The Near Memo, Episode 18

Is there a solution to fake reviews?; Google’s Grand Vision for search is nothing if not ambitious; How a local bookstore is fighting for his business and his community in his fight against Amazon.

The Near Memo, Episode 18

Part 1 Video: With fake reviews being so prevalent, how can incentives be aligned to achieve possible solutions?

Part 2 Video: Google’s future vision for search that uses AI to deliver expert answers beyond the Knowledge Graph

Part 3 Video: Local businesses must tell their stories & build community then explain: Amazon puts all at risk

Reference Articles:

  1. Fake patient reviews are making it increasingly hard to seek medical help on Google, Yelp and other directory sites
  2. Google: Rethinking Search: Making Experts out of Dilettantes
  3. Resisting Amazon: An Interview with Author and Bookstore Owner, Danny Caine


Greg: okay. Welcome episode 18. Boom. Mike David and Greg. Here we are get let's get right into it. So the three, three items, we had a lot of choices this week. Um, we're talking about three things. Uh, I'm going to start with the discussion of the piece that came out today. Which would be Friday if you're listening to this in the future, uh, in the wa the Washington post.

Uh, Laura Sydell wrote a piece on fake reviews and the pervasiveness of the problem. She's just focused on the healthcare field, but it's mirror. It's just a representative example of a much, much wider problem. We've talked a lot about fake reviews, review fraud, and, uh, Mike was interviewed extensively in the piece.

I was interviewed extensively, but did not appear in the piece for which I'm bitter. Uh, no, I'm kidding. I'm kidding. Uh, anyway. Um,

Mike: Yeah, I was wondering though, if he thought I should go back and ask her for a LinkedIn.

Greg:No, don't ask her for it. When people, people used to ask me for links, I would just smile, delete the email.

David: It made me smile.

Greg: so, so, so, so, so the problem is, is pretty pervasive. Um, some unpublished. Uber all research that we feel that, um, you know, two thirds of consumers now recognize fake reviews as a problem. And I think I forget the precise number, but it was, it was about half of that, saw it as a big problem.

And the way that people are dealing with that is by going to multiple review sites to try and validate the, the, the, uh, assessment of any given business, uh, or hotel or whatever it is. So it's, it's, it's gone from bright locals a couple of year ago, a couple of years ago, survey. Yes. I've seen a fake review to now.

People recognize it as a problem and all the articles like this, bring it into much more, uh, visible relief. And so what we want to talk about is really what are the solutions realistically? I mean, Google claims to be doing a lot. Amazon claims to be doing a lot. All these companies Trustpilot TripAdvisor, oh, we're, you know, have various algorithms and we remove X million fake reviews.

And Google was saying, it's shut down. Billions of. Fake listings last year, but, but yet the problem is real, really, really pervasive and, um, you know, exists in the millions and millions, uh, of, of, of deceptive reviews, many of which are generated by the business owners themselves through third parties or through friends and family, because there's so much at stake.

So in that article, they talked a lot about section two 30 and. You know whether or not reform of section two 30 would be impactful. And Mike, you had some, some strong ideas about that.

Mike: Uh, obviously I've been watching and criticizing Google in the review space for many years and, you know, they always make those statements that you said, "oh yeah, we didn't know this was such a big problem". Or, "oh yes. We found some fake reviews addressing it and we'll take them down and we'll address it. And we have a new algorithm I think is really good". And "it's going to fix the problem going forward". And six months later, you know, after the news is died down, It's always it's the same as it always was.

And it became clear to me that, and as I escalated this from local press to regional press, to national press. And the same thing happened over and over again. And I realized that Google has no incentive to work on this problem. None, zero, because there's no harm to them when their platform is abused. For this purpose, there is no financial repercussions.

There's no social repercussions. They've figured out a PR strategy that. They're like Teflon in the matter they, oh, it's not us. It's, you know, whatever. Yeah.

Greg: Well, "we're dealing with such scale that we're doing the best we can, or we're doing great".

Mike: "It's only one half of 1%" or whatever bullshit number they pull up.

And I realized in doing all that, that the only way to change this was from a societal point of view to look at every part of that chain. And either make parts of it illegal, which they're not, it's not illegal to have a fake business listing right now, for example, as far as I can tell, and if it is illegal, it might be illegal.


Greg: it's it's consumer op repair,

David: uh, pilots.

Mike: It is going strong. Yes. I once created a fake Google listing called illusory laptop repairing and still gets calls, uh, which I forward to my brother. But. It's to be in fact to repair laptops, he does on the land. Then it's mostly,

Greg: it's not an illusion, then it's a real list,

Mike: but there's problems that every part in the chain, right where the business is incentivized to buy fake reviews, where the marketing agency working for them is incentivized to buy fake reviews.

