select all

‘We’ve Lost 10 Years of Innovation. This Decade Has Been Boring for the Web.’

Over the last few months, Select All has interviewed more than a dozen prominent technology figures about what has gone wrong with the contemporary internet for a project called “The Internet Apologizes.” We’re now publishing lengthier transcripts of each individual interview. This interview features Ethan Zuckerman, an early staff member at widely credited with inventing the pop-up ad. Zuckerman is now a professor of media at MIT.

You can find other interviews from this series here.

I wanted to reach out to you because you work at MIT, but also because of your writing about having a hand in building the pop-up ad. So, to start, given that you’ve written so extensively about advertising as the internet’s original sin, I’m kind of curious about what role you think advertising has played in getting Silicon Valley to this particular moment.
Well, it’s really been the heart of the argument I’ve been making for a couple of years now. I think the main thing that most of us didn’t really consider is that once we got into a world where everybody could publish, the thing that becomes scarce is attention. And when you’re fighting each other for attention, suddenly, fighting advertisers for attention seems like a really bad idea. So, once content is incredibly plentiful, and attention is what becomes scarce, you’re locked into this really strange economy, where you’re trying to provide a service, you’re trying to get people to pay attention to it, and simultaneously, you’re also trying to get people to go away, and go pay attention to someone else.

It’s a very different tension than someone like Google has. So Google has the first model that really works on the internet, which is directed advertising. Where someone basically says, “Hey, I’d like someone to come fix the hole in my roof; I’m in this town, find me a roofer.” And Google doesn’t really want to hold on to you; Google wants to send you on to a roofer very quickly. The roofer wants you to show up as a qualified lead. Everybody’s happy; everybody benefits.

But when I go onto Facebook and I grouse about how much it sucks that it’s still snowing in western Massachusetts and my roof is leaking, I don’t really want someone to lure me away at that point. I particularly don’t want to hear about a vacation; I don’t want to hear about the new car that would make my life happy. What I really want is to grouse and get sympathy from friends, so Facebook at that point is in conflict with me.

And I think what we basically fail to realize is that, in the vast majority of these cases, we’ve sort of picked a revenue model in which we’re in conflict with one another, rather than working coherently in a way where we all end up feeling good about incentives being aligned. For me, the moment where it went wrong was when we failed to realize that there were two kinds of advertising, and that one of them might be healthy for the internet, and the other one probably wasn’t.

A thing that comes up a lot is the idea that the business model of these companies might be in direct conflict with the interests of its users — that growth, that getting as many eyeballs as possible, becomes the goal of business that cut against the users. Is scale a problem?
For me, the distinction is intention versus a graphic psychographic. So when someone’s saying, “Hey, I’m looking for this, would anyone like to help me?” and you help them, I feel like that, generally speaking, is a pretty ethically sound business model. It’s one that works reasonably well. It’s certainly worked very well for Google. When we’re at the point of “I don’t know why anyone would want this, but let me go out there because I think young men would like this, if they’re 18 to 25,” that’s where I think things get trickier. What happens is, suddenly, you’re looking not only for patrons and attention so you can sell the ad, but you’re looking for ways to tell you that they’re 18 through 25 and male.

As soon as you’re saying, “I need to put you under surveillance so that I can figure out what you want, so I can meet your needs better.” I think, at that point, you really have to ask yourself the question, Am I in the right business? Am I doing this the right way? It’s not even so much scale. I think it’s really that question: “I don’t need to know what you want. I don’t want to know what you tell me. I want to figure out who you are and what you might want.” And I realize that in saying that, I am condemning all brand-based advertising, all psychographic advertising. I think I’m actually kind of comfortable with that. I think what we got wrong is, we didn’t take advantage of the ways in which the internet makes it possible to state your intentions. That, to me, just feels like a much healthier way to do this than trying to intuit your intentions based on who we think you are.

Given that there’s no shortage of people looking to try out good ideas in Silicon Valley, you have to wonder, well why isn’t there some competitor to this, that has tried this kind of thing?
So you’re actually thinking about concentration even more than scale. I mean, I completely agree with you that one of the real things that all of us sort of cyberoptimists or cyberutopians got wrong was that we didn’t understand the concentration power of the internet. I think all of us thought, because it was so easy to start our little businesses in the 1990s, that this was going to be a sort of infinitely open space. And I think, for a lot of us watching, say, the rise and fall of Yahoo, even in the 1990s, and sort of looking at this and saying, “Wow, that company got big and then real irrelevant real fast,” kind of led to this thought that this was a really easy space to start something new and see what happened.

