Google Translate Continues to Improve
September 10, 2019
Google Translate is a handy tool, imperfect though it may be. Google made some cosmetic changes to the UI earlier this year, and now is rolling more substantial improvements. Pakistan’s Technology Times reports, “Google AI Translation Adds 60 New Languages.” Not only are more languages, from Afrikanns to Zulu, now included, the translate-images function has gotten a boost. Writer Sayyed Shehzer Abbas tells us:
“The prevalent tech giant Google is rolling out a significant update to the camera feature on its Translate app. The new version of the app adds support for 60 new languages. It’s great news for regular users of Google Translate, where the camera feature is fantastically useful for translating things like menus and signs. Key to the update is the integration of Google’s AI translation methods, known as neural machine translation (NMT). These models have been incorporated into Google Lens and the web version of Translate, but they are now supporting instant camera translation, too.
We noted these languages:
“New languages supported in the update include: Afrikaans, Arabic, Bengali, Estonian, Greek, Hindi, Igbo, Javanese, Kurdish, Latin, Latvian, Malay, Mongolian, Nepali, Pashto, Persian, Samoan, Sesotho, Slovenian, Swahili, Thai, Vietnamese, Welsh, Xhosa, Yoruba, Urdu and Zulu.
We learned:
“The updated version of the Google translate app will also automatically detect what language it’s looking at, which is handy if you’re traveling in a region where multiple languages are common.”
The write-up emphasizes the importance of language translation to companies like Google. It has become a benchmark for evaluating an enterprise’s AI capabilities, and it underpins software many have come to rely on, like AI assistance, commerce, and social media. Will Translate put Google ahead of the competition?
Cynthia Murrell, September 10, 2019
See (at Least Some) of the Data Google Collects About You
September 10, 2019
This SGT Report headline may not be strictly true, but the write-up is interesting nevertheless. They claim, “Google’s File on You is 10 Times Bigger than Facebook’s—Here’s How to View It.” Keep in mind that SGT Report can publish some interesting and often difficult to verify information.
The article extrapolates its figure from the experience of one developer:
“Curious about just how much of his data Google had, web developer Dylan Curran says he downloaded his Google data file, which is offered by the company in a hub called ‘My Account.’ This hub was created in 2015, along with a tool called ‘My Activity.’ The report issued is similar to the one Facebook delivers to its users upon request. Whether or not these reports are comprehensive is still up in the air, but Curran says his was 5.5 GB, which is almost ten times larger than the one Facebook offered him. The amount and type of data in his file, Mr. Curran says, suggests Google is not only constantly tracking our online movements but may also be monitoring our physical locations.
We noted:
“Curran’s Google report contained an incredible amount documentation on his web activity, going back over a decade. But perhaps more importantly, Google had also been tracking his real-life movements via his smartphone device or tablet. This included fairly random places he’d frequented, many of the foreign countries and cities he visited, the bars and restaurants he went to while in these countries, the amount of time he spent there, and even the path he took to get there and back.”
Though we cannot tell whether this much Google-gathered data is typical, it is true big tech companies gobble up a lot of user data. It is also clear that one should take the promises of Chrome’s “incognito mode” with a grain of salt. Concerned readers may want to navigate to the links the article shares for taking some control over this data: Here Google account holders can turn off location tracking and other features of Google apps; at this link you can set advertising preferences; and this is how to download that Google data file like Curran did.
We are not sure Google is really collecting 10 times more data than Facebook, but how one’s personal data is being collected and used online does warrant attention.
Cynthia Murrell, September 10, 2019
Read Two Google Ads and Call Me in the Morning
September 10, 2019
We know Google has been branching out every which way it can, largely through divisions like X Lab, Sidewalk Labs, Jigsaw, and Deep Mind, among others. Now Health Impact News reports that “Google Joins the Pharmaceutical Industry” through its healthcare divisions Verily Life Sciences and Calico Labs. Writer Kate Raines seems suspicious of Google’s motives and its ties to “Big Pharma.” She describes some projects that Verily is working on with partners in the pharmaceutical industry. (That company started out pursuing miniaturization tech and machine learning for projects like smart lenses.) What we find interesting here is Google’s heavy push into the healthcare arena—are they chasing Amazon?
