Has Google Smart Software Become the Sad Clown for AI?
April 20, 2021
“Is Google’s AI Research about to Implode?” raises an interesting question. The answer depends on whom one asks. For the high profile ethical AI Googlers who are now Xooglers (former Google employees), the answer is probably along the lines of “About. Okay, boomer, it has imploded.” Ask a Googler who still has a job at the GOOG and received a bonus for his or her work in smart software and the answer is probably more like, “Dude, we are AI.” With matters Googley, I am not sure where the truth exists.
The write up states:
in making certain “corrections” to large datasets, for example removing references to sex, the voices of LGBTQ people will be given less prominence. The lack of transparency and accountability in the data makes these models useless for anything other than generating amusing Guardian articles (my words, not the authors). But they have substantial negative consequences: in producing reams of factually incorrect texts and requiring computing resources that can have a major environmental impact.
Ah, ha, the roots of bias.
Google has not made enough progress is making its models neutral. Thus, human fiddling is required. And where there are humans fiddling, there are discordant notes.
The write up concludes with this statement:
What concerns me is that when Google’s own researchers start to produce novel ideas then the company perceives these as a threat. So much of a threat that they fire their most innovative researcher and shut down the groups that are doing truly novel work.
Right now, I think the Google wants to squelch talk about algorithmic “issues.” Smart software appears to be related maximizing efficiency. The idea is that efficiency yields lower costs. Lower costs provide more cash to incentivize employees to find ways to improve, for example, ad auction efficiency. Ethics are not an emergent phenomenon of this type of system. The result is algorithmic road kill, a major PR problem, a glimpse of the inner Google, and writers who are skeptical about the world’s largest online ad vendor’s use of “smart” technology.
Stephen E Arnold, April 20, 2021
The Google Is Busy: Use the Maps to Buy from Advertisers Already
April 15, 2021
Talk about ungrateful. Australia is annoying a US tech giant with unwarranted criticisms. Just because a highway has a petrol station every 500 or 600 miles, what difference does it make if the data on a free map are incorrect. That electric vehicle should have solar panels. Carbon fuel machines need to have auxiliary gas tank.
“Google Maps Under Fire for Incorrect Information, Outdated Imagery” is grousing and to the really busy Google. The write up asserts:
The errors on Google Maps go as far as claiming that the town of Eromanga is some 85 kilometers from its actual location, so drivers who may use the navigation to drive to this city could end up in a completely different place.
Like the Googzilla has time to figure out where a dirt road goes in Eromanga? Ho ho ho. Buy an ad. The Google may add Eromanga to its customer database. Well, maybe.
The write up continues:
Furthermore, according to local reports, the local Street View imagery is more than a decade old…
I have concluded that the article in auto Evolution has been assembled by individuals who are not Googley. The fix is directly from a Crocodile Dundee film; to wit:
As a result, the authorities recommend people stick with the traditional way of navigation and use the street signs to find a specific location. “If you see a signpost saying a town is ‘this way’ and Google Maps is telling you something different, don’t trust Google Maps,” Quilpie Shire Council Mayor Stuart Mackenzie said.
Australia. Consistently annoying. Eromanga, really? Just buy some ads. The Google cares about ads even if these messages are for enterprises in where was it? Oh, right, Eromanga.
Stephen E Arnold, April 15, 2021
Google: Its Feedback Loop Explained
March 23, 2021
I read “Google Profits from Spreading Fake News — Here’s How.” Google’s been leveraging its “inspiration” from Yahoo-GoTo-Overture ad innovations for decades. Imagine my surprise when the “truth” of feedback was finally revealed. (Yep, it took decades for whiz kids to crack the somewhat high-school auditorium sound system concept.)
Here’s a passage I found revelatory about how little awareness “expert” Google watchers know about the systems and methods of the online ad giant:
When you click on a search result, the search algorithm learns that the link you clicked is relevant for your search query. This is called relevance feedback. This feedback helps the search engine give higher weight to that link for that query in the future. If enough people click on that link enough times, thus giving strong relevance feedback, that website starts coming up higher in search results for that and related queries. People are more likely to click on links shown up higher on the search results list. This creates a positive feedback loop – the higher a website shows up, the more the clicks, and that in turn makes that website move higher or keep it higher.
