - if you project out infinite growth ( of traffic ) does it become a major cost?
Um probably, and depending on how that a growth is structured it might in fact be a crippling cost.
Ok well Google is an example, but the argument is about Internet companies not just Google, but ok Let's take Google and use that to go back to the more general level.
aside from any other arguments -
is it possible that variable cost approaches zero because
Google is a monopoly with a high profit margin. If the profit were to decrease a lot from each transaction then they would be at a position where variable cost becomes more relevant.
That is to say Google's special place has nothing to do with it being a magical internet company but rather its being in a seller's market that they dominate.
If it were suddenly to no longer be in a seller's market or to have to do stuff like save data longer to get value from it, or to earn much less per search because advertisers didn't want to pay, or to find that there are side effects from having too much traffic that ends up decreasing ability to serve relevant data because of signal to noise ratio thus causing a loss of profit etc. etc. we would end up with the cost to Google of a transaction being much closer to what they earn from that transaction.
In the last scenario I put up there, which I believe could actually be a problem if you project out infinite growth (search growth) does it become a "major cost", yes, in such a scenario too much traffic decreases the value of all traffic.
In fact too much traffic decreasing the value of all traffic is a well known feature of Internet companies, because internet companies actually use physical devices (just wanted to point that out, since there seems to be this special thing about Internet companies not having physical inventory) to serve their customers or users (depending on which we're talking about here) and this is why DDOS is a thing and why even natural spikes in usage can be problematic. So in fact, I would actually say about just about any online company there is - if you project out infinite growth ( of traffic ) does it become a major cost?
Um probably, and depending on how that a growth is structured it might in fact be a crippling cost.
Ben Thompson explicitly states that he is not talking about all Internet companies, and that is one of his beefs with Tim Wu's critique—that it is reducing his actual argument to a caricature.
To your point, yes I agree that this hinges on the specifics of Google's business. If, for example, their advertising business was made illegal in its current form it would completely change the economics.
To your point, yes I agree that this hinges on the specifics of Google's business. If, for example, their advertising business was made illegal in its current form it would completely change the economics.
if the cost was 0 geocities would still be running.
Wal-Mart putting their own brand of toilet paper on more prominent end cap shelves is fundamentally different than Amazon promoting their own Amazon Basics toilet paper on a search results page on Amazon.com. Amazon has a much greater ability to drive conversion to its own products than a brick-and-mortar retailer (e.g. limited screen space, UI, information hierarchy, better data, etc.)
Thompson's argument that "it's all the same," "private labeling has been going on forever," is flimsy and intellectually lazy.
Once we agree on this basic set of facts the more interesting question becomes: is this dynamic bad for consumers long-term?
The crux of this round of tech anti-trust scrutiny is time horizon. Many of these practices are neutral or beneficial to consumers in the short-term, but that's always how monopolies operate. Undercutting prices to put a competitor out of business benefit consumers immediately, but hurt consumers after the competitor goes out of business and the monopolist can raise prices. Amazon selling Amazon Basics products at a loss or near-loss is good for customers now, but if it puts too many other online retailers out of business, eventually they'll raise prices and hurt consumers.