Listen to the Podcast
16 Oct 2020 - Podcast #715 - (19:59)
It's Like NPR on the Web
If you find the information TechByter Worldwide provides useful or interesting, please consider a contribution.
If you find the information TechByter Worldwide provides useful or interesting, please consider a contribution.
Perhaps you've noticed that Facebook shows you sponsored ads from left-leaning organizations if you're a Democrat and sponsored ads from right-leaning organizations if you're a Republican. A case can be made for this being a bad thing because it confirms our existing biases with posts that are slanted to fit our own point of view, and often are factually inaccurate.
Ads can be targeted this way because political organizations and businesses have so much information about each of us. With enough information, big data providers can create predictive algorithms that may be correct, but are often wrong.
With a PhD in mathematics from Harvard, Cathy O'Neil has a good understanding of big data. She's a mathematician, data scientist, and author. The Cleveland Public Library presented a webinar with O'Neil in September.
O'Neil has written several books, including Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. In September, she talked about algorithms that affect each of us and our society. From the ads you see on Facebook to the interest rate you'll get on a car loan, big data is there.
These systems all work on a scoring system and usually that system and the algorithm behind it are invisible. The scores cannot be appealed. So it's a score that will have an effect on your life, but you can't see the formula that created it and you may not even be aware that it's happening.
I had hoped to be able to share some of the audio from the Cleveland Public Library's presentation, but the library's agreement with Cathy O'Neil prohibited any use of the recording for any than archival purposes.
O'Neil says that the results are often wrong and deny people opportunities that shouldn't be denied. The companies that create the systems are not held accountable for errors, and the errors compound each other to create what she calls "feedback loops". These systems exist in bureaucracies that need to make difficult decisions: College admission, human resources departments, insurance companies, banks, police departments to determine where officers are deployed, and in the justice system to determine the length of prison sentences.
Currently health insurance companies cannot refuse insurance for pre-existing conditions, but that could change, and O'Neil says that insurance companies could even use big data to determine whether you might develop a costly condition in the future. But it's mainly guesswork disguised as intelligent systems. The systems fail, she says, because they simply predict the future based entirely on the past.
Political parties are big users of this information because they want to know which candidate people will vote for, and perhaps of more importance, who will donate to a campaign. O'Neil pointed out that whether or not you voted is public record. How you voted is secret, but whether you voted is not.
Political campaigns have also become adept at harvesting social media information that people have posted publicly, and many apps that people download and install also capture user data and sell it to businesses and political campaigns.
O'Neil offered examples of how big data is used to send tailored ads to specific groups -- ads that people outside the target audience will not see. For-profit "colleges" that often have high costs and low graduation rates send ads to people whose income is low, but who would qualify for federal financing. These are ads that more affluent people will never see.
She offered another example of tailored ads that are shown to a specific group of people on Facebook, but Facebook users whose profiles didn't fit the target audience didn't see them.
Although big-data information is currently being misused, it doesn't have to always be that way. O'Neil is hopeful that things will change. She compared big data to automobiles. In the early part of the 20th Century, as cars were being developed, people wanted to drive. Early cars had few safety features, but the technology was new and exciting. Automobile-related deaths skyrocketed, and eventually people saw the need to make cars safer.
In the middle and latter parts of the century, people demanded safety features -- better steering systems, disc brakes, seatbelt, airbags, crash testing, and more. Highway engineers designed roads that are safer.
Big data and artificial intelligence are exciting. The technologies can do a lot of good, but they can also do a lot of harm. The harm is mostly invisible, O'Neil says: People not getting a loan they should get or being passed over for a job they're qualified to do.
Today, we're just beginning to see the harm these technologies can do. Once we understand the problems, we can solve them. "We just have to do the work."
The webinar by the Cleveland library reminded me again how important libraries are, particularly now.
Libraries can't currently offer in-person meetings with hundreds of people who want to hear an author speak, so they've stepped up to present webinars instead. It's an improvement in some ways. When the Cleveland Public Library offered its webinar with Cathy O'Neil, the topic was one of considerable interest to me. It's unlikely, though, that I would have driven more than 240 miles to and from Cleveland for a one-hour program. Offering the event as a webinar made it possible to attend a session I would have missed even in a world without COVID-19.