There's companies that have been set up, I think, largely criminal, but then that sell fake reviews. There's no penalties for them. They're national and there's no penalties at Google. Now I've been taking the approach from Google's point of view. And to me. The blanket carp launch that they're given is through section two 30, which absolves them of any responsibility for use of their platform.

It's a content moderation issue. Right? And my contention would be that if they are held to a slightly higher, after the fact standard, where they are expected to respond quickly to accusations of fake reviews, where they're expected to analyze them quickly and respond honestly to people who report them, then that would drive their costs up after the fact without changing their.

Protection from liability, but forced them to then write better algorithms so that they aren't dealing with so many calls about,

Greg: well, th th the, the problem. So this, so the FTC put out a little guide, which, which I think you originally pointed out to me, um, uh, about how, uh, for consumers about how to evaluate online reviews.

And it was like the weakest thing ever. Not only is no one gonna see it. It's just, it's just kind of, you know, no, duh not helpful. Uh, information. And so, and the Supreme court sort of gutted the FTCs power to impose monetary damages or penalties, uh, on companies, uh, as, as kind of regulatory overreach. So, so now they only have kind of injunctive capabilities, meaning they could stop the behavior going forward, but they can't do anything to raise the cost as you're suggesting.

So. You know, there's no mechanism here unless Congress does something and

Mike: amending to 30 is my opinion is that I have to

Greg: give the, the, the F the power of, uh, financial penalties back to the FTC.

Mike: Well, that, and, and put more post after the fact. Moderation responsibility that you know, that things have to be responded to in a reasonable timeframe with transparency.

And they have to like right now, if, if somebody puts a fake review up that's libelous, Google has no obligation under current law. Even take it down. I mean, how well

Greg: the, so the states, like as, as, as what as happened with privacy, the states could step in and do something under consumer fraud. Yes, or, or whatever.

Mike: And the FTC has recently announced that they're partnering with states on that issue, but that's just, just a very onerous and difficult. And again, the FTC job is mostly the civil don't do it again. And if there is a fine, it's minor, fine. If they could do a final once again, but it's like, it's not enough.

So there needs to be a policy change that, that starts at the top with Google and works its way all the way down that makes it explicitly illegal. To do these things.

Greg: Well, as you say, Google is not going to do this on its own. So, so let me, let me just sort of shift slightly, but by the way, Yelp has less review fraud than Google and Yelp could take that information and market aggressively to consumers, which might put some pressure on Google too.

You know, you can't trust Google. You can't trust her to read on Google. Because, you know, there's all this fraud that, that might be three. What you

David: read on Yelp either because they filter out so many

Greg: legitimate, it's a different discussion, but there is a lower statistical incidence of. Of fake reviews on Yelp, but that's a slightly different discussion.

So let me segue into, into something that's going to get into your, your, your discussion, Mike, which is, um, which is the idea of trust and credibility in, in online content. So reviews, if reviews are perceived to be less and less credible, and they certainly are in my world. Um, you know, I'm increasingly turning to expert reviews and turning away from consumer reviews around products.

Um, and, and I find myself more and more dissatisfied with search results in a lot of situations, like just third parties promoting themselves and crap content. Um, you know, You, you, you ha uh, found an item that is an interesting alternative to the existing sort of treatment of search results. Why don't you talk about that?

Mike: So Google, as we know it, I O they announced. Several major AI initiatives, mam, which is a language analysis initiative to replace Burt, which is already very sophisticated, but it looks at words in context. So, and understands more about the sentence. They talked about Lambda D I think it's called, which is a AI to actually complete sentences based on starting words and, and.

Okay. Uh, categorical sort of definition. So kind of like GPT three AI, you can actually, it actually writes documents for you. That's their Lambda product. And they talked about their image ability to understand the content and images, which I know from my personal experience has been growing exponentially over the last several years.

And so there's those three sort of foundational things. And then I, I, I found this paper called rethinking search making experts on agility, gelatin that was written by four Googlers, uh, for the AR XIV conference. And they discuss how there's a need that their goals require both a rethinking of how information is organized top to bottom using these three tools, getting rid of.

The knowledge graph, which they, they not getting rid of it, but, but they perceive the knowledge graph as being very limited, to only expose limited number of facts and they call that the dilutant. And so their goal is to create a new indexing system that removes this whole idea of, uh, of traditional retrieval sort of, uh, storage, retrieval, ranking, picking out the right piece, dance, a question into a, a model based.

Sort of storage using these three elements to literally, I mean, the best way I can summarize is you ask your question, they, or drop in an image or drop in an image and a question they can then write a detailed Wikipedia, like response with references so they can show you where all their knowledge came from, why they're saying it and what, and, uh, basically right.

It's a full exposition as an answer, include images and compare them. So they gave the example at IO that you're used certain shoes to climb Mount hood. Would they be good enough to climb Mount Fuji? Well, they have to understand what Mount hood is Mount Fuji is, and then extrapolate the type of hiking conditions based on things they've read in any language and then write a new answer to it.