What I think a lot of us are realizing now is, actually, the internet naturally concentrates. And that the economies of scale are such that, if you’re any good at what you do, you’re going to generate a set of users. And you can ideally, at that point, sort of shut out a lot of other users. You can basically say, “These are going to be my users; you’re not going to get them onto your system. Or, if you do, I’m going to buy you and absorb you.”

I’ve been thinking a lot lately about how you’d launch rival social networks. My sense is that unless you can find something that lets you stay in touch with your friends that are already on Twitter and Facebook, you sort of have no prayer of launching anything new. And so my analogy for this is to say, imagine that you had a web browser that could only look at Facebook. It couldn’t look at any other website. Well, that’s what we had more or less in the days of AOL and CompuServe. They finally had to open up, otherwise they would die. But that was that walled-garden model.

In many ways, that’s now what we’re all dealing with on our phones, you know? My Facebook app won’t let me look at Twitter, and it won’t let me look at Mastodon, and it won’t let me look at anything else. I would really like to get back to the moment where I could have a single application that could let me look at existing social networks and new social networks. And that seems like the sort of direction we’d need to go in if we actually wanted more competition and more creativity than we’re getting right now.

I’m still curious about your distinction between Google and Facebook, given that there are people now who are even questioning the idea of advertising on the internet, period.
I think intention-based advertising’s a bit of a different thing. So I think, when I go and say, “I’m looking to adopt a dog,” and people show up with dog-adoption services and dog breeders. That is meeting my needs. It may not be meeting my needs well, and I may want to find ways of reviewing that, and sort of making sure that that’s happening sterilely. But, I don’t see that as being fundamentally counter to my interests. Whereas, essentially dropping into the middle of a conversation, or dropping into the middle of a search, and saying, “Hey, by the way, can I interest you in a new pair of pants?” That just seems significantly unhelpful. So I do see those things’ intention.

My thought on this is, I don’t want bad advertising. I want advertising to be on the side of the user. And for me, one of the few ways I can sort of see that happening is the sort of intention-based search advertising that I was talking about. But I definitely agree that this sort of notion of advertising that’s essentially competing for your attention makes things very, very challenging.

Look, there’s a bunch of things that are complicated and worth interrogating when you’re looking at something like Google Search. It’s worth asking, are the answers Google gives you fair? Are they influenced by commercial services? Are they the quality that they want them to be? And there’s also the question of whether the advertising is pulling away from the other results. At least the ad at that point is saying, “You asked about a roofer for your house, here is a roofer for your house who is willing to pay money for your attention.”

And that, for me, at least, is potentially helpful. I can also look at that and say, “You know, I’m not really interested in whoever’s paying the most money for me. I am potentially interested in whoever the Google algorithm thinks is best.” But it’s tricky. It’s really hard to know at that point. Should I value more that someone is trying to purchase my attention for that very specific intent?

What bothers me is this other piece of this. I am a 45-year-old white man who drives a pickup truck and lives in rural America. And therefore, there are certain demographic assumptions that can be made about me. It doesn’t seem especially fair, and it certainly doesn’t seem aligned with my interests when I start either getting light beer or Bushmaster AR-15 ads while I’m trying to have a conversation about something else.

If we want to get rid of advertising all across the board, I’m open to sort of brainstorming this model. It’s really hard for something you don’t use often. For the things that I use the most — Gmail, Twitter, Facebook — I would much prefer to see those things as subscriptions where I pay for them and I get certain protections from them.

It’s much harder to do that for something that is just emerging, and I just want to try it out. Or something that is a publication that I’m not necessarily going to read every day. Either we need some way to measure our online attention and give out some sort of monthly content fee, which is really hard to do without putting people under very heavy surveillance, or you probably need some sort of model that does free supportive content. And for me, that’s where things get very, very tricky.

I’m interested in hearing about the moments, the key inflection points you think, in the development of the internet and the platforms that put us in this position.
For me, it’s way earlier than Facebook. It’s probably earlier than Google.

How early?
Like, ’95.

What happened in 1995?
So, we started Tripod in ’94. And we thought of this as a content site that was mostly going to be about college students, and postcollege kids, and we were planning on doing it with sitewide sponsorship. You know, sold per year or per quarter for tens of thousands of dollars. It’s maybe early ’96 when we started taking our traffic statistics seriously and looking really seriously at how many page views we had, and then trying to sell ad impressions. So, for me, it’s really very, very early on with the development of CPM-based display advertising.

For me, those are sort of the turning points. And I like to tell the story about the pop-up ad, and the ways in which trying to figure out how to put a display ad on user-generated content may have been the really dark move here. And you could argue that everything that Facebook has done — essentially putting display ads in context of user-generated content — comes from that. But for me, it’s probably the first ad servers.

