Another Way to Inject Ads into Semi-Relevant Content?
May 25, 2021
It looks like better search is just around the corner. Again. MIT Technology Review proclaims, “Language Models Like GPT-3 Could Herald a New Type of Search Engine.” Google’s PageRank has reigned over online search for over two decades. Even today’s AI search tech works as a complement to that system, used to rank results or better interpret queries. Now Googley researchers suggest a way to replace the ranking system altogether with an AI language model. This new technology would serve up direct answers to user queries instead of supplying a list of sources. Writer Will Douglas Heaven explains:
“The problem is that even the best search engines today still respond with a list of documents that include the information asked for, not with the information itself. Search engines are also not good at responding to queries that require answers drawn from multiple sources. It’s as if you asked your doctor for advice and received a list of articles to read instead of a straight answer. Metzler and his colleagues are interested in a search engine that behaves like a human expert. It should produce answers in natural language, synthesized from more than one document, and back up its answers with references to supporting evidence, as Wikipedia articles aim to do. Large language models get us part of the way there. Trained on most of the web and hundreds of books, GPT-3 draws information from multiple sources to answer questions in natural language. The problem is that it does not keep track of those sources and cannot provide evidence for its answers. There’s no way to tell if GPT-3 is parroting trustworthy information or disinformation—or simply spewing nonsense of its own making.”
The next step, then, is to train the AI to keep track of its sources when it formulates answers. We are told no models are yet able to do this, but it should be possible to develop that capability. The researchers also note the thorny problem of AI bias will have to be addressed for this approach to be viable. Furthermore, as search expert Ziqi Zhang at the University of Sheffield points out, technical and specialist topics often stump language models because there is far less relevant text on which to train them. His example—there is much more data online about e-commerce than quantum mechanics.
Then there are the physical limitations. Natural-language researcher Hanna Hajishirzi at the University of Washington warns the shift to such large language models would gobble up vast amounts of memory and computational resources. For this reason, she believes a language model will not be able to supplant indexing. Which researchers are correct? We will find out eventually. That is ok, we are used to getting ever less relevant search results.
Cynthia Murrell, May 25, 2021
Google Smart Software: Bert and Ernie Subbed for Joe Fourier
May 24, 2021
The story “Google Replaces Bert Self Attention with Fourier Transform: 92% Accuracy, 7 Times Faster on GPUs” may be more smart software hog wash. Against the colorful background of Google’s artificial intelligence melodrama, now Bert (shorthand for the wonderfully named Bidirectional Encoder Representations from Transformers) has some flaws. It seems accuracy and performance are among the ones which caught someone’s attention at the Google.
What’s up next?
The 19th century methods of Joe Fourier, who dabbled in fancy math in the 18th and 19th century until his vibrations petered out in 1830. Joe may be gone, but this math has new life. Why? Bert, it appears, was not exactly a high performer in the 100 meter sprints needed to make ad centric smart software work.
The write up is filled with jargon. That’s okay since anyone who reads about Fourier transforms knows something about the methods. However, what’s left out is an explanation of how the DeepMind research team has reacted to the Sesame Street infused acronym.
My hunch is that there are some sharp differences of opinion because, well, Google, smart software, algorithmic issues, and management methods.
Stephen E Arnold, May 24, 2021
Google: Technology Which Some Day Will Be Error Corrected But In the Meantime?
May 20, 2021
I read “Can Google Really Build a Practical Quantum Computer by 2029?” which is based on the Google announcements at its developers’ conference. The article reports:
… the most interesting news we heard at I/O has to be the announcement that Google intends to build a new quantum AI center in Santa Barbara where the company says it will produce a “useful, error-corrected quantum computer” by 2029.
Didn’t Google announce “quantum supremacy” some time ago? This is an assertion which China appears to dispute. See “Google and China Duke It Out over Quantum Supremacy.” Let’s assume Google is the supremo of quantumness. It stands to reason that the technology would be tamed by Googlers. With smart software and quantum supremacy, what’s with the 96-month timelines for “a useful” quantum computer?
