OK, I’ll admit it: I don’t know if Windows Update, DriverUpdate, DriverAgent, or something else is involved, but I’m seeing the number of outdated and/or unneeded (duplicated) drivers mount up in my various Windows 10 installations. I just cleaned up half-a-dozen of my machines using the excellent Windows DriverStore Explorer, aka RAPR.exe, and found no machine with less than 7 unneeded or extraneous drivers. My Surface Pro 3, for some unfathomable reason, picked up more than 100 bogus drivers (nearly all of them from Intel for things like system devices, IDE ATA/ATAPI controllers, USB controllers, and so forth: I started out with over 90 Intel drivers in the store, and wound up with a total of 7 after clean-up).
That tells me it’s time for network/system admins to check client systems to see what kind of driver detritus they may have picked up, and perhaps to schedule some pruning or clean-up in the near term.
Before cleanup: 141 drivers; After cleanup: 37 drivers! (Surface Pro 3)
I’ve also been reading stories about Windows 10 stability issues in the past day or two (one about “Windows 10 glitches…” from James Kendrick, and another on “…Instability” from Mary Jo Foley, both of ZDNet). This is both amusing and vexing, in that as of last Friday’s (9/12) cumulative update, I’ve actually experienced a significant uptick in system stability (Reliability Monitor had been reporting anywhere from 1-3 application crash events daily per machine prior to this update, and lately, I’m down to less than 1 such mishap per machine for the entire 4 days that have now elapsed since the update was released). I can’t tell if that means I’m ahead of the curve, or playing in a different ballpark!
All in all, though, I’ve been able to function just fine using Windows 10 for both production and test machines without difficulty or major causes for concern. My wife and son have also made the switch to Windows 10 without voicing any (or many) issues or complaints — though the Edge browser definitely took the Boss by surprise, and was quickly replaced with IE on her rig. I guess it all depends on where the users are coming from, and the quality of tech support available to them! 😉
Something very interesting has happened to my Windows 10 production desktop since the last Cumulative Update was installed on that machine last week (I’m talking about KB3081455, dated 9/8/15). Prior to that update, I was getting corruption errors from both the sfc /scannow and the DISM ... /restorehealth commands on my production Win10 desktop machine. Following that update, no problems whatsoever, as this output from the second command illustrates:
Before the latest CU, I couldn’t find a source that would fix my reported corruption programs; afterwards, no problems reported!
I’m not able to tell whence this unexpected good news/result originates, because MS isn’t divulging details about its cumulative updates anymore (or right now — I suspect this will have to change, once enterprise adoptions start going up). What I can say, however, is that system stability on this formerly mildly flaky system has jumped appreciably since that update was applied. I’d love to know more about what MS did to make this happen, but that wish to know does not dampen my surprise and delight at the results of applying that update.
The question then becomes: “If you don’t know why things get better, does that reduce the resulting goodness?” For the answer is yes, because I like to understand what makes Windows tick and work (or work better or worse, as the case may sometimes be).
This morning, I read Adrian Kingsley-Hughes’ ZDnet story entitled “Windows 10 installer could be on your PC whether you want it or not” with some amusement. Then I asked myself the question “I wonder if it’s already on my Windows 8.1 test machine.” And sure enough, there it sits in the (hidden) $Windows.~BT directory, weighing in at a fairly hefty 4.77 GB:
MS has already downloaded the Win10 files to my 8.1 test partition, though so far I’ve declined to upgrade.
Kingsley expresses a certain amount of shock and outrage that MS would presume to clutter up its users’ disk drives with this much stuff without obtaining permission, and makes the point that for users subject to bandwidth caps or limits this could involve some unwelcome resource consumption. I can kind of see this one both ways, believing that users should be allowed to choose what they do and don’t download, but also understanding that MS wants to get those files out the door and onto the wire at off hours when bandwidth is more rather than less available to its (or its service providers’) servers (such as Akamai, a frequent host for large MS downloads). But my understanding is that MS makes this presumption only for users who’ve indicated an interest in a Windows 10 upgrade by signing up for “Get Windows 10,” so I don’t think it’s necessarily an unwarranted assumption on Microsoft’s part to believe that such users might want to upgrade at some point in time. Even so, I think he’s right to assert that MS should still ask for user permission before downloading almost 5GB of files onto somebody’s boot/sys disk.