What other Google magic awaits discovery?
Remarkable. That feedback has baffled for so long.
Stephen E Arnold, March 23, 2021
Google and Cookies: Crafting Quite Tasty Bait
March 19, 2021
I read “Alphabet: Five Things We Know about Google’s Ad Changes after Cookies.” I approached the write up with some interest. Cookies have been around for a long time. The reason? They allowed a number of interesting functions, including tracking, cross correlation of user actions, and a covert existence.
Now, no more Google cookies.
The write up explains what Google wants keen observers, real journalists, and thumbtypers to know; to wit:
- Privacy is really, really important to Google—now. Therefore, the GOOG won’t support third party cookies. Oh, shucks, what about cross site tracking? Yeah, what about it?
- Individuals can be targeted. Those with a rifle shot orientation have to provide data to the Google and use the Google software system called “customer match.” Yeah, ad narrowcasting lives.
- Google will draw some boundaries about its data leveraging for advertisers. But what about “publishers”? Hey, Google has some special rules. Yeah, a permeable membrane for certain folks.
- FLOC makes non-personalized ad targeting possible. I want to write, “You’ve been FLOC’ed” but I shall not. Yeah, FLOC. But you can always try FLEDGE. So “You’ve been FLEDGED” is a possibility.
How’s this work? The write up does not shed any light. Here’s a question for a “real news” outfit to tackle:
How many data points does a disambiguation system require to identify a name, location, and other personal details of a single individual?
Give up. Better not. Oh, the bait, pivoted cookies. Great for catching prospects I think.
Stephen E Arnold, March 19, 2021
What Makes the Web Slow? Really Slow?
January 28, 2021
I read “We Rendered a Million Web Pages to Find Out What Makes the Web Slow.” My first reaction was the East Coast Internet outage which ruined some Type A workers’ day. I can hear the howls, “Mommy, I can’t attend class, our Internet is broken again.”
Here’s a passage from the “Rendered a Million Web Pages” which I found interesting:
Internet commentators are fond of saying that correlation does not equal causation, and indeed we can’t get at causality directly with these models. Great caution should be exercised when interpreting the coefficients, particularly because a lot confounding factors may be involved. However, there’s certainly enough there to make you go “hmm”.
Yep, I went “hmm.” But for these reasons:
- Ad load times slow down my Web experiences. Don’t you love those white page hung ads on the YouTube or the wonky baloney on the Daily Mail?
- How about crappy Internet service providers?
- Are you thrilled with cache misses?
- Pages stuffed full of trackers, bugs, codes, and spammy SEO stuff.
Hmm, indeed.
Stephen E Arnold, January 28, 2021
Oracle: Dons White Hat and Tries to Lasso the StreamScam Stallion
January 6, 2021
Oracle is only of passing interest to me. I paid attention to Oracle’s search efforts which I think once involved Applied Linguistics, TripleHop, Endeca, and a natural language processing company crafting smart software to replace trained human customer support types. Oh, I did check out Oracle SES. As I recall, the first “s” meant secure. The idea was that Oracle had discovered that most vendors of enterprise search provided systems which were not focused on security. Just getting these overhyped and jargon infused puppies to behave consumed quite a bit of developer, MBA, and ultimately CPA energy.
I read “Oracle Exposes largest CTV Ad Fraud Operation Ever” reminded me of the Oracle Secure Enterprise Search hook: Enterprise search vendors did not provide Oracle-grade security. That was, as it turned out, mostly true. The problem was that organizations purchasing enterprise search were not buying security. The organizations were trying to find a way to deal with the increasing flows of digital information. Prior to the implosion of Delphes, Entopia, and Fast Search & Transfer — tools were not widely available. Today, of course, an enterprising and enthusiastic developer can download or just use a variant of Lucene/Solr. Amazon and IBM support these open source solutions, and developers of proprietary systems like Coveo, Mindbreeze, and others have to put in extra hours on their sales Peloton’s to generate sustainable revenues. Oracle’s run at enterprise search as security disasters went nowhere.
