21 Jan 2019
Just a quick technical note, for the benefit of future Googlers… A
few weeks ago, after reading some surprisingly positive comments about
Windows 10 from people I trust, I decided I’d give Microsoft the
benefit of the doubt and give it a shot on a spare laptop I had
The laptop in question was a Dell Latitude E6410, a business-oriented
machine from a decade ago. For its day, this thing was a bit of a beast:
dual graphics (chipset and discrete), 8GB RAM, 500GB SSD, Gigabit,
decent WiFi, all the good stuff. Oh, and it had a TrackPoint-ish
pointer in addition to a trackpad, which is still hard to find
today. By modern standards it has absolutely rubbish battery life,
even on a brand-new set of cells, but whatever, it spends most of its
life locked into a docking station anyway.
To be honest, it’s still a pretty good machine. The specs treadmill
started to slow down just after that machine was made, and it’s still
well within the minimum system requirements published by Microsoft for
Windows 10. And its Windows 7 license key (helpfully printed on the
sticker under the battery) was, allegedly, still good for a Windows 10
But I had a bastard of a time getting Windows installed on it, and
Dell is unhelpful because the machine is technically not supported to
run Win10 (naturally, they want you to buy a new machine, even if it
wouldn’t be much of an upgrade).
I’ll spare you the entire tale of woe, except that I got variety of
boot failures after trying to install Win10 from a USB stick. At
first, I couldn’t even get the machine to boot from the USB stick with
Microsoft’s installer. Once I got that working, and it appeared to
install Win10, I couldn’t get the machine to boot off of the HDD.
There were two big takeaways:
- You need a Windows machine to successfully create the install media.
There are lots of tutorials around, allegedly telling you how to
burn the Windows install ISO to a USB device, using Mac OS or Linux.
None of them worked for me; all of them produced unbootable
media. Didn’t matter if I used
dd, if I wiped the MBR, if I used
special-purpose bootable-USB tools… the machine wouldn’t boot from
them. But after borrowing a friend’s Windows machine for an hour, I
was able to make one that worked using Rufus 3.4.
- The installer senses how you boot from the install media, in terms
of BIOS/MBR or EFI/GPT, and uses that during the install process as
it writes to the hard drive. This took me a long time to notice.
The USB installer, once burned correctly, is bootable either with
“legacy” BIOS from the device’s MBR, or “modern” EFI boot, with GPT.
For reasons that aren’t clear to me even still, the E6410 wanted to
boot from the USB installer using BIOS/MBR, but needed the OS on the
internal hard drive to be configured for EFI, or else it wouldn’t
boot. But you can’t actually do this using the installer.
Instead, I had to use the F10 boot order menu to force the machine to
do an EFI boot from the USB stick, then install Windows. This
So, the process I used was:
Download the Win10 ISO from Microsoft (I used
Burn the ISO to an appropriately-sized USB drive with Rufus, on a
Power down the E6410, plug in the USB drive, and power it up. Hit
F10 quickly to reach the boot-device menu (not the full BIOS menu,
just the boot selector is fine).
Choose EFI boot, and then from under the EFI options, choose the
USB stick. Don’t choose the tempting option of the USB stick that’s
just on the root level of the boot device tree. If you choose that
one, the computer will boot using BIOS/MBR, resulting in a MBR install
that won’t boot correctly.
Allow the machine to boot into the Windows installer. At the
partitioning screen, delete all partitions (I had several, because my
machine was being abused as a server for a while) and install to the
HDD device. The Installer will create a single new partition spanning
the whole device. I suppose at this point you could also select a
pre-created partition if you were going to dual boot, but I didn’t
Let the installer do its job, reboot, and pull the USB stick out
promptly so it will boot from the HDD. Now enjoy the dulcet tones of
Cortana, talking you through the rest of the install process.
I’m reserving judgement on Windows 10 until I have a chance to really
try it out, but so far I can’t say I hate it. I don’t love it,
and the hard sell on the Edge browser (“please use it! It’s really
nice! Really!”) is a bit annoying, and some of the telemetry privacy
options are ominious as hell, but it seems to run fine without
slipstreaming drivers or any of the other crap I’ve learned to hate
Read more »
21 Jan 2019