The Cleveland Public Library's Director of Community Engagement, Aaron Mason says "one of the positive aspects of streaming events is we reach people who are out of town." Attendees came from as far away as Israel and Italy, both of which are considerably further from Cleveland than is Columbus.
As difficult as these times are, it's hard to imagine the situation without libraries.
It's also a good reason to belong to more than one library. In Ohio, for example, my membership in the Worthington Public Library also gives me access to more than 5 million resources from a consortium of 17 central Ohio library systems, but I also am a member of the Columbus Metropolitan Library, the Cleveland Public Library, and the Cincinnati Public Library.
Libraries in some areas have returned to near-normal operation, but others remain closed or are offering curbside pickup. Libraries that are partially open may offer concierge service, and many libraries make it possible to use their Wi-Fi from the parking lot as they continue to find ways to offer essential services.
Microsoft released Windows 10 version 20H2 in late September. It will arrive on your computer at some undetermined time, but it's easy enough to obtain if you want it. For the second year in a row, the fall update is tiny when compared to version 2004 (aka 20H1).
Click any of the small images for a full-size view. To dismiss the larger image, press ESC or tap outside the image.
Among the small changes: "System" is still an option in the old-style Control Panel, but it no longer opens the system section. Instead, it opens System in Settings. Microsoft has been working to retire the old Control Panel for the past five years and is getting closer. With each update, the Control Panel has fewer options.
It also appears that Microsoft is planning to standardize the version number on the code name. The first version's code name was Threshold 1, followed by Threshold 2. Then Redstone 1 through Redstone 5. Version 1903 had code name 19H1 and 1909 was 19H2. The May 2020 update, version 2004, had code name 20H1 and the October 2000 update uses 20H2 as both the version number and the code name. Apparently somebody at Microsoft thought that version 2004 made it appear that the version had been released 16 years ago and that 2009 (or 2010) would appear to be 11 or 10 years old. That problem would be eliminated with the first 2021 release, which would have been 2104 -- but that might have people wondering how they got an operating system from the next century.
So we have 20H2 instead and, next year, 21H1 and 21H2.
This should be an uncommonly stable release because there are so few new features. The update felt more like a standard monthly update. The computer had to be rebooted, but the entire process, including the download, took less than ten minutes. If you're not running version 2004, though, expect a long update process.
The Start Screen now has transparency, and if Windows notices that you use an application frequently, it will pin it to the Task Bar. How important these changes will be to you will depend on how much you use the Start Screen and whether you want the operating system to make decisions about what's on the Task Bar.
The new Chromium-based Microsoft Edge browser is built in. The new version has been available since January and now new Windows installations will have the Chromium Edge version instead of the old version.
So it's probably good to think of the just-released version of Windows as a service pack.
If the wait for 20H2 is anything like the wait for 20H1, you might not see the update until a few weeks before next year's update. Maybe you'd prefer to install the update at a time of your choosing. It's easy. Well, relatively easy. Right now it requires that you sign up for the Windows Insider program, but you can choose the Release Preview Channel instead of the more hazardous Beta Channel and the risky Dev Channel.
The Release Preview Channel is typically at the release candidate stage, a version that has been extensively tested and is sufficiently stable to be generally released. This is the one you want if you desire to have the update earlier rather than later.Here's how:
Then you'll need to reboot the computer, and return to settings after you're logged in.
If it's not present, you'll probably find it by selecting View Optional Updates. Computers running the 2004 (20H1) version should complete the process in a few minutes. Those running earlier versions will take considerably longer.Support for Office 2010 ended on 13 October 2020. If you're thinking this is no big deal, think again. The end of support means the end of security updates and bug fixes.
The best option for most people, especially if there are two or three or even five people in your family who need the Office applications on their computer, is Microsoft 365. That's the new name for Office 365. For about $100 per year, you can install the apps on up to six computers and any number of portable devices.
That's not $100 per month or $100 per quarter. It's $100 per year.
Microsoft also offers Office 2019 with a perpetual license, but it's expensive and it's limited to use by just one person. The software rental system (software as a service -- SAAS) is the future and there's nothing much anyone can do to get away from it.