Now Oracle has donned its white hat and is now lassoing or trying to lasso the online advertising sector. In my research, we have encountered numerous reports of online advertising fraud. Making the charges stick to the Google, its subsidiaries, outfits like Facebook, and the third party intermediaries has been difficult. Gobbledygook explanations and intentional complexity are designed to keep those advertising dollars flowing.
The StreamScam is probably the first in a series of oracular pronouncements about alleged fraud: Click, view time, reach, etc., etc. The write up states:
StreamScam perpetrators capitalized on vulnerabilities in the technology used to improve the video viewing experience in CTV. Known as Server-Side Ad Insertion (SSAI), the technology combines content and ads into a single video stream that can play seamlessly with no delays on end-user devices, such as Roku, AppleTV, and FireTV.
The idea is that advertisers don’t know if online advertising generates sales. Marketers emphasize misty notions of brand and reach. But what’s an advertiser to do? The answer is that newspaper, television, and radio have been replaced among certain desirable market segments by streaming, Twitter, and podcasts. Therefore, a device like a smart TV seems to straddle some of these technology. Put money into connected televisions, YouTube, Facebook, and maybe — just maybe — TikTok.
I agree with “CTV Ad Fraud Schemes Like the One Oracle Exposed Will Become More Common But That Won’t Affect Marketers’ Spend.” I would, however, add a caveat to this write up’s assertion about not impacting what marketers spend and where; to wit: What choice does an advertiser have? Old fashioned direct mail? When the post office worked, that was an option. How about telemarketing? Some desirable demographics don’t answer their mobile’s chirps. What about billboard advertising? Covid and work from home may have reduced the impact of these view enhancing objects of art.
With more and more GenX and millennials positioning themselves as experts in social media, online ads are necessary complements to influencer campaigns. The Google ad purveyors are reassuring and armed with data illustrating that online ads really do work. Don’t like the Google rep, check out the Facebook pitch with micro and nano targeting that really works better than Googzilla’s approach. Amazon, Apple, yes. Options.
Net net: Ad fraud is endemic. No one survives who documents it. But Oracle has a white hat and maybe will own TikTok some day. There’s gold at the end of the digital rainbow even though one end is in Beijing.
My take: “Whoa, StreamScam! The Lone Oracle is gonna break you down. Right, Don Quixote.” (Tonto is in Covid lockdown.)
Stephen E Arnold, January 6, 2021
Another Xoogler Explains Algorithmic Manipulation
December 25, 2020
I don’t want to make a big deal about a former Google engineer talking about algorithmic manipulation. I know what happened to the Google AI expert who pointed out that training data can and does bias how numerical recipes make decisions.
I spotted this 2019 statement in an ancient write up called “‘YouTube Recommendations Are Toxic,’ Says Dev Who Worked on the Algorithm.” The speaker is a Xoogler (the semi official name for someone who worked at the Google) who allegedly worked on the YouTube recommendation algorithm. Now keep in mind that the Google is a pretty chaotic outfit, and it is possible for people to “work on” something and the outside world will have zero idea whether the contribution was a quality test or something substantive like fiddling thresholds to meet a harried superior’s goal. (Bonus time causes some interesting activities I have heard.)
Here’s the quote, and I have put in bold face the important statement which I found important and possibly accurate:
“It isn’t inherently awful that YouTube uses AI to recommend video for you, because if the AI is well tuned it can help you get what you want. This would be amazing,” Chaslot told TNW. “But the problem is that the AI isn’t built to help you get what you want — it’s built to get you addicted to YouTube. Recommendations were designed to waste your time.”
The write up does not dig into wasting time. I want to share my perception of the time wasting angle. In the good old days, Web sites wanted to be sticky. That’s why mere search engines became portals and eventually massive one stop shops with everything in one “experience.”
For YouTube, the more time a person invests in watching videos on YouTube, the more ads Google can slam into the video. If you think there are a lot of ads for a video now, just wait until the “game plan” is rolled out to the Googlers in the Spring of 2021.