We learn:
“Verily now partners with a number of pharmaceutical companies that develop vaccines on projects ranging from smart lenses with Alcon (a subsidiary of Novartis) and surgical robotics with Johnson & Johnson to early identification and intervention in chronic diseases with Merck Sharp & Dohme and diabetes management with Sanofi. Verily is partnered with Gilead on profiling the immune system to clarify the biological mechanisms of autoimmune disease and with Verve Therapeutics on nanoparticle formulations. Verily is also partnered with GlaxoSmithKline, the world’s largest vaccine manufacturer, in the development of bioelectronic medicine. With the creation of Galvani Bioelectronics in collaboration with GlaxoSmithKline, Verily now has its own pharmaceutical company that is working to ‘enable the research, development and commercialization of bioelectronic medicines,’ which aim to treat disease using miniaturized implanted devices. Another of Verily’s projects is the development of the ‘sterile insect technique’ to manipulate mosquito populations by releasing sterile male mosquitoes that will reduce the populations of insects carrying such diseases as dengue, Zika, chikungunya and yellow fever.
The company has also entered the clinical study arena, first with its own study called Baseline, which seeks to connect potential study participants with clinical research groups.”
That is indeed a lot of cooperation; whether that is a good or a bad thing we leave our gentle readers to decide. Raines spends less time describing Calico, which is focused on increasing the human lifespan. It was established by the former CEO of vaccine developer Genentech and now employs a Head of Drug Development who was once a VP at vaccine developer Amgen. Very suspicious.
Raines concludes by noting that Google’s search algorithm specifically makes it difficult to find certain information about vaccines that runs afoul of the “government and pharmaceutical industry.” I think others call that addressing the scourge of fake news, specifically anti-vaxer propaganda in this case.
Cynthia Murrell, September 10, 2019
Amazonia for September 9, 2019
September 9, 2019
At the airport. Waiting. Here are a few nuggets the Bezos bulldozer left in its wake last week.
Arrests at Amazon
Forward reported that “hundreds of protesters organized by the Jewish group Never Again Action” were making their voices heard. The scene was Boston. Allegedly 12 activist were arrested after entering the Amazon facility and refusing to leave. DarkCyber thinks that protesting ICE is likely to get one put on ICE in the local lock up. The key point in the write up:
Amazon does not provide services directly to ICE, but does provide cloud hosting services to ICE’s subcontractors, according to the tech website The Verge. Amazon Web Services also hosts Department of Homeland Security databases that allow officials from numerous agencies to track immigrants.
Amazon Wants in Your Auto
Google landed General Motors. Amazon wants to go along on your rides. “Alexa, Roll Down the Windows!” Inside Amazon’s Quest to Get in Your Car” explains:
Amazon has been working hard on Alexa Auto for the past two years. Now it hopes to convince automakers to embed the platform into their new cars.
A revised auto SDK is forthcoming. We learned:
Amazon is set to announce the second version of the Alexa Auto SDK. This update will allow Alexa to do more things when the car’s internet connection is interrupted, by switching to to a mobile phone (connected to the car via Bluetooth or USB) for the connection needed to call, message, or stream music from services such as Amazon Music and Pandora. The new SDK also enables a couple of new offline car-control features, including the ability to turn on defrost and in-cabin lighting. However, Alexa Auto SDK still does not support the ability to control the ignition, door locks, or headlights using voice commands, whether the car has an internet connection or not.
DarkCyber assumes that the Bezos bulldozer is already equipped with these capabilities.
Amazon Personnel Management Gets the Evil Eye
We spotted “The Human Cost of Amazon’s Fast, Free Shipping” in the New York Times and then on the MSN.com Web site. The write up appears to be a research summary with the original work done by Pro Publica. In short, Microsoft was keen to get this tabloid-esque exposé in front of Azure tinted eyeballs. Our take: Amazon is a bad personnel management outfit. Boo.