It's pretty ambitious. And as you pointed out could deal with local search.

Greg: Well, it could, it could deal with a lot of things, I mean, remains to be seen, but it's, it's sort of, uh, you know, they're doing a bit of that with the snippets now where you, you get a highlight on the page where the snippet is extracted from.

So it's interesting. Yes, Dylan, Dylan Thomas is not a positive word by the way. I know

Mike: my, in my mind, there's papers either. I can't imagine the people on the line, local liking being called deal with that.

Greg: Well, it's, it's, it's, it's, it's not, uh, it doesn't inspire confidence, but, but as you pointed out, I think, uh, before we started today that there was, uh, we were talking about duplex and, um, you know, one of the other things that we're not really talking about is Google's test in Italy to use.

Uh, Google the Google assistant and Google duplex to solicit business information as a way to set up a GMB profiles, as opposed to the current way, which is manually claim and then populate your profile. And, um, you know, you were talking about the time lag between the introduction of that technology, you know, several years ago and the 2018 IO.

Yeah. So the implementation, so three, three years. So this, this, um, model that you're describing could be many years away if it ever appears.

Mike: Correct, but it, it reflects several things and we, we, one, we can see the building blocks happening and they're using sort of the building blocks. Like we're seeing in the image analysis piece Mo dramatically improved understanding I've recently gotten access to their API.

And so just fascinating what they see in images, all the entities, their ability to label it, their ability to extract what are called semantic triplets from an image. So, uh, Uh, kind of yes, exactly. Precisely

Greg: those, those are the primary categories into which everything is organized,

Mike: but so they can put these, they can take an image and literally extract a logic that they can then query about the image.

And that's fascinating. So we're seeing parts of this and we know their ambition is massive.

Greg: Well, imagine, imagine though think about this in the context of antitrust let's we don't know what this ultimately will look like, but what if the whole world and all the content of the documents and websites and underlying, uh, data is then just becomes a source material for Google's own proprietary presentation of that content.

I mean, that's even more agregious. If I can use that word, then what we're dealing with today, the, the, the zero click stuff. Yeah, lots of times, I'd

David: say a couple of, I mean, Google has been moving in this direction with a zero click results and this is just sort of potentially a smarter way to compile those results.

Um, so, and it fits in with their mission to organize and make accessible all the world's information. So, I mean, to me, this is what they've been trying to do since they started. And they've finally figured out a technical way that they might be able to achieve it. Um, so I don't see like, cause this doesn't come as a surprise to me.

Uh, necessarily. Um, but I do think there's a certain point at which there, um,

Greg: if they push it too far,

David: publishers are going, are going to opt out of Google because there's literally no, there's no value if you're not getting any click-throughs from Google. Uh, to, to make your content accessible to them. So I think it could have a, you know, if they push it too far, it could have a significant downside not to say, I don't think it should be regulated out of existence, uh, if we get to that point.

But, um, I do think that there's, there might be some, uh, unintended consequences from,

Greg: from Google. It's also interesting to think about in terms of like non-screen, uh, experiences, like, you know, audio or smart glasses or. Whatever else may be forthcoming the car, you know, that, that this, this mechanism, uh, might be well-suited to those other kinds of the, the smart speaker, you know, those other kinds of, uh, treatments or, or formats as well.

Um, there's no easy segue into the, into the third item, except, um, uh, to say the Google is a yes. Yes. Google is a giant company and so is Amazon, Amazon. Yeah. And so, yeah. So go ahead.

David: I just wanted to highlight Miriam Ellis' excellent article that she wrote for us this week, which was really a, sort of a conversation that she had, uh, with the owner of Raven bookstore.

Uh, Danny came in Lawrence, Kansas, and, um, just a really interesting piece about how he has sort of oriented his entire business around. Exposing the societal ills that Amazon imposes on all of us. Um, and I thought I found it interesting for a couple of reasons. The first was the anecdote, uh, that he shared with which sort of, I don't know, pushed him over the edge, uh, to some degree of doing this with his business, where a customer came in and said, oh, I can get this, this book that you're selling for 26 99.

I can get it for $15 on Amazon. Sort of asking them to sort of match the price and he's like, that's below our costs. There's no way I can do that. And so the first thing I thought was I thought it was, you know, our, I don't know, to the extent that it's possible. I know that apple is very, uh, restrictive right.

About what you can sell an apple product for, because it's the same price on Amazon best buy and the apple store. Um, and so I'm wondering specifically in the case of books, why this hasn't, um, why publishers haven't sort of banded together. To, um, enforce the same price, uh, at a minimum across the Amazon and have that are local outlets.

Um, there's still the issue of, you know, same day delivery and all these things that Amazon can do that, that Raven bookstore, uh, probably will, will

Greg: never be able to, and also eBooks also

David: any books as well. Um, but you could, again, you could enforce the same price for an book as you do for a print print book.