But if you're adamantly opposed to SAAS, Microsoft will offer Office 2022 with a perpetual license. Buy it once and you can use it "forever" or until Microsoft decides to no longer support it. Microsoft 365 dings you for $100 per year, but the applications are always up to date.
Whether the new version will be called Office 2022 or Microsoft 2022 or something else is an open question. Office 2019 was released in 2018 and the new version of Office with a perpetual license will be released in 2021, so it's easy enough to presume that it will be called Office 2022.
If you're using Office 2013 or Office 2016, you may still have support. Mainstream support for Office 2013 ended in 2018, but extended support continues until 2023. Mainstream support for Office 2016 ended in mid-October, so no new features will be added, but extended support will continue until April 2023, which means that Microsoft will continue to fix bugs and address any security issues that arise between now and then.
The termination of extended support for Office 2010 means that users are on their own for security issues. No support of any kind will be offered for that version. Maybe this is a good time to remember that software is "licensed" and not purchased. This is true for applications that are offered online as well as those with a "perpetual" license. Support ends when the software publisher says it ends, not when users wish it would end.
Sometimes I get the crazy idea that there are scammers who intentionally create email messages with so many errors that no sentient human would fall for them. Why? There's no logical reason.
Sure, the scammers who claim to be dying heiresses, fraudulent bankers, or Nigerian princes who want nothing more than to give you millions of dollars depend on the greed and stupidity of their marks. But I don't see how social engineering scams benefit from assuming that the potential victim is stupid.
Click any of the small images for a full-size view. To dismiss the larger image, press ESC or tap outside the image.
It just doesn't make sense unless the scammers themselves are astonishingly stupid, and that is not an unlikely supposition.
How could anyone fail to see ten obvious clues: (1) The message claims to be from "cardusage.com" instead of "americanexpress.com", and the cardusage.com site doesn't even exist.
The message was sent to (2) multiple recipients, and it did not arrive in the email inbox that I use for messages from American Express.
The (3) padlock graphic is both poor quality and ridiculous.
The scammer addresses me as (4) "Dear Card Member" instead of by name. American Express knows my name. The (5) headline, "Your Important Notification" does not reflect a solid understanding of US English vernacular.
The singular (6) "Our record shows ..." is not common US usage.
Messages from large corporations rarely use (7) exclamation points.
Good design would place the (8) View Notifications button in the center of the screen and would provide a link to americanexpress.com, not to an obfuscated address.
A real message from American Express would have (9) functional Privacy Statement and Update Your Email links.
A message from American Express would not use both (10) e-mail and email in the same message, and would actually include account information to help me "recognize this as a customer care e-mail from American Express".
Could the scammers be any dumber? (Don't answer that; it was a rhetorical question.)
You probably use Zoom. These days, it's hard to find anyone who doesn't. Users can set up virtual backgrounds, and Zoom even offers a few for your use -- the Golden Gate bridge, the Earth from space, and several others.
My preference is a photo I took several years ago in the New York Public Library's reading room in the building at 5th Avenue on 42nd Street, but you can use any image.
With Halloween just around the corner, maybe you'd like to use an image with a Halloween theme. Apparently the folks at Tech Republic had the same idea and they're offering 13 images that you can download and use for free.
Previously they've created several other groups of photos, so if you want something different and there's no image on your computer that quite fits, take a look.
In 2000, I had a long discussion with Roadrunner's national security director, Mark Herrick. Roadrunner became TimeWarner and is now Spectrum after Charter Communications acquired the company in 2014.
I wrote "No matter how secure you make your computer, real security still depends on you. You can't depend on antivirus software to catch every new bug that somebody launches. Several days can elapse between the time the antivirus software companies first hear about a problem and when they identify the threat, find a solution for it, and make new definitions available for download."
What Herrick said then applies now: A good general rule is never to open any e-mail attachments (except for those that you have verified to be plain ASCII text) from anyone until you've confirmed that the apparent sender intended to send you the file. This applies particularly to word processor files, database files, and spreadsheet files, too, because most of these programs use Microsoft's Visual Basic for Applications (VBA) to provide "macro" capabilities.