This all raises the very interesting question of how many such downloads (like mine) are still sitting there, ready to be used but not yet put to work. Methinks it might be even bigger than the rumored 80-100 million installations of Windows 10 already running.
For years and years, I’ve subscribed to the notion that a clean Windows install — as opposed to an upgrade — can be a desirable thing. In cruising the Web, I see that my belief is still shared by the vast majority of Windows experts and mavens. However, a response to my most recent blog from Windows wizard and security wonk Russ Cooper now has me questioning or rather, researching and rethinking, this notion. Here’s the substance of the interchange between Russ and myself:
1. I post a promo to the blog post that reads: “Finally figures out how you can perform a clean install on a Windows 10 upgrade without supplying a key.”
2. Russ responds by saying that the old paradigms from previous versions of Windows don’t apply to this one, thereby implying that a clean install is no longer necessary.
3. I respond to Russ by asserting conventional wisdom that “the occasional clean install is an effective technique for de-gunking … Windows installations” while giving a nod to Jeff Duntemann (and co-author Joli Ballew) for their excellent books on this topic (run this Amazon search for more info). I show him this recent Reliability History graph for my production PC, upgraded in early September, by way of motivation:
All of the red Xs indicate somewhat serious errors, most originating from IE or hardware errors. This system is none too stable.
In addition, I have been unable to come up with a set of Windows 10 sources that will allow me to run the DISM ... /restorehealth command on this installation to a successful completion, which tells me that I do have lingering corruption issues with this particular install. Though reinstalling the raft of apps on this machine (PSI tells me I have 85 such elements on my PC, and I’m pretty sure I’ve got at least one or two dozen more installed that it doesn’t track) will take me the better part of a full day, I am still of the opinion that it will be worth the effort to restore my desktop to more normal and stable operation.
All of this said, Russ’s assertion has me questioning this conventional wisdom that a clean install is in fact an effective way to restore sanity and stability to an increasingly wonky Windows installation. That’s why I’d like to throw this topic open to the readership, and ask you collectively for your thoughts and experiences in the pros and cons of clean Windows 10 installations versus upgrades. I’m also reaching out to other Windows wizards I know to get their thinking on this topic and will follow up with more information as it becomes available.
It’s a great subject, though, and one worth digging into. If you’ve got something to say on this subject, or experiences to share, please do post a comment here, or reach out to me via e-mail through my contact page at EdTittel.com. Thanks!
[Note added 9/10/2015:]
In this morning’s edition of the Windows Secrets newsletter Fred Langa provides a definitive explanation of what’s going on with license data in Windows 10 upgrades. Here’s a quote from his piece entitled “How to clean-install a Windows 10 upgrade”:
First, you must— at least temporarily — upgrade your current Win7/8 system to Win10, the standard way. During this initial upgrade, Microsoft’s activation servers create and store a unique and permanent machine ID that’s based on your old Windows key plus the system’s hardware.
During the upgrade, Microsoft will also automatically issue you a new, generic Win10 product key. But it works only after your PC has been successfully upgraded to Win10 and activated. (This is how Microsoft intends to prevent piracy of the free Win10 upgrade.)
After your system has successfully completed an initial upgrade to Win10 and it has been registered with Microsoft’s activation servers, you then can wipe out the Win10 upgrade setup and perform a thorough, from-scratch, clean install.
At the end of that process, your PC will again check in with Microsoft. But because your system was previously whitelisted on the MS activation servers, your new clean-install setup will pass muster — recognized as 100 percent legitimate.
And FWIW, Fred remains a staunch member of the “clean install restores stable, reliable Windows operation club,” as do many other experts whose opinions I’ve researched since writing this blog post (including Ed Bott, Brink at TenForums.com, and lots of other folks).
I’d been wondering for some time how one could perform a clean install on an upgraded Windows 10 machine without providing a key. Thanks to the nice folks at the Windows Ten Forums, I found a pointer to this Windows 10 Web page entitled “Installing Windows 10 using the media creation tool,” which helps to shed some light on this apparent mystery. Given a copy of the Windows 10 ISO — readily available using the Media Creation tool available from the Download Windows 10 page — simply mount that ISO or use it to create an installable USB flash drive or DVD, then run setup.exe from the root directory of that installation environment (for an upgrade install). To perform a clean Windows 10 installation to replace an upgrade install, you must boot from the upgrade media itself (which works without requiring the key to be manually entered, for reasons I am still chasing down — common sense argues that either Win10 install reads the key from the install being upgraded, or “phones home” to an MS server to obtain that information automatically, and I’ve seen both of those possibilities hazarded by other veteran Windows-watchers like myself).