Mike: So, I don't know if you remember that apple tried to help the publishers established baseline pricing and they were sued by Amazon for antitrust and Amazon won that case. So that's one issue.

David: The publishers should be able to decide what they, they should be able to fix their prices on all platforms.

This the same way apple does or Sony or any other

Greg: manufacturer. You, you, if you, if they all, if they all. Agree in principle to that policy that isn't necessarily, uh, you know, whatever, it would be a price, price fixing it's, it's just the, you know, they're, they're just saying we're, we're not gonna allow our, our distributors, right?

Yeah. It's not a collective thing.

David: So that, that struck me as one way that, that, uh, industries at least could to band together to support. Local businesses. And then the second thing is, uh, Miriam highlighted a number of various things, civic tactics that Danny is doing, um, to foster a sense of community at his local business.

Um, and I think leverage some of this sort of latent consumer interest in supporting local businesses. I mean, he makes it really tangible, right? He's hosting local events. He's partnering with other local institutions, uh, in Lawrence, in the surrounding area, um, to host festivals, um, Miriam pilot, Mariam highlighted his, uh, store kitties.

Um, that's his sort of community reading area. Um, so there's just a number of experiential things that he is doing that actually give him a competitive advantage over Amazon. And I think. That's really the way to build consumer loyalty at the local level. Um, to the extent that you as a local, I mean, not everyone's going to be able to do this right.

If you're a plumber, I suppose Amazon doesn't yet really provide plumbing services, but, um, there may not be this, this, this notion of building community, you know, you know, is sort of not, uh, not applicable to the same extent across multiple industries, but I just thought that Danny has done some really interesting things, um, to actually play up this competitive advantage over Amazon, that they can't possibly replicate.

Greg: I agree. It was a really, um, interesting and, and strong piece that she wrote and interesting sort of to, to, to, to, to, to see those tactical things. You know, it's, it's in a way it's kind of a, the same thing that a lot of large retailers are dealing with, trying to build better in store experiences as a way to entice customers, uh, to, to continue to come in.

I have some skepticism. I mean, simply I'm very sympathetic to everything in that article. I have some skepticism about consumer willingness to go any extra distance to support a local businesses. Some people, some, you know, some percentage will, um, but a lot of people will take the path of least resistance, which for them is, uh, Amazon.

And there was that piece about Amazon prime, the, the, the Matt Stoller piece in sub stack that was talking about how prime. Um, actually raises prices for consumers because Amazon extracts, you know, they, they do all, all kinds of things in the back end to, to compensate for that. And, but prime is a, is a powerful, it's probably the most successful loyalty program ever.

And it's, it's driving enormous value for Amazon and there's a lot of inertia around it for consumers and I'm guilty of it too. I mean, I'm getting packages every day from Amazon, even though philosophically I'm in the same boat. As a lot of people, I want to support local businesses.

Mike: I mean, to some extent there's a businesses.

I think have evolved to understand that certain products can be valuable. It books is one of those. You need curation, you need display, you need community. And if you enhance and pants and embrace those things, bookstores are now on the increase in have been for about three years. Uh, once B Dalton went down and, uh, borders went down and left Barnes and noble is the only.

Bookstore standing and they had problems. And that interesting article in the financial times where they are using many of these same tactics to recreate the ambiance of a bookstore by relying on the employees in the bookstore to add that value. Uh, and, and to do that at scale. So to do it, they have to trust that employees love books, recommend good books, and they have some interesting techniques about how to display them, but their, their books lend themselves to that more than say, uh, laundry detergent or at some other commodity.

Right. And to some extent in the true commodity area, Amazon is providing a service as is Walmart getting a commodity to the end users cheaply as possible. It's just the commodity, right? So they're not evil in doing that. The problem is that they have such ambitions and they have to grow into all these other areas that they become detrimental to society at large.

Um, so.

David: Yes, as we always try to do here on the near memo, uh,

Mike: which I continually seem to submarine, but I, I,

David: Danny has holding his own and, uh, I have actually never been to Lawrence, Kansas despite driving across  basically every summer as a kid. Um, so I certainly look forward to visiting sounds like a great town where he's partnering with a lot of other similarly minded, local businesses.

And there's a real sense of. Of camaraderie. And I think that that is, uh, a model that can and should be replicated by local businesses

Mike: across the country. So kudos to Miriam for writing a great article for us.

Greg: This is something we want to do more of. We want to showcase these local businesses and their experiences and learn what's really going on in their world and on the ground and, and, uh, tactically.

So w th look forward to more, more of these pieces. All right. And with that goodbye for this week, join us next week. Be sure to subscribe to a near media and thanks again for listening.

Mike: Thank you. Have a great weekend. Bye bye.