Then I came across a news item from Fox10 in Arizona. The TV station’s story “Driverless Waymo Taxi Gets Stuck in Chandler Traffic, Runs from Support Crew” suggests that a Google infused smart auto got confused and then tried to get away from the Googlers who were providing “support.” That’s a very non-Google word “support.” The write up asserts:
A driverless Waymo taxi was caught on camera going rogue on a Chandler intersection near a construction site last week. The company told the passenger at the time that the vehicle was confused with cones blocking a lane.
The Google support team lifted off. Upon arrival, the smart taxi with a humanoid in the vehicle “decided to make for a quick getaway.”
According to Business Insider, “Waymo issued a statement that it has assessed the event and used it to improve future rides.” If you want to watch smart software and the incident, you can navigate to this YouTube link (YouTube videos can be disappeared, so no guarantees that the video will remain on the Googley service.)
Amusing. Independent, easily confused smart Google vehicles and error-corrected quantum computers. Soon. Perhaps both the Waymo capabilities and the quantum supremacy are expensive high school science club experiments which may not work in way that the hapless rider in the errant Waymo taxi would describe as “error corrected”?
Stephen E Arnold, May 20, 2021
Surprise! Prabhakar Raghavan Runs Google
May 20, 2021
The revelation that the Google boss is Prabhakar Raghavan, a former Verity executive who jumped to Yahoo and bounced into Google. The article/interview “Prabhakar Raghavan Isn’t CEO of Google—He Just Runs the Place” is an entertaining romp through search and a number of other topics. Please, read the interview. I want to highlight three statements from the article and offer a few observations.
Dr. Raghavan is quoted as saying:
Even now, with all the resources of Alphabet, if you remove one of the biggest constraints, such as competition, the problem doesn’t become easier.
My interpretation of this statement is that Dr. Raghavan has admitted that Google has no competition. Perhaps the “if” is the word that allows my conclusion to be disputed. Nevertheless, search is tough; there’s no competition; and the first page of most query results are ads and Google-ized content. “Objective” and relevant search results may appear on subsequent results pages. Hunting for the optimal result is as tedious as digging through the pages of the paper version of the Readers’ Guide to Periodical Literature circa 1964.
Here’s another interesting statement:
For instance, if you ask, “What’s today’s weather?” I show you a module of the weather, I don’t send you off somewhere else.
I think this means that Google engineers stickiness into its search results. Google does not want the user to click away and possibly not enjoy more Googley goodness. Over the years, the stickiness has ossified into what is a walled garden. Google caches the most clicked on content and serves it from a point close to the user. The originating site is not involved in some cases. The walled garden is replacing Wild West swinging doors with those one-way doors found in secure facilities. Easy in but no way out.
My final circled item from the article/interview was:
There is no circumstance under which I would say let’s toss that out and become a conventional company, because that is not what is going to lead us to innovate and serve humanity.
Google does not want to be conventional.
Interesting. Google seems to be evolving into an entity which has not been seen for decades: A quasi-country which operates in a way that gives it as much control over what information flows as an old-fashioned dictatorship. The list of Google prohibited words is fascinating. The banned creators on YouTube is varied and evolving. The company’s approach to innovation is “me to, me to”. The short attention span for products and services is a challenge to developers.
How will Google fare in the harsh light of more legal scrutiny? Will Google’s handling of “ethical AI” become the standard for the emerging smart software sector? Will the learnings imparted at Verity transfer to the Google environment? Will Google deliver relevant search results without the curve balls that seem to be the go-to pitch for the core service?
No answers are available for these questions. Maybe in the next round of European Union and US legal activities.
Stephen E Arnold, May 20, 2021
Google Says, Roku You!
May 19, 2021
YouTube TV is a popular streaming service accessible through various sling boxes a.k.a. carriages. Apple TV, Amazon Fire, and Roku are the top carriages that offer YouTube TV. Roku, however, could soon be off that list says Android Police in: “Google Tells YouTube TV Customers To Spam Roku Support.”