MS handles a clean (re)install of Windows 10 as a kind of upgrade and retains the upgrade key.
The secret is briefly mentioned at the very tail end of the afore-linked “Installing windows…” page, where the text explaining installation media includes this comment:
Both of these options allow you to upgrade the PC if it’s already running Windows 7, Windows 8, Windows 8.1, or Windows 10.
The two options mentioned are the ISO mount or install from UFD or DVD already mentioned, where a clean install from such an environment is apparently able to extract the key information from the running Windows 10 installation it is about to replace before overwriting (or adding to) the boot/sys volume on the PC that’s acting as the installation target. I’ve tried this now with an install onto an existing Windows partition, as well as an install that begins by repartitioning the boot/sys drive (and volumes) before actually copying files, and can confirm that both methods work.
Among other things, this means you can select a target disk for your new clean install to replace an existing Windows 10 install on a drive different from the one that housed the initial upgrade installation. This can be very handy when replacing the boot/sys drive, or when switching from an older, slower boot/sys drive to a newer, faster one. I’ve also been able to confirm that capability on my Lenovo laptops, both of which made the switch from an older, slower OCZ Agility3 boot/sys drive to a newer, faster Plextor mSATA boot\sys drive as part of the Windows 10 adoption process.
And for those who may be concerned about the old, tried-and-true method of capturing the Windows 10 key, and then reusing it during installation: that still works, too, as I confirmed early in my massive upgrade-to-10 sequence in late August and early September here at Chez Tittel. But it’s nice to know you aren’t forced to capture the key if you don’t want to (I always grab mine as a matter of course, however, being paranoid enough to realize that just because Windows 10 isn’t out to get me doesn’t mean I might not have to recover from a catastrophic crash some time in the future).
Cheers, and Happy Labor Day!
For months during the Technical and Insider Preview stages of the Win10 version track, I noticed that some of my PCs would switch from Private to Public network status following sleep (tablets, mostly — both Surface and Dell Venue Pro 11 proved occasionally susceptible). The easy fix on my home network which lacks an Active Directory server is to launch the HomeGroup applet in Control Panel: this immediately informs me that the network is Public and must be reset to Private before HomeGroup access is available. On an AD network, machines that fall prey to this gotcha must instead be reset from Public to Domain, as explained in this excellent tutorial on the TenForums.com website “How to Set Network Location to be Public or Private in Windows 10.”
Once you get into the properties for your adapter (Ethernet or Wi-Fi in most cases), you should turn this setting On.
For those to whom the HomeGroup method is not applicable, the quickest method to regain access to proper network status is to dig into the Network and Internet silo in the Settings app in Windows 10. Once into that silo, pick Ethernet for wired PCs and Wi-Fi for wireless ones, then make sure the setting for “Find devices and content” is turned on, as shown for Ethernet in the preceding screen cap. This will automatically confer the proper network status for those who have administrative access to their PCs. Those who lack such access will have to get some help from their Help Desk or local administrator, to restore their PCs to normal (and expected) functionality.
Here’s hoping that MS will be able to finally find and fix this occasional glitch in some forthcoming update to Windows 10, preferably sooner rather than later. In the interim, the preceding approach will help users get back on the network (and back to work) with minimal muss and fuss. Please consult the afore-linked TenForums tutorial for alternate fixes based on registry hacks, local security policy, or PowerShell (all of which work for both HomeGroup and Domain situations with varying degrees of success; Domain Admins will also want to investigate edits to domain security policy as well).
There was a long period in my checkered career as someone who works with and writes about Windows, where I simply couldn’t get by without a high-end screen capture tool. For most of that period, in fact, TechSmith’s outstanding SnagIt was my tool of choice for such things. But when Microsoft added the Windows Snipping tool to Windows Vista back in January, 2007, my appetite for high-end, high-dollar screen capture programs started to abate.
Sure, there are things you can do with such tools that no simple screencap utility can match — such as the ability to capture entire Web pages to an image including what’s off the screen as well as what’s on it, the ability to time the “snapshot” so that users can mouse over to active objects and cause them to appear for inclusion in screen captures, not to mention tools for cropping, annotating, rotating, and working on the resulting image files.