The main point is simple:
In its relentless push for e-commerce dominance, Amazon has built a huge logistics operation in recent years to get more goods to customers’ homes in less and less time. As it moves to reduce its reliance on legacy carriers like United Parcel Service, the retailer has created a network of contractors across the country that allows the company to expand and shrink the delivery force as needed, while avoiding the costs of taking on permanent employees.
Efficiency is okay. Efficiency which harms employees is not okay. But the human factor is likely to be shaped. Amazon wants robots. A capital investment is a two-fer: Lower taxes, no overworked humanoids burdening the online bookstore with benefits, health care, and on the job incidents.
Harsh Words for AWS
DarkCyber does not know if these assertions in this Reddit post are accurate. However, one may want to apprise oneself of these issues. Check out this Reddit post.
Summer Sale
If you use AWS EFS infrequent access, you get a deal. Silicon Angle reported that Amazon has cut prices AWS EFS Infrequent Access. How much of a price chop? For some customers, a $1.00 charge could become $0.08. Storage is just $0.08 per gigabyte. Also, Lifecycle Management service for EFS have been trimmed as well. Why? DarkCyber is hypothesizing but grousing about the “hidden” costs of AWS seem to be cropping up in online discussion groups. Plus, there’s some bad publicity about AWS reliability. Team Azure keeps pecking at AWS. Will more price cuts follow? Tough to predict the future.
Partners, Resellers, Integrators
Accenture. The accounting/consulting/billing machine has team with Amazon to offer managed blockchain services for “small scale producers into the value chain.” No we don’t know what this means. Source: Forbes
Baffle. This cyber security firm is now an “AWS Database Ready Technology Partner.” Source: Help Net Security
Esono. This consultant provides a VMware cloud on AWS. The function is “the new manager of cloud environment. Source: CIO Review
ICL. This global specialty minerals and specialty chemicals company will use AWS to deliver its digital services to agricultural professionals. ICL is based in Israel. Source: MarketWatch
Mobvista. The company is now part of the Amazon partner network. Source: Yahoo
NRGene. This AI and genomic outfit is not an Amazon advanced technology partner. Source: Digital Journal
Privo. This AWS consulting firm is now a premier consulting partner. Source: Marketwatch
Pureport. This multi cloud networking provider said that its Multi cloud Fabric platform now supports AWS Transit Gateway over AWS Direct Connect. Source: Capacity Media
Verimatrix. The Paris-based service provider has announced “interoperability between the Verimatrix Multi-DRM solution and the Secure Packager Encoder Key Exchange (SPEKE) API developed by Amazon Web Services (AWS)”. Bloomberg
Interesting Google Items
September 9, 2019
Waiting for a flight. Noted three interesting Google items.
The first is “Apple Criticizes Google for Creating False impression about the Scale of iPhone Hack.” The points seems to be that Google presented information that focused attention on Apple. We noted this assertion:
In a statement, Apple spokesman Fred Sainz said that the sophisticated attack on iPhones was “narrowly focused” and was not “a broad-based exploit of iPhones”, as described by Google Project Zero security researchers. Sainz also claimed that the attacks affected fewer than a dozen websites that provided content about the Uighur people living in China.
The write up added this fascinating factoid:
Last week, some reports also suggested that the hacked websites targeted Android and Windows users as well, but Google didn’t provide any details about that aspect of the attacks. Google claimed that it had not been aware that Android was affected in the attacks.
Is Google weaponizing information to discredit Apple? DarkCyber does not know.
Second, we spotted “Google Finally Confirms Security Problem For 1.5 Billion Gmail And Calendar Users.” The idea is that Google allegedly reacted to the security problem news this way:
A Google spokesperson responded to my story by insisting that “Google’s Terms of Service and product policies prohibit the spreading of malicious content on our services, and we work diligently to prevent and proactively address abuse.” That statement went on to say that Google offers “security protections for users by warning them of known malicious URLs via Google Chrome’s Safe Browsing filters.” Now, it seems, Google is finally taking this security problem somewhat more seriously.