Google owns YouTube and its affiliate TV streaming service and the search engine giant is pursuing a new contract with Roku. Google wants Roku to push YouTube TV to the front of its interface. Roku, however, claims Google is using its monopoly power to force them into a new contract.
“…Google is looking to position YouTube as the premier streaming destination on Roku, requesting a dedicated search page, boosted music results, and a full block on search returns from other sources while the YouTube app is open. In perhaps its most brazen move, the company has reportedly asked Roku to meet specific hardware requirements to continue carrying its apps. On paper, none of these demands have anything to do with YouTube TV. Instead, Roku alleges that Google is using negotiations for its cable streaming service to gain leverage for YouTube as a whole, a move labeled “anticompetitive” in this morning’s email.”
Roku reasonably requests its users to petition Google to keep YouTube TV on its carriage. Google denies that it wants access to user data or it would interfere with search results. They also told YouTube TV subscribers to message Roku’s customer support in order to keep the cable alternative.
Bicker, bicker, bicker. Google wants more power over Roku, but the company will not cave. Solution: Just stick the YouTube TV thing into the YouTube free thing. That’s a “Roku you” for sure.
Whitney Grace, May 19, 2021
Google Tesla: A New Play for the Final Frontier?
May 14, 2021
I read some real “news provided by Google Cloud”. The story was “Google Cloud and SpaceX’s Starlink to Deliver Secure, Global Connectivity.” The write up said:
Google Cloud and SpaceX today announced a new partnership to deliver data, cloud services, and applications to customers at the network edge, leveraging Starlink’s ability to provide high-speed broadband internet around the world and Google Cloud’s infrastructure. Under this partnership, SpaceX will begin to locate Starlink ground stations within Google data center properties, enabling the secure, low-latency, and reliable delivery of data from more than 1,500 Starlink satellites launched to orbit to-date to locations at the network edge via Google Cloud. Google Cloud’s high-capacity private network will support the delivery of Starlink’s global satellite internet service, bringing businesses and consumers seamless connectivity to the cloud and Internet, and enabling the delivery of critical enterprise applications to virtually any location.
Tweets ensued. Pundits pundited.
Most of the comments focused on the key words the wordsmiths at the GOOG included; for example:
- Edge
- Organizations
- Private network
- Rural
- Seamless connectivity
Google knows what words are popular and elicit clicks. Very terrestrial.
However, the tie up — if it works out and does not get marginalized like Google’s Dodgeball, Orkut, and WebAccelerator efforts — could be something slightly more ambitious than connecting people to persistent advertising.
The deal makes the on-again, off-again lovebirds Google and Tesla the hot couple. The immediate payoff is publicity, usually good for a company’s stock price. The deal also may irritate some of the space-crazed wizards at Amazon, who also offer cloud services. Advertisers may like the idea of global delivery of compelling messages about vacation rentals, chicken sandwiches, and grammar checking software.
I see three different angles.
The first is that the Google – Tesla thing may represent a method for providing a new type of meta-fabric for innovation. If the service works, advertising may be available for those wanting to distribute messages globally to specific individuals who are going to buy something no matter where they are.
Second is that the service will be space-based. That alone provides a marketing and hyperbole edge. Even if the service burns up in the earth’s atmosphere, the deal is a juicy one. Think about those mid tier consultants writing “reports” this weekend. Boom. Consulting opportunities.
Third is that if the visions of the Musk-its (not Muscovites mind you) and the Googlers come to pass, we have a new type of information company in development. If the blended system works, the old AT&T is going to look even more old fashioned than it does at this time.
Net net: Satellites, a Saturday Night Live host, smart advertising, the “cloud” — what’s not to like? Will regulatory authorities buy the terrestrial spin in the Google news release? Probably. In the absence of regulatory controls, the sky’s the limit. Maybe I should say “the solar system” or maybe the “universe” is the limit. Amazon, Microsoft, Apple? Your move.