I still keep SnagIt around, but was amused that I hadn’t renewed my license for it when I fired it up the other day. That’s because for most of my needs — static screencaps to enliven my blog posts, UI snapshots to explain systems and programs, bits and pieces of photos or artwork to add visual interest to my prose, and so forth — I don’t need the kinds of advanced features that a high-end screencap utility can deliver. And such capabilities come with certain costs, above and beyond the purchase precise and license renewals: such programs go into the startup lineup, run constantly in the background, take over global hotkeys for activation and use, and so on.
That’s why I was tickled to see a “Delay” option show up on the (miniscule) menu bar for Snipping Tool in the RTM version of Windows 10. It looks like this:
On the menu bar, Delay now appears between New and Cancel.
Though this new capability might not sound like much (and only offers delays from 0 to 5 seconds, in whole second increments), it provides the ability to request a screen shot, and then offers users time to mouse over to a program or Web page, open a menu or pull-down list or whatever, and capture that dynamically produced information for posterity. Here’s a snap that shows what I’m talking about, taken from the GoSquared Windows 10 marketshare graphs I wrote about in my previous blog:
What you can now do with Snipping Tool is set a delay, then go off to a program or Web page, and capture pop-up info, like the stats read from the Win10 tracking curve shown here.
This is a great little feature, often useful, and provides yet one more reason why I’m thinking about not renewing my SnagIt license, though I do really like the program. However, in a delicious irony, that thinking is countered by my observation that Snipping Tool adamantly refuses to make a snapshot of itself: I had to use my extended trial version of SnagIt to capture the first of the two screen shots that appear in this very blog post. LMAO!
[Note added 9/2/15 11:15 AM CDT] OK, so by researching my E-commerce folder in e-mail I was able to ascertain that I’d upgraded my license with 1 year extended support for SnagIt on 8/26/2015. Five days after the cutoff, I went to the TechSmith website and found myself still able to qualify for update pricing on the software ($25 for another year, plus $13 for another year’s worth of upgrade access). I went ahead and upgraded because the tool is too handy to have around but for those who don’t need its more advanced features (I do, sometimes) or who don’t need to take snapshots of the Snipping Tool itself (ha! ha!), SnagIt may be overkill, however great a tool it happens to be…
On August 26, 2015, MS made it known that the Windows 10 installed count had surpassed 75 million. A quick look at the calendar shows me that 8/26 is exactly four weeks after 7/29 (the day Windows 10 hit RTM status, and Build 10240 became available to non-Insiders). A fascinating article on Supersite for Windows entitled “Windows 10 Momentum” digs into these numbers, and also exposed me to a new and very different, UK-based source for online user population data named GoSquared.com.
GoSquared reports a very different fraction of Windows traffic from Win10 as compared to NetMarketShare.
Last week, I blogged about an earlier 67-million-installed figure from unofficial sources that apparently equated to a fraction under 1 percent of all desktops it monitored through its tens of thousands of sites. GoSquared, on the other hand, reports some very different statistics — namely, instantaneous readings that vary from a high of just over 20% on August 23 to a low of under 1% right out of the starting gate on July 28 (as far back as its data goes). I believe these numbers reflect the breakdown by Windows version (and the line graphs underneath the circular indicator shown above show the breakdown across Windows 7, 8, and 10 for the entire time line covered) at any given point in time graphed. What’s not clear to me is the size of the population that’s being monitored and how that population is composed. GoSquared is a subscription service which makes the population self-selecting (those willing to sign up and pay for its services).
What’s fascinating is the idea that if Windows 10 is indeed roughly 10% of the Windows population active on the Internet at the time the graph was generated, the total number of Windows users online falls under 1 billion. This makes more sense than the 17 billion figure I calculated last week, predicated on the assumption that all Windows users were online and that the number of online Win10 users scales smoothly into the overall number of Windows users across all versions. Alas, that’s not really the way that it works, because at any given moment we’re measuring only a fraction of the global population, excluding the approximately 1/3 of the globe that’s in its prime sleeping hours at that time. Assuming instead that we can measure a maximum of 2/3s of Windows users at any given moment, that raises the global Windows population to something in the neighborhood of 1.12 billion. But given the wide range of fluctuation in the Windows 10 numbers available by tracking the graph on GoSquared across the past week (which runs in a range from over 8 percent to under 15 percent of active users at the time of monitoring) we get a sense that the population is pretty fluid and changeable in size.
