Tech Brew Ride Home - Wed. 05/29 – Did The Secret Sauce Of Google Search Just Leak?

Episode Date: May 29, 2024

A big leak at Google might be the tech equivalent of the secret formula for Coca-Cola being revealed for the first time. The ex-OpenAI board members are starting to explain why they tried to fire Sam Altman. Did the Biden administration pass on TikTok’s concessions to avoid a ban? And remember delivery apps? How they doin’ these days? Sponsors: Lumen.me code: RIDE Links: An Anonymous Source Shared Thousands of Leaked Google Search API Documents with Me; Everyone in SEO Should See Them (SparkToro.com) Ex-OpenAI Director Says Board Learned of ChatGPT Launch on Twitter (Bloomberg) Anthropic hires former OpenAI safety lead to head up new team (TechCrunch) YouTube’s free games catalog ‘Playables’ rolls out to all users (TechCrunch) How the U.S. ignored a chance to make TikTok safer (Washington Post) Food delivery apps rack up $20bn in losses in fierce battle for diners (Financial Times) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the Texas. me right home for Wednesday, May 29th, 2024. I'm Brian McCullough today. A big leak at Google might be the tech equivalent of the secret formula for Coca-Cola being revealed for the first time. The ex-open AI

board members are starting to explain why they tried to fire Sam Altman. Did the Biden administration pass on TikTok's concessions to avoid a ban and remember delivery apps how they do in these days? Here's what you miss today in the world of tech. Weird one to start the day with. A leak has reportedly revealed pages of internal API documentation for Google Search, some of which the details appear to contradict Google's public statements about how search actually works. In other words, the black box that is the Google Search algorithm's secret sauce might have been revealed. This is from noted SEO guru Ran Fishkin, who got the documentation from a leaker, quote,

The leak appears to come from GitHub, and the most credible explanation for its exposure matches what my anonymous source told me on our call. These documents were inadvertently and briefly made public. Many links in the documentation point to private GitHub repositories and internal pages on Google's corporate site that require specific Google credentialed logins. During this probably accidental public period, between March and May of 2024, the API documentation was spread to Hexdocs, which indexes public GitHub repos, and found slash circulated by other sources. I'm certain that others have a copy, though it's odd that I could find no public discourse until now. According to my ex-Gougler sources, documentation like this

exists on almost every Google team explaining various API attributes and modules to help familiarize those working on a project with the data elements available. This leak matches others in public GitHub repositories and on Google's cloud API documentation, using the same notation style formatting, and even process slash module slash feature names and references. If that all sounds like a technical mouthful, think of this as instructions for members of Google's search engine team. It's like an inventory of books in a library, a card catalog of sorts, telling those employees who need to know what's available and how they can get it. But whereas libraries are public, Google Search is one of the most secretive, closely guarded black boxes in the world. In the last quarter

century, no leak of this magnitude or detail has ever been reported from Google's search division, end quote. What has been revealed here? I mean, you might need to be fairly SEO conversant to go deep into the weeds on this. If you are such a person, please click through on the links to read everything. But for the rest of us, I'll share instead maybe the most SEO-centric headlines that normal people would understand. For example, why do you think they really created the Chrome web browser inside of Google? It's the reason that we all thought from the beginning, apparently, quoting from Fiskin again, but jumping around a bit. Many of the leakers claims directly contradict public statements made by Googlers over the years. In particular, the company's repeated denial that click-centric user signals are employed,

denial that subdomains are considered separately in rankings, denials of a sandbox for newer websites, denials that a domain's age is collected or considered, and more. My anonymous source claimed that way back in 2005, Google wanted the full click stream of billions of internet users, and with the Chrome web browser, they've now got it. The API documentation suggests Google calculates several types of metrics that can be called using Chrome views related to both individual pages and entire domains. This document describing the features around how Google creates site links is particularly interesting. It showcases a call named top URL, which is, quote, a list of top URLs with the highest two-level score,

i.e. Chrome trans clicks. My read is that Google likely uses the number of clicks on pages in Chrome browsers and uses that to determine the most popular slash important URLs on a site, which go into the calculation of which to include in the site links feature. Also, Google regularly employs white lists around sensitive topics like politics and during COVID times as well, or other health-related emergencies. In other words, they put their thumbs on the scales sometimes to make sure certain searches surface certain things at certain times. Google has numerous ways to identify entities, sort, rank, filter, and employ them. entities include brands, brand names, their official websites, associated social accounts,

etc. And as we've seen in our ClickStream research with Dados, they've been on an inexorable path toward exclusively ranking and sending traffic to big, powerful brands that dominate the web over small independent sites and businesses. If there was one universal piece of advice I had for marketers seeking to broadly improve their organic search rankings and traffic, it would be build a notable, popular, well-recognized brand in your space outside of Google search. For most small and medium businesses and newer creators slash publishers, SEO is likely to show poor returns until you've established credibility, navigational demand, and a strong reputation among a sizable audience. SEO is a big brand, popular domains game. This is almost certainly