And I spotted “Google Woes: Antitrust Concerns, YouTube Fine, and Corporate Culture Clashes Keep Tech Giant on the Hot Seat.” Google has a lot of billions to prevent serious burns.
Net net: These items create an impression that lovable Google may not be in touch with its inner flow. On the other hand, maybe these examples are Google.
Stephen E Arnold, September 9, 2019
Research Suggests Better Way to Foil Hate Groups
September 9, 2019
It is no secret that internet search and social media companies have a tough time containing the spread of hate groups across their platforms. Now a study from George Washington University and the University of Miami posits why. Inverse reports, “‘Global Hate Highways’ Reveal How Online Hate Clusters Multiply and Thrive.” This is my favorite quote from the article—“In it, [researchers] observe that hate spreads online like a diseased flea, jumping from one body to the next.”
The study tracked certain hate “clusters” across international borders and through different languages as they hopped from one platform to another. Current strategies for limiting the spread of such groups include the “microscopic approach” of banning individual users and the “macroscopic approach” that bans whole ideologies. Not only does the latter approach often run afoul of free speech protections, as the article points out, it is also error-prone—algorithms have trouble distinguishing conversations about hate speech from those that are hate speech (especially where parody is used.) Besides, neither of these approaches have proven very effective. The study suggests another way; reporter Sarah Sloat writes:
“The mathematical mapping model used here showed that both these policing techniques can actually make matters worse. That’s because hate clusters thrive globally not on a micro or macro scale but in meso scale — this means clusters interconnect to form networks across platforms, countries, and languages and are quickly able to regroup or reshape after a single user is banned or after a group is banned from a single platform. They self-organize around a common interest and come together to remove trolls, bots, and adverse opinions. …
“A better way to curb the spread of hate, the researchers posit, would involve randomly banning a small fraction of individuals across platforms, which is more likely to cause global clusters to disconnect. They also advise platforms to send in groups of anti-hate advocates to bombard hate-filled spaces together with individual users to influence others to question their stance.
“The goal is to prevent hate-filled online pits that radicalize individuals like the Christchurch shooter, an Australian who attacked in New Zealand, covered his guns with the names of other violent white supremacists and citations of ancient European victories, and posted a 74-page racist manifesto on the website 8chan.”
The researchers’ approach does not require any data on individuals, nor does it rely on banning ideas wholesale. Instead, it is all about weakening the connections that keep online hate groups going. Can their concept help society dissipate hate?
Cynthia Murrell, September 9, 2019
Yeah, We Are Sorry. Very, Very Sorry
September 8, 2019
If you do not remember the name James Damore, he was a former Google employee who authored the Google’s Ideological Echo Chamber aka the Google memo that described Google’s diversity as an “ideological echo chamber,” where the company believes that disparities are caused by oppression and attempts to fix them through reverse discrimination. Google fired James Damore after the memo made the rounds inside and outside the company. One Redditor named TiredOfLying4Google posted, “I Helped Google Screw Over James Damore” in the James Damore subreddit.
TiredOfLying4Google started that he/she saw the memo internally a month before it went viral. Google human resources did nothing, except send the memo up the reporting chain as internal feedback. When Damore’s memo leaked, Google took action:
“Unfortunately, the memo started spreading within the company. The floodgates opened and previously silent employees started talking. To quell dissent, we: told executives to write to their employees condemning the memo; manipulated our internal Memegen to bias the ratings towards anti-Damore posts (the head of Memegen is an “ally” to the diversity cause); and gave every manager talking points on what to tell their reports about the memo. In all our communications, we concentrated on how hurt employees purportedly were and diverted attention from Google’s discriminatory employment practices and political hegemony, never mind the science.”
TiredOfLying4Google continued that the company wanted to make an example of Damore, so they spied on him and tried to find a reason to terminate his employment. They did not discover anything, but his devices became extremely slow and probably prevented him from rallying support. Upon his dismissal, Google employees were afraid to speak up. TiredOfLying4Google also said that Google’s reputation took a hit.
Damore apparently knew about Google’s dubious practices, including Dragonfly-the censored Chinese search engine. TiredOfLying4Google was surprised Damore did not report those secrets, claiming Damore probably cared about Google.