Therefore, the purpose of the YouTube algorithm is to create opportunities to display ads. Are these relevant or irrelevant. I must say that I am quite adept at clicking past blandishments for Grammarly, Liberty Mutual, and many other hapless companies dumping cash into the coffers of the world’s most wonderfulest Web search system. Grammarly, isn’t “wonderfulest” a real word when used with “Google”? Maybe I should as DeepMind? Oh, right. DeepMind is busy doing healthy things and losing hundreds of millions of dollars every year.
Burn that ad inventory. Absolutely.
Stephen E Arnold, December 25, 2020
A Facebook Promise: Good As Gold
December 3, 2020
Oops. A Facebook algorithm’s mistake is causing the company to offer apologies and refunds, we learn from CNBC’s article, “Facebook to Reimburse Some Advertisers After Miscalculating Effectiveness Data.” Citing a report from Ad Exchanger, writer Lucy Handley informs us:
“The company’s ‘conversion lift’ tool suffered a glitch that reportedly affected thousands of ads between August 2019 and August 2020. Facebook fixed the error in September and is now offering a credit to clients ‘meaningfully affected’ by the bug. Conversion lift helps brands understand how ads lead to sales, using a ‘gold-standard methodology’ that links ads on Facebook’s platforms, including Instagram, to business performance, according to an explanation of the tool on Facebook’s website. The free tool shows ads to separate test and control groups and then compares sales conversions for each. Then, based on the results of the study, an advertiser can decide how much to spend on the social network.”
Though the error was discovered and fixed in September, the company is just now getting around to informing clients. According to Facebook, a “small number” of advertisers were affected, though what that behemoth considers a small number is unclear. Handley reminds us:
“This isn’t the first time Facebook has admitted mistakes in reporting. In September 2016, it said it overestimated the average time people spent viewing video ads over a two-year period, and in 2017 a report found that Facebook claimed to reach more people in some U.S. states and cities than official population data said existed in those areas.”
Yep, Facebook is starting 2021 with its true colors flying. I suppose it is nice to see some things remain consistent.
Cynthia Murrell, December 3, 2020
The European Competition Commission Goes for the Throat
November 3, 2020
I wanted to note the October 30, 3030, Reuters’ story “Online Giants Will Have to Open Ad Archives to EU Antitrust Regulators.” At last regulators are taking steps to gain access to the systems and methods used by Google and other online ad giants. The news story helps cement Margrethe Vestager as someone who uses her position to do more than posture. Also, the news story points out that there is a research agency called Algorithm Watch.
The problem is that the companies asked to provide information have legal options. The delays are likely to slow the regulators’ quest for data. If sufficient time goes by, the landscape can be reworked. Internet time is different from regulators’ time.
There is a counter point. Navigate to “Monopoly Power Is Less Dangerous Today Than in Past.” The argument set forth in this Telegraph Herald write up is unlikely to have a significant impact on the good ship SS Margrethe.
Stephen E Arnold, November 3, 2020
Google Filtering: How Smart Is Software?
October 27, 2020
I included a screenshot illustrating YouTube search results which make it clear how to obtain without charge copy-protected commercial software. You can read that story and see the screen shot at this link. I want to document a Reuters’ report called “Italy’s Communications Watchdog Fines Google for Betting Ad.” The news item documents that Google was fined for running gambling ads. The DarkCyber research team has been monitoring some of the questionable video streaming sites. Advertisements are appearing on these sites in greater numbers. What vendors are providing these paid messages? At this time, there’s no open source information about the intermediaries involved.
Questions:
- Why doesn’t filtering by key word work for Google advertisements? Gambling seems to be a no brainer.
- Why are Google YouTube search results providing recently updated links to video content which appears to violate a number of rules and regulations? The word “crack” is like gambling a seemingly obvious yellow caution light.
- What are the names of the ad agency intermediaries providing advertisements to what appear to be illegal video streaming sites?
Interesting? The DarkCyber research team finds the subject engaging. Smart software seems to have some blind spots.
Stephen E Arnold, October 27, 2020
-
- Subscribe to Beyond Search
Feature archive
News archive
-