What were the servers?
We all wrote them, at first. We all had our own, initially.

Who’s we?, GeoCities — I mean, everybody who was putting display ads on web pages.

How would you draw a direct line from that moment, or when you had the first display ads on the internet, to where we are now?
So, Net Gravity releases an ad server in ’96 designed to deliver online advertising for Yahoo and Pathfinder. And so, in early ’96, all of us are making build decisions. We’re trying to decide, are we going to build our own one in-house, are we going to go out to someone like Net Gravity? And all these guys are essentially saying, “We want to be able to deliver the ads, and we want to measure how many we delivered for the clients.” The reason for this was that, as the very first on the web, we saw ourselves as magazines, but even better. You work in magazine journalism; you know that magazine ads can sell for $40, $50, $100 CPM, right? Full page of a magazine, you might be getting five or ten cents per impression for everyone that’s out there in the wild. On the web, we could say, “I got served once”; we could see it. But we could argue that I was even better than a magazine ad because it had a hyperlink, you could click on it, and you could go to the website.

The pitch was that this isn’t just an ad that you see in a magazine, where you have to go buy the thing advertised in the ad; this is an ad that takes you straight there.
You’ve got it. It’s even better than a magazine ad. So we were arguing early on that we deserved $100 CPM. And we got it, briefly. For a little while, in 1995, we were getting $100 CPMs. And when, as more and more and more people came out there, they started dropping really rapidly.

Keep in mind, you’ve got to go way back in time. This is before conversion. Conversion doesn’t exist at this point. People don’t know whether their ads get clicked or not. This is the very early days of this. And the very early days of this are,How many impressions did you generate?” Click-based pricing comes later. What really happens is just a surplus of content. Everybody’s out there. Everybody has tens of millions of page views. And people are able to go out there and say, “Everybody else is asking for $100; we’re going to ask for $50.” Then it goes down to $5. Then it goes down to $1. So everyone is trying to figure out how to take their existing inventory and make it more worthwhile.

And the first thing that everyone does is say, “We’ll give you behavioral targeting or contextual targeting.” Behavioral targeting is, “We will look around and see where you’re going on the site” — you spend a bunch of time in our car section, we’ll send you a car app. Contextual targeting is, “This is our cars section; we will sell you a car ad.” So, ad servers sort of get the ability to do that. And then, what starts happening after that is demographic targeting. “Oh, I know you, you’ve signed up for my site. I know that you’re an 18-year-old boy, and I’m going to target based on that.”

Somewhere in all of that, for me, is where it starts going off the rails. But if you want my moment, it’s that moment where I’m sitting and having a debate with my boss about whether we’re going to buy the Net Gravity ad server or write our own. Because it’s very, very clear that that’s the game that we have to get into.

How did Google’s arrival on the scene with something like AdWords change things? To my mind, it was sort of like Google stepping in, cracking its fingers, and saying, “Now listen up, this is how it’s done.”
First of all, it took a while for Google to become Google. The other thing that you have to remember is that when Google came on the scene, almost immediately everyone realized that it was a much better search engine than everything else out there. But it was unclear how much power and dominance that was going to give them.

I worked for Lycos, and companies like Excite and AltaVista were actually taken seriously. There were debates — were we all going to go the way that Google wanted us to go? Or were we going to do something different? In retrospect, we were idiots, obviously Google was going to eat everybody. But it was really hard to tell at that point. And AdWords doesn’t really come in and dominate the marketplace in that way until 2000, 2001.

But one of the things that everyone has to remember is that we already had a dot-com boom and crash by 2001. We’d been through one full business cycle on all of this, and a lot of these decisions that got made, got made in that first cycle before we even get to that.

So Google accrues power over the years following the crash, and advertising on the internet begins to take shape. AOL collapses in the mid-2000s and reemerges as an ad-tech company. You see emerging players in this. And then Facebook arrives in the thick of that. And they created something almost like their own internal AdWords. The afterburners didn’t really kick in until after the IPO, but it was growing really rapidly, and it built that initial business. Is that a fair way to look at it? How does Facebook fit into this?
Let me give you a slightly different narrative than the one that you just put out. It’s only a little different.

Google comes out with a very, very good search engine, and that search engine is powered by an advertising model that is just flat-out better than anything else out there. That is the combination of intention-based advertising, auction for search-engine keywords, and pay-per-click. You put those things together, and Google just owns the space. And Google does for the first time the thing that everybody starts doing on the internet, which is, they buy their only relevant competitor, Open Share.