Google took more extreme measures by cancelling an employee town hall to address the controversy, placed the blame on “alt-right trolls”, planted information with journalists, and controlled the entirety of the NLRB case and class action lawsuit. Google used its money, influence, and power to create false information to support dismissing Damore and keeping their employees in line. Damore does not hold any power and Google will continue to hold sway.
Whitney Grace, September 8, 2019
Google: A Friday Get Together in the Shadow of Dorian
September 7, 2019
When I worked in Sillycon Valley, Friday was a big deal. I am not sure why. Once or twice a month, I would trek to some local joint and hang out with others who worked at our whiz bang technology and cyber data company. In general, the mood was upbeat. We were making money. We did not have vulture capitals roosting on our shiny vehicles. We were not responding to US government mandated document collection tasks.
If the information in Mr. Jeff Bezos’s Washington Post is correct, the Googlers must have concluded that Dorian was pummeling them with rain, high winds, and untethered plastic pool floats. The story is titled “Google Receives Demand for Documents from Justice Dept., Acknowledging Federal Antitrust Scrutiny.” (I was able to read it after wading through the begging-for-dollars pop ups. Really, Mr. Bezos?)
I noted this statement, which may or may not be affected by someone who is breathing the fumes from the Bezos bulldozer idling in front of the Washington Post’s headquarters.
the Justice Department has requested records related to its prior antitrust investigations, marking the tech giant’s first major acknowledgment that it’s a subject of a federal competition probe. The civil-investigative demand — acknowledged in a securities filing and a blog post — comes weeks after Justice Department officials said they would open a broad review of big tech, including search.
Records requests are interesting. On the surface, the request is simple: Gather up the information from “past investigations.” On the other hand, fast-moving, high-tech companies are not really into archiving. Sure, there are document management systems, files on Google Drive, data tucked into USB sticks, paper stored in file cabinets (although some Googlers may not be familiar with actual records management conventions), and maybe –– just maybe — data in a Google social media system.
The unknown, as I understand the document landscape, is to comply with this simple government request.
But — and there is often a but — associated with a simple government request. The content Google provides will be compared with information that the investigators, lawyers, and analysts have.
Anomalies are, in general, not desirable. For example, if the government document reviewers have a document NOT in the Google collection delivered in compliance with the request, an int4eresting question can be raised:
Why did you Google not provide the same information you delivered in the prior antitrust matters? (Translation: We have info in our files from our previous look at you and you a leaving stuff out.)
Now let’s assume that there is information in the government’s file (usually maintained in accordance with assorted guidelines and regulations about US government document retention). Here’s the question:
Why did you provide a document pertinent to a prior antitrust matter that you previously did NOT provide? (Translation: The trove of documents you Google have just delivered includes information we have not seen before. Why?)
You can generate quite a string of questions from this type of matching exercise. Neither question trigger unencumbered joy of pre-demand Friday staff get togethers. (Did you know that Google owns the Sports Page in Mountain View?)
Worth monitoring for two reasons:
- Is Google’s record keeping up to snuff?
- Are the data provided congruent with what the lawyers, analysts, and investigators have in their files both paper and digital?
A digital Dorian in Mountain View?
Stephen E Arnold, September 7, 2019
Thinking about Real News
September 7, 2019
Now that AI has gotten reasonably good at generating fake news, we have a study that emphasizes how dangerous such false articles can be. The Association for Psychological Science reports, “Fake News Can Lead to False Memories.” While the study, from the University College Cork, was performed on Irish citizens ahead of a vote on an abortion referendum, its results can easily apply to voters in any emotional or partisan contest. Like, say, next year’s U.S. presidential election.
