Saoud Khalifah: CEO, CTO & Founder of Fakespot – Ep. 139

[From Teenage Hacker To Fake-Review Fighter]
*3:49 – “I was a hacker in my teenage years now so that gave me a good look into how these dark web, black hat, grey hat people would actually think. I was a white hat hacker. So basically I would find bugs and submit them to bug bounty programs publicly. For example, there was a big one with Microsoft. I found a vulnerability in Microsoft PowerPoint; it basically worked this way: I send you a PPT file, a PowerPoint file, and I can take over your computer. But I was a white hat so I reported it to Microsoft through one of the vulnerability bug bounty programs. And you know, these black hat hackers will never do that. They will try to game every system around the Internet.”

[Getting Fake Amazon Reviews Is Easier Than People Realise]
*5:58 – (Ross): “So if you have no following like me and you want to release a book, I’ve got a book that I want to put out. And that book is called Why Doughnuts are the One True Religion. I have zero audience and I want to have an Amazon bestseller. What do I need to do to game the system?”
(Saoud): “Yes. So usually what we see is people give it out to friends and family. So they get that first tranche of reviews and you will see that they’re very positive about the author, more so than a normal consumer would be, right? And then they go out to the services and you can even see them on Fiverr. And you’re one Google search away from finding these services. So you want to Google ‘buy fake Amazon reviews’ or ‘buy five star reviews’ or something like that, you check out on their website and you start getting those reviews on your book. So that’s usually the play we see here. The other kind of more, I would say intrusive one is using influencers to pump your product. You pay them and they pump that product on their channel on TikTok or YouTube.”

[Is All Future Discourse Going To Be Bots Chatting With Other Bots?]
*10:39 – (Ross) “We know that on platforms like Twitter, anywhere, such a large percentage of all of the Internet’s traffic and content is produced by bots. And you have this sense of, in the future, the majority of all discourse is just going to be bots chatting with other bots. You’ve got harmful bots communicating with good bots, but still there’s not a human involved and we know how that can affect obviously the political situation, right? If I see that there’s all of this support for a certain political candidate, even if all of that is generated by bots, it still gives the impression that there are two sides to an issue when in reality there may not be, or there are two opinions, when in reality there might not really be. But we can’t really tell. And I made a video a little while ago about Google and LaMDA and Blake Lemoine, and Google achieving sentience in his opinion. You might have read about that when it was all in the news. And the point that I was trying to make there was that it doesn’t matter whether Google is sentient or not or whether A.I. has reached that level of intelligence, what matters is that it can fool somebody. What matters is that Blake Lemoine was fooled enough to write about this. That’s what matters. The significance isn’t whether it’s actually conscious.”

13:50 – (Ross) “I think the big shock in certain presidencies, we don’t have to go deep in the details here, but I think everybody will understand what we’re talking about. But the big shock was to have this sense shattered for most people who haven’t been thinking about this as long as you have, to have this sense of ‘I think I know what the average person thinks and what they feel,’ completely shattered. Where I used to think, ‘okay, there’s a set of common values that we can all agree on. We may disagree on certain points, but there’s a set of common values that we can agree on, regardless of our political affiliation.’ And we sort of saw that, or we had the perception that, oh, wait, there is this huge tidal wave of dissenting opinions or people who don’t agree and you don’t know whether those are real people or whether that is just bots. Is there incredible support for this candidate, really, or is that just an army of bots that are going into every post that mentions certain keywords and flooding it with this pro or con viewpoint, as the case may be? So I think our concept of what is actually happening was completely destroyed. At least if you’re paying attention.”

[FIRST CLIP] – this clip will need to be edited as I’ve omitted some text (see ellipses) – the green highlighted text is a shorter version for Reels but the longer quote is for a grid post
The Battle Against Fake Reviews
*22:36 – “I ordered this supplement from Amazon in my final year in college and it had hundreds of five star reviews. The yellow stars were very enticing and glowing. So I was like, okay, cool. I trust this rating. I have trusted it to this point. And this was about Christmas 2014. I got the product, I did the one-click checkout and it looked like someone made it in a garage. The tape was falling off the packaging, like the glue was like completely messed up. And then the pills had chip dust from a woodworking shop inside the pills. So I was like, okay, there’s no way this is getting five star reviews from anyone. Like no chance, and at the volume they had. So I went back and I looked at the reviews and I realized they were written with Markov chain generators. At that time that was the way – so believe it or not, the new neural networks are using Markov chains underneath the hood, but not the way that was the primary, where in the past where you give it a theme like a text and it just spins the text, like replaces synonyms, it changes the sentence structure, makes it look like it’s a different author…And that’s actually when the idea for Fakespot started. I was like, okay, what I just did with my mind, looking at reviews/reviewer profiles, let me launch an A.I. model or a bunch of A.I. models and look at these different attributes. Collect as much data as possible so that we’re very accurate with the representation of the trustworthiness and the veracity of this textual content in that case…And I got the domain and put out the website. It was very simple. It was like a URL bar, like a search, kind of like Google. You paste in the product page and you click ‘analyze,’”

[Plato’s Allegory of the Cave]
*28:23 – (Ross): “I still think that Plato’s Allegory of the Cave is the most prescient, you used the word prescient, the most prescient text ever. I still think that’s the greatest model for what we’re witnessing right now of anything–”
(Saoud): “–It still applies today, like people are stuck in that cave. Using Twitter, using Facebook, using Instagram, using TikTok. TikTok is just promoting the cave. Hide in the cave.”
(Ross): “Yeah, it’s only the cave. And we worship the people who are the most successful in the cave of TikTok.”
(Saoud): “We worship the shadows, we worship the shadows that are fake. And these platforms are about fakeness.”

32:10 – “If you remember I just mentioned earlier on how Fakespot started – 2014/2015, something happened on Amazon. Amazon needed to scale. How did they scale? They realized they couldn’t be selling everything from their warehouse. They get the product from Clorox or Levi jeans and they sell it. To scale it they needed to introduce all these third-party sellers, millions of sellers across the world, and be the platform where people come and buy that stuff and they can maybe take care of the logistics for fulfillment or they connect you directly to that third party. So but they need that commission. That’s how they scale this to a trillion-dollar company. That’s exactly what they did. But by doing that, they introduced, they brought in, a lot of fraud. These sellers are not ethical and they will engage in any means necessary to make as much money as possible.”

[You Have To Do Things That Matter]
*36:58 – “I think going back to the focus of your podcast, you have to do things that matter to you personally. Like, you cannot be doing things that don’t matter to you, right? I think you are doing this podcast, and you can correct me if I’m wrong, because you want to disseminate information and knowledge to your viewers and listeners. And that’s a critical I think, that’s something very critical to humanity. And the same way, with Fakespot, our team, us, we want to disseminate trust. We want to tell you what you can trust and what you can’t trust. And you have to be passionate about it.”

51:04 – [On the next 5 years] “Well we want to continue building an impactful platform that, you know, we’re solving the problem for consumers, for shoppers. And we may look at different content types across the Internet. We kind of hinted at it throughout our interview here. And that’s something that we would be very passionate about, bringing more value to people’s lives through technology in an open and transparent way. And our mission is to always bring trust and transparency to the Internet. It’s always been that from day one. So continue building on that mission and continue impacting people’s lives in a very positive manner and show everyone that there can be a company built away from the way that we’ve seen with the big tech in a positive way.”