Stephen E Arnold, May 14, 2021
Speak Using Our Words, Or Do Not Speak
May 14, 2021
Google has learned from legal misfortune, both its own and other companies’. That is why, “To Head Off Regulators, Google Makes Certain Words Taboo.” The Next Web post outlines several of the major antitrust investigations the company currently faces at home and abroad. It also describes the role language played in past lawsuits brought against Google and, notably, Microsoft. We learn employees are given specific instructions on their language and other parts of communication both inside and outside the company. Having acquired some internal documents, the journalist known as The Markup writes:
“The taboo words include ‘market,’ ‘barriers to entry,’ and ‘network effects,’ which is when products such as social networks become more valuable as more people use them. ‘Words matter. Especially in antitrust law,’ reads one document titled Five Rules of Thumb for Written Communications. ‘Alphabet gets sued a lot, and we have our fair share of regulatory investigations,’ reads another. ‘Assume every document will become public.’ The internal documents appear to be part of a self-guided training session for a wide range of the company’s more than 100,000 employees, from engineers to salespeople. One document, titled ‘Global Competition Policy,’ says it applies not only to interns and employees but also to temps, vendors, and contractors. The documents explain the basics of antitrust law and caution against loose talk that could have implications for government regulators or private lawsuits. In one of the documents, which appear to be written by the legal team, employees are advised to choose their words carefully and use only third-party data when referencing Google’s ‘position in search’ in sales pitches. They are further cautioned never to print or hand out their slides.”
The documents helpfully suggest alternative words, including “industry,” “space,” “area,” or the name of a region instead of “market;” “valuable to users” rather than “network effects;” and “challenges” instead of “barriers to entry.” Though employees may (mis)use terms innocently, history tells us lawyers and regulators can and will seize upon certain definitions to build their cases. The higher in the company one is, the riskier careless language becomes. Especially sensitive are phrasings that suggest Google dominates any market, intends to “crush” its competition, or makes any choice for its own advantage rather than for the benefit of users. Because, of course, Google would never do that.
Cynthia Murrell, May 14, 2021
The Amusing Antics of Big Tech Monopoly-Type Companies
May 13, 2021
If I use my imagination, I can hear the comments in the TV room of a fraternity house near the Chambana campus of the University of Illinois. “Dudes, we can make the losers at Sigma Nu look really stupid.” Then the snort, snort, snort of perceived victory over lesser beings.
I thought about this hypothetical bro-moment when I read two stories this morning.
The first is “Microsoft Edge Blocks Firefox Installer, Says It’ll Hurt Your PC.” Firefox has had its share of challenges. There’s the money thing, the management thing, and the number of users thing. Microsoft, the all-time leader in security, has determined that Firefox is allegedly a danger. The write up reports:
“Firefox Installer.exe was blocked because it could harm your device,” the warning read, with users only able to click through to see more details rather than continue the download. Techdows says that all versions of the Firefox Installer, including release, beta, dev, and nightly, appear to be affected, with multiple Reddit threads detailing download issues. Some users were able to download and install Firefox using Edge after disabling Microsoft Defender SmartScreen, a program.
That seems like a predictable response from those who have witnessed commentary in the hypothetical frat house.
The second is “Google: We Put YouTube TV in the Main YouTube App. What Now, Roku?” The idea is that Roku, the hardworking salary man of online video, is going to be reminded that the Google is the top dog. The write up states:
Google announced in a blog post that it was just going to run an end-around on Roku and stick the YouTube TV app in the YouTube app.
No one fools around with Mother Google.
What do these frat mentality actions by two large companies tell us? Perhaps these are routine business practices in the regulation and consequence free datasphere of 2021? Could these actions indicate that fraternity type thinking remains a core part of the technology world in the US? Or is there a darker implication; for instance, these actions are perceived as just what has to be done to ensure that big outfits get larger?