And by the way, for any antitrust regulators sitting in the background, that’s the moment that someone should’ve done something. Because at that point, you suddenly have hyperconcentration about how we’re going to sell ads, and ad pricing; its decisions are all being made in the same company.

So Google, at that point, rules the open web. They rule the web where people care about URLs, they care about hyperlinks. They jump from one site to another, the Tim Berners-Lee view of the web.

What changes when Facebook comes on? Facebook basically says, “We want all of your time. We want all the time you spend online. We will try to get you from one page to another, one post to another, as quickly as possible. We will try to hold on to you as much as we possibly can. And, in fact, we so want you to be part of this that we’re going to open up our ecosystem and allow other people to sort of build games and build toys using our environment, just so we can keep you in here. And since we’re going to keep you in here, we also need to have our own app server to be able to deliver to you at the same time. So we’re going to do that. And since we’re doing this now, and now this is the modern world, we’re going to do that in a way that we have behavioral data, we have psychographic, we have demographic, we have contextual data; we can go for all of those things at the same time.”

What do you think that they’ve wrought? What do you think of the consequences and benefits of the deepening of those models?
I really feel like we’ve lost about ten years of innovation. I feel like this last decade has been pretty boring for the web. And I think because Google and Facebook have had such a stranglehold on the ways to make money on these services, the innovation and the interesting stuff have come from pretty far off the web. Things like Uber, things like Airbnb — trying to figure out, essentially, how to monetize in the physical world. Or they’ve happened outside of the U.S. ad market, in the Chinese market, which is wholly protected by the great firewall. And that’s where you’ve seen Weibo, WeChat, stuff like that, coming up in some interesting ways.

But basically, I think that the web’s been pretty boring the last ten years or so. And I think it’s been pretty boring because those platforms have been working to make it almost irrelevant. Facebook would, in many ways, prefer that we didn’t have the internet. They’d prefer that we had Facebook. Google seems fine as long as we are the way to get to the public web. But even with things like Google Plus, you can understand them recognizing Facebook as an existential threat in the long run. But for me, the disappointment is, it just hasn’t been that exciting a decade.

The innovation that has happened has happened at these very high levels of professionalism. You used to be able to do something novel online for a modest amount of money. And in many ways, it’s still pretty easy to program something new. But it’s very difficult to build an audience for it unless you are essentially willing to pay quite a bit of money to an existing partner. Or essentially try to buy that audience through Facebook or Google.

So that ability to sort of launch something and find that audience has become much, much harder. In the meantime, what’s happened is, these large platforms that have captured a lot of our attention have not been working all that hard on addressing some of their known bugs. And those include things like trolling and harassment. They include things like echo chambers and filter bubbles. And those are things that we’ve got to find some way to either force or help the platforms to address because they’re controlling so much of the attention, and it’s incredibly irresponsible that they haven’t found a way to sort of address these toxic environments.

Some people argue that solutions to this could come in the form of increased competition. That you could have other platforms or social networks rise up that prioritize these things. There’s also the view that there also needs to be some sort of public intervention, or government intervention, setting the standards and rules by which these platforms have to abide, in addition to, or instead of, either one, of breaking up the big platforms.
I think there are multiple responses here. I think the first is that regulating tech is hard. And there seems like — things that people could start to agree would be really helpful to do. You should be able to find out why you’re receiving a certain ad. I would argue that you should be able to find out why you’re receiving a certain piece of content. My lab released a tool called that lets you filter your own news. And one of the cool things about it is that on every post, you have the option to see why this was filtered in or out of your feed. And I think that’s potentially very, very helpful.

At the same time, I think that many of these problems are probably too hard. I have a hard time essentially saying, “Platforms have to stop bullying.” I’m not sure what that means; I’m not sure how that gets implemented. I agree, it’d be great if we had open competition. The trick at the moment is that you have all these people on Facebook saying, “I’ve had it; I’m going to Instagram.” Well, you know, Facebook owns Instagram. The competition’s been bought, and the same people are sort of making the rules.

The concrete proposals that I’ve put forward, which are pretty simple, come down to this: We need GDPR-type restrictions that allow us to own our own data, control where it goes, own our social graph, own the content we create; and then we need some sort of right to aggregate that makes it possible to be able to say, “I should be able to see multiple networks at the same time in the same tools,” so you have a competitive space for it.

I don’t know that that’s all that you need to do, but I would argue that those are things that you absolutely have to start doing, so that people can start running other groups under other different rule sets.

Ethan Zuckerman: ‘We’ve Lost 10 Years of Innovation’