Researchers recruited 3,140 likely voters and had them read six articles relevant to the referendum, two of which were accounts of scandalous behavior that never actually happened. We learn:
“After reading each story, participants were asked if they had heard about the event depicted in the story previously; if so, they reported whether they had specific memories about it. The researchers then informed the eligible voters that some of the stories they read had been fabricated, and invited the participants to identify any of the reports they believed to be fake. Finally, the participants completed a cognitive test. Nearly half of the respondents reported a memory for at least one of the made-up events; many of them recalled rich details about a fabricated news story. The individuals in favor of legalizing abortion were more likely to remember a falsehood about the referendum opponents; those against legalization were more likely to remember a falsehood about the proponents. Many participants failed to reconsider their memory even after learning that some of the information could be fictitious. And several participants recounted details that the false news reports did not include.
We note:
“‘This demonstrates the ease with which we can plant these entirely fabricated memories, despite this voter suspicion and even despite an explicit warning that they may have been shown fake news,’ [lead author Gillian] Murphy says.”
Indeed it does. Even those who scored high on the cognitive test were susceptible to false memories, though those who scored lower were more likely to recall stories that supported their own opinions. At least the more intelligent among us seem better able to question their own biases. Alas, not only the intelligent vote.
In addition to fake articles that can now be generated quickly and easily with the help of AI, we are increasingly subjected to convincing fake photos and videos, too. Let us hope the majority of the population learns to take such evidence with a grain of salt, and quickly. Always consider the source.
Cynthia Murrell, September 9, 2019
A Reason Why Governments Tip Toe around Facebook?
September 6, 2019
Is Facebook a criminal evidence preservation resource?
How important is Facebook to law enforcement, and how important should it be? How broad should efforts be to preserve potential evidence on the platform? These are some questions we pondered when we read the Daily Caller’s article, “Facebook’s New ‘Clear History’ Tool Hits a Legal Roadblock Over Criminal Evidence Concerns.” The tool in question, Off-Facebook Activity, addresses privacy concerns by giving users control over what data other apps and websites share with Facebook. (It does not, however, completely “clear” user’s browsing history, as some had hoped.)
The Texas civil case at hand is one of human trafficking facilitated, alleges the plaintiff, through Facebook’s platform. The victim’s attorney, Annie McAdams, filed a temporary restraining order to block Off-Facebook Activity altogether, reasoning the tool would obscure data relevant to her case. In doing so, she cited the federal Fight Online Sex Trafficking Act and the Stop enabling Sex Traffickers Act, both enacted in 2018. This would seem reasonable but for a couple of factors, reporter Audrey Conklin argues. For one thing, U.S. Code Section 230 protects social media companies from being held responsible for content published by others on their platforms. Furthermore, Facebook has other mechanisms for preserving data relevant to legal cases. Citing tech lawyer Brad Shear, Conklin writes:
“Shear explained that law enforcement can still access information behind Off-Facebook Activity under proper legal procedures because ‘Facebook still has that information on the back end because they don’t delete any information on their platform.’”
Interesting. The article continues:
“Facebook is used regularly as a resource to catch both active and potential criminals, including mass shooters and drug traffickers, so long as law enforcement subpoenas the information needed to identify them. …
We note:
“Facebook also works with Polaris and the National Human Trafficking Hotline ‘to provide resources and assist victims of human trafficking,’ according to a page on its site dedicated to human trafficking information, which also includes a list of human trafficking hotlines in various countries. The social media giant complied with 88% of 41,336 U.S. government requests for information in 2018, according to Facebook Transparency. There were 23,801 search warrant requests made for 36,652 accounts, 90% of which were produced. Of 8,360 subpoena requests for 13,728 accounts, data was produced for 83%. Hundreds of requests related to national security threats were also requested.”
So, according to Conklin, McAdams is approaching the issue all wrong. Instead of filing a broad restraining order against the Off-Facebook Activity tool, she could have asked Facebook nicely. However, the Transparency page states it will preserve information upon request from government law-enforcement, not private attorneys. Perhaps the issue is more complicated than the write-up suggests, and McAdams is not simply tech-ignorant, as Conklin charges.
Whatever the case here, there is no denying Facebook data has been playing an increasingly important role in law-enforcement efforts. Is that why punishments are mostly hand waving?
Cynthia Murrell, September 06, 2019
-
- Subscribe to Beyond Search
Feature archive
News archive
-