From my point of view, I find the frat-style a reminder that what characterizes those in extended adolescence appears to be the warp and woof of high technology: Competitive products are harmful or too stupid to cope with Googley reality.
Stephen E Arnold, May 13, 2021
Googley Logic: Can Money Buy Success, Innovation, and Insecurity?
May 12, 2021
Two items caught my attention this morning. No, I am not talking about warfighting, a horse trainer’s about face on the use of controlled substances in the horsey world, or melting the Eredivisie trophy into 42,000 champion stars. (Almost everyone is special in the Ajax fan world!)
Nope. Nope. Nope.
The first item is the Wall Street Journal’s story “Google Plans to Double AI Ethics Research Staff.” The story will demand payment for Mr. Murdoch’s gem of journalism. The main idea for the story is that Google will pump more cash into “the team tasked with evaluating code and product to avert discrimination…” How will this management tactic work out? I cannot predict the future. My hunch is that when it comes to figuring out averting discrimination, the “team” may find itself in some interesting chalk board meetings with the coach. If have fiddled with models to make the numbers flow, you may not appreciate how those subjective decisions can cascade through smart systems. Toss in the knock on effects of unmonitored feedback loops, and you get some darned thrilling moments in smart software outputs. Is it possible that by spending money be demonstrates the decider thought process used by a high school science club to determine what toppings are put on a vegetarian pizza?
The second item is “YouTube Announces a $100 Million Fund to Reward Top YouTube Shorts Creators over 2021-2022.” I don’t think the 2021-2022 refers to the age of the YouTube stars who qualify. The main point of this write up is to illustate that the Google is serious about short form video. Like other me-too innovators, the GOOG is using money to signal its innovative excellence. YouTube appears to have been blissfully ignorant of TikTok’s become a hot property — what? — four or five years ago. Now the Google is prepared to spend big to develop its new, revolutionary service. TikTok, the cash splash says, you are toast. Maybe Alphabet will issue a TikTok video to the song “I’m Gonna Get You” once the rights issues are worked out. Here’s a verse from the tune:
When you’re driftin’ off to sleep
Close your eyes and think of me
Make it easy on yourself
Don’t dream about nobody else
Is this a nightmare or a lullaby?
Let’s step back.
These two management actions raise some questions for me; for example:
Can money buy trust for Google’s management with regards to ethical AI?
Can money buy innovation in short form video and other new media formats?
Can Alphabet Google YouTube management respond in effective ways to the legal challenges about how it does business, the Amazon surge in product search, and the surprising proliferation of ads in Google search results, maps, and other “free” services?
I have zero insight into the workings of the Google. The firm’s management decisions are fascinating to observe: So big, yet so darned high school science club and certainly “diverse.”
Stephen E Arnold, May 12, 2021
Xoogler Meredith Whittaker Explains How to Be Google Grade
May 7, 2021
I read the interview called “Ex-Googler Meredith Whittaker on Political Power in Tech, the Flaws of ‘The Social Dilemma,’ and More.” Very Silicon Valley. You will need to work through the transcript yourself. Here are the points I circled as checkpoints for being Google Grade. The phrase in my lingo means “How to keep your job at the GOOG.” I identified six behaviors; your mileage may vary.
- Be a white male.
- Float above the concerns of non-Google grade type people.
- Emulate senior Google leaders; for example, the affable, other directed Jeff Dean.
- Ethics. Ho ho ho. Embrace phenomenological existentialism within the Google context.
- Respond like a Pavlovian dog or pigeon when money and power are the payoff.
- Fight the impulse to be a contrarian.
And the interview ends on an interesting note. The Xoogler allegedly said:
It’s going to be really hard to repurpose that toward democratic, people-driven ends, given the consolidation of power that is right now dominating those infrastructures and given the neoliberal capitalist incentives that are driving those who dominate those infrastructures.
Maybe not hard, just too late.
Stephen E Arnold, May 7, 2021
-
- Subscribe to Beyond Search
Feature archive
News archive
-