true for other creators, publishers, and small and medium businesses. The content you create is unlikely to perform well in Google if competition from big popular websites with well-known brands exists. Google no longer rewards scrappy, clever SEO savvy operators who know all the right tricks. They reward established brands, search measurable forms of popularity, and establish domains that searchers already know and click. From 1998 to roughly 2018 or so, one could reasonably start a powerful marketing flywheel with SEO for Google. In 2024, I don't think that's realistic, at least not on the English language web in competitive sectors, end quote. In a podcast interview with my former colleagues over at TEDX OpenAI board member Helen Toner

revealed that the board learned of chat GPT's launch on Twitter. She also criticized Sam Altman's leadership on safety and more, quoting Bloomberg. In a podcast called The Ted AI show, Toner gave her fullest account to date of the events that prompted her and other board members to fire Sam Altman in November of last year. In the days that followed Chief Executive Officer Sam Altman's sudden ouster employees threatened to quit, Altman was reinstated, and Toner and other directors left the board. When ChatGPT came out in November 2022, the board was not informed in advance about that, Toner said on the podcast. We learned about ChatGPT on Twitter. The company's launch of ChatGPT was relatively quiet. Open AI simply called the chatbot an artificial intelligence model that, quote, interacts in a

conversational way. But over the following days, In recent weeks, ChatGPT's ability to generate human-sounding text made it a massive hit and helped pave the way for the current boom in AI. OpenAI did not immediately provide a comment in a statement provided to the TED podcast. OpenAI's current board chief, Brett Taylor, said, quote, We are disappointed that Ms. Toner continues to revisit these issues. He also said that an independent review of Altman's firing, quote, concluded that the prior board's decision was not based on concerns regarding product safety or security, the pace of development, open AI's finances, or its statements to investors, customers, or business. partners, end quote. In the podcast, Toner also said that Altman didn't disclose his involvement with

OpenAI's startup fund, and she criticized his leadership on safety. Quote, on multiple occasions, he gave us inaccurate information about the formal safety process that the company did have in place, she said, meaning that it was basically impossible for the board to know how well those safety processes were working or what might need to change, end quote. Toner said that after years of such events, quote, all four of us came to the conclusion that we just couldn't believe things that Sam was telling us, end quote. In an article in The Economist over the weekend, Toner and Tasha McCulley, another former director, expounded on their thinking, saying that Open AI was not positioned to regulate itself and that governments should intervene

to ensure that powerful AI is developed safely, end quote. Meanwhile, Anthropic has hired former Open AI safety lead Jan Leakey to head up a new superalignment team there. A source says Leakey will report to Chief Science Officer Jared Kaplan, quoting TechCrunch. Jan Leakey, a leading AI researcher who earlier this month resigned from OpenAI before publicly criticizing the company's approach to AI safety has joined OpenAI rival Anthropic to lead a new superalignment team. In a post on Axleaky said that his team at Anthropic will focus on various aspects of AI safety and security, specifically scalable oversight, weak to strong generalization and automated alignment research. A source familiar with the matter tells TechRunsh,

that Leake will report directly to Jared Kaplan, Anthropics' chief science officer, and that Anthropic researchers currently working on scalable oversight techniques to control large-scale AI's behavior and predictable and desirable ways, will move to report to Leakey as Leakey's team spins up. In many ways, Leakey's team sounds similar in mission to Open AIs recently dissolved superalignment team, the superalignment team, which Leakey co-led, had the ambitious goal of solving the core technical challenges of controlling super-intelligent AI in the next four years, but often found itself hamstrung by Open AIs leadership. Anthropic has often attempted to position itself as more safety-focused than OpenAI.

Anthropic CEO Dario Amadai was once the VP of Research at OpenAI and reportedly split with Open AI after a disagreement over the company's direction, namely OpenAI's growing commercial focus. Amo-Dai brought with him a number of ex-open AI employees to launch Anthropic, including Open AI's former policy lead, Jack Clark, end quote. YouTube has rolled out Playables, a collection of over 75, free games available on mobile and desktop to all users, quoting TechCrunch. YouTube's, quote, App Store for Games is rolling out more broadly. The company announced on Tuesday that its collection of lightweight free games dubbed Playables will soon start to appear in the YouTube app for all users

in addition to the YouTube homepage. Previously, the games had been made available to select users for testing before arriving for YouTube premium subscribers last November. Because they don't monetize as paid downloads or via in-app purchases, YouTube's playables don't directly challenge the App Store model or break Apple's rules. However, they do compete with the App Store's free games, which are often downloaded by casual gamers and generate revenue via ads. As Google is shifting its focus to integrate AI, there are questions about the technology's impact on its ads business cash cow, driven by the sponsored links that appear above search results. Free games on YouTube, in theory, could become another place to serve ads further down the line. For now, though,

Google hasn't signaled its intention to monetize its playables. The games could, however, provide a distraction for YouTube users in between their browsing and viewing sessions and help keep them engaged with YouTube's app. The lineup of Playables includes a handful of more popular titles like Angry Birds Showdown, Words of Wonder, Cut the Rope, Tomb of the Mask, and Trivia Crack, among others. It also pulls in titles like StackBounce, a game that Google offered on its HTML minigames service, and Game Snacks, which had been developed out of its internal incubator, Area 120. With Game Snacks, the goal was to bring gaming to users in emerging markets, a place where Android dominates. Today, there are currently over 75 minigames in the Playables catalog, Google says.

