Partner, Sue, or Block?
These tend to be three options available to publishers when it comes to dealing with artificial intelligence companies by cutting content to train large language models.
Nextdoor, a Hyperlocal Social Network, chooses by highlighting blocking paths.
“If you spread content widely, one of the larger operating companies like ChatGPT or Microsoft doesn't need to take all the content and bring the consumers to you.
Tria said his stance is “ideology,” and that comes after he learns harsh ways of thinking about the dangers of being too dependent on big tech platforms.
During Dot-Com Boom, Tolia was an executive at Shopping.com. This is a site that helped shoppers track the price decline of the products they are about to buy. Tria said about 80% of the traffic came from Google search. The company was published in 2004, but its stock price quickly crashed, following changes to Google's algorithm that destroyed traffic. (Sold on eBay in 2005)
“When I was starting Nextdoor with my co-founder, we said it's much harder to grow without Google's profits, but we don't sleep well at night so we want to do everything so we don't depend on them,” Tria said.
“We never allowed to scrape away any content. None of the search engines have raw,” he added.
That's an unusual approach. Many website owners focus on lasers on how their brands appear on their search pages and how to optimize Google's crawler sites.
That approach is also why we don't see Nextdoor reducing licensing agreements like Reddit's partnership with Google. The search giant says the deal, reported by Reuters, is worth around $60 million a year to Reddit, not only trains AI models, but also surfaces more information derived from Reddit in search results.
Tria questioned the long-term viability of these types of transactions, even in the short term. (Disclosure: Openai has a license partnership with BI parent company Axel Springer.)
Tolia said that if users decide to go to Google to retrieve data derived from Reddit, Google can reduce the cost of their licensing agreement over time.
“These consumers can't even know that it's reddit information,” Tria said.
Google previously stated that AI overviews send you a “high quality” visit to some websites when a user clicks on the source citation link and then clicks on more information. However, individual analyses from independent researchers suggest that AI overviews may undermine click-through rates.
Nextdoor is preparing for the next era
Tria's stance on licensing does not mean Nextdoor has no AI ambitions.
Tolia wants to have AI agents in every neighborhood on Nextdoor, so consumers can ask about chatbots. NextDoor's agents will then troll users' posts for 14 years and recommendations to express their answers.
“It's just as easy as ChatGpt, it's a coincidence for us at Nextdoor,” Tolia said.
Nextdoor's trip to Cannes Lions comes as you prepare to start a new redesign with the aim of creating more structured feeds and more timely notifications about events happening in your neighborhood, and stacking information from sources such as city councils and local news outlets.
The company's goal is to become a daily habit. Nextdoor said in its revenue call last month that it had 100 million verified users, but only about a quarter of them are active on the platform.
Nextdoor is also sharpening its pitch towards advertisers. Once the new redesign launches, we plan to deploy more ad formats and other advertising services later in the year. Tolia said Nextdoor emphasizes its hyperlocal roots and the authentic conversation that occurs on a platform that is rooted in real events rather than online.
“I think marketers want authenticity, so if you think about the ideas of validated neighbours in your real area, then in the private ecosystem – that's very different from places like Instagram, for example, “a lot of content from influencers,” Tria said.
Axel Springer, the parent company of Insider Inc., is an investor in Nextdoor.