What I do like about these numbers is that they show a more realistic notion of what the overall population of Windows users might be, and put Windows 10 in a more realistic position vis-à-vis the other common versions in use (though XP and Vista are not on this radar). It remains very interesting to watch Windows 10’s explosive growth, and to see how that curve continues to climb. To me, that makes upcoming quarterly and yearly milestones equally interesting, especially as indicative of overall trends.
[Note added 9/1/2015 AM:]
When I checked NetMarketShare for Windows 10 yesterday, it was still reading under 1% (0.93, to be more precise). As of this morning, however, that figure has zoomed to 5.21%. This is much more in line with what I thought it should be, at about half of what GoSquared is reporting. I’m sure there’s a story in how those numbers could jump by 560% in one day, but I’m also pretty sure we’re not going to learn the no-doubt “oops!”-related details involved. Here’s the screencap:
What’s truly amazing about this 5.21% number is that Windows 10 achieved in 28 days what it took Windows 8.1 almost one year to achieve (the 8.1 number was 5.92% by the end of July 2014, about one year after the RTM date). At this rate (which obviously cannot be sustained for too much longer, though it will be interesting to see how long this run lasts), Win10 could conceivably surpass Win8.1 some time in early November! We’ll see…
I’ve been aware of Windows Hello (the ability to use biometrics for login authentication and other access controls) in Windows 10 since MS first started talking about it late in 2014. Intel’s RealSense cameras, which exploit built-in 3-D recognition capabilities, have gotten the lion’s share of attention in promoting what Windows Hello can do. They are also new devices, and unlikely to be supported on anything but the latest and greatest of laptops, notebooks, and tablets. That’s why I was intrigued to read recently that Hello also works with any fingerprint readers that Windows can recognize. As it turns out, this means even my three-to-four-year-old Lenovo laptops support the Hello login capability in Windows 10.
I didn’t realize I could use Windows Hello to login with a fingerprint reader until I reinstalled Lenovo’s Fingerprint software on those laptops.
Once the right software to link the device to the Windows login facility is in place — I proffer the afore-depicted instance for the Lenovo ThinkPad laptops as an illustration — Windows Hello will quite cheerfully work with even fairly old devices like those installed in my X220T and T520 ThinkPads (purchased on 3/2/2012). This will produce the Windows Hello response when logging into a Windows 10 machine using the fingerprint reader, and helps speed that process while providing tighter integration with the OS than was available in Windows 7 or 8 versions. Certainly, for those devices that already include fingerprint readers, it’s worth taking advantage of this facility.
Every now and then, I like to drop in on the market share stats at NetMarketShare.com to see what’s up in the desktop world as monitored via that company’s observations of the Internet. When I checked this yesterday, I saw that for the first time in the three years or so since Windows 8 versions made their debut, that a single Windows 8 version (8.1) has surpassed the installed base for Windows XP, at 13.09 and 11.74 percent respectively.
On 8/25, the baseline for XP finally gave way to Windows 8.1
I’m not sure where Windows 10 figures into all of this, but with that OS in RTM status for a month now, and over 67 million installations leaked to Windows Central on 7/31, it still doesn’t show up on the NetMarketShare pie chart as a separate entry. The 8/26 readout shows Windows 10 at a mere 0.39% of respondents, so if all of the 67 million were on the Internet today, that puts the total PC population at 17 billion or thereabouts — a patently absurd figure, given that Internet World Stats puts the total number of Internet users worldwide at just over 3 billion.
It’s going to be interesting to watch Windows 10 emerge from the lowest ranks by market share, and begin to make its way up that ladder. I predict it will catch up with and pass Windows 8 versions much more quickly than they have come up that same path, but we’ll just have to wait and see how those numbers play out. I do have to question the leaked numbers reported in Windows Central, though, simply because the actual counts appear to be much lower as a fraction of the overall active user population measured through NetMarketShare. It’s also fascinating that the leading desktop, Windows 7, is now two full generations back from the current OS version, now out of mainstream support but with four-and-a-half years still to go on extended support (see the MS Lifecycle Fact Sheet for details).