Players who use the feature will be able to save their game progress and track their all-time best scores. Not everyone will see the Playables right away, but the feature should complete its rollout in the coming weeks, end quote. Apparently, TikTok was not caught quite as flat-footed by the threat of a ban in the U.S. as we thought. Apparently, they had tabled a proposal to let the U.S. pick its board of directors, have veto power over hires and other concessions, but the Biden administration went for the sell your company or your banned option instead, quoting the Washington Post. The video app owned by a Chinese company also said it would pay an American company that contracts with the Defense Department to monitor its source code according to a copy of the company's proposal. It even offered to give federal

officials a kill switch that would shut the app down in the United States if they felt it remained a threat. The Biden administration, however, win its own way. Officials declined the proposal forfeiting potential influence over one of the world's most popular apps in favor of a blunter option, a forced sale law signed last month by President Biden that could lead to TikTok's nationwide ban. The government has never publicly explained why it rejected TikTok's proposal, opting instead for a potentially protracted constitutional battle that many expect to end up before the Supreme Court. Since federal officials announced an investigation into TikTok in 2019, the app's user base has doubled to more than 170 million U.S. accounts, including Biden's

re-election campaign. But the extent to which the United States evaluated or disregarded TikTok's proposal known as Project Texas is likely to be a core point of dispute in court where TikTok and its owner bite dance are challenging the sailor ban law as an unconstitutional assertion of power. The government essentially threw up its hands at the possibility of any kind of regulation or cybersecurity measures said Anapum Chander, a Georgetown University law professor who researches international tech policy. TikTok proposed this incredible array of protections, but none of it mattered, he added. In the government's thinking it wasn't, can this app be protected? It was, there's a Chinese owner. That became the death now. The government had a complete absence of faith

in its ability to regulate technology platforms because there might be some vulnerability that might exist somewhere down the line, end quote. Finally today, let's check in on something we've not talked about in a while. Remember when delivery startups were all the rage? Well, a new analysis says that food delivery apps Deliveroo, Just Eat Takeaway, Delivery Hero, and DoorDash have racked up more than $20 billion in combined operating losses since they went public. So remember when I used to talk about these companies, I would say, I guess they finally solve the unit economics on this sort of thing. Well, looks like that continues not to be solved, though there may be a light at the end of the tunnel. Quoting the Financial Times,

shares and Deliveroo, Just Eat Takeaway, Delivery Hero, and DoorDash, the four largest standalone publicly listed food delivery businesses in the U.S. and Europe are all trading well below their pandemic-era peaks as investors scrutinize their business models. Following a period of pandemic lockdown-fueled growth, the four companies are now contending with a tougher macroeconomic environment that has hit consumers. As they make a renewed push to demonstrate profitability to investors, their cumulative annual operating losses have now hit $20.3 billion. Calculations by the Financial Times and Industry Analysts, the Delivery. World found. The figure covers the seven years since Deliveroo, Delivery Hero,

and DoorDash went public, and after Just Eat Takeway was created following a merger in 2020. It includes substantial write downs related to acquisitions and stock-based compensation. Investor's willingness to fund losses has changed, and now they want food delivery businesses to demonstrate sustainable, profitable growth after a rise in interest rates, said UBS analyst Joe Barnett Lamb. Rival Uber does not break out the profitability of its Eats business, but marked its first full year of operating profitability at a group level in 2023 following a concerted push to boost margins, a moment the company held as an inflection point. Nevertheless, stock market analysts are becoming more optimistic that the companies can improve financials. In April, the three European players said they expected this year to follow Doordesh in becoming free cash flow positive on an annual basis.

The focus on free cash flow follows a longstanding emphasis among companies in the sector on an alternative measure of profits, adjusted earnings before interest, tax, depreciation, and amortization that strips out a range of costs such as legal provisions. But many people, quote, don't see adjusted earnings metrics as a true level of profitability of the underlying business, says Joseph McNamara and analyst at City. Operating losses offer, quote, the best standardized view across all companies that minimized adjustments and other non-cash and non-operational impacts, said Amanda Biencasa Arena, a partner at management consulting firm Aeon. Whether the companies could show they were generating more cash than they were spending was the next major litmus test.

McNamara added now that the stage of growth at all costs was over, end quote. Song lyrics, why don't you tell your dad, get off my back. Tell him what we said about paint it black. Talk to you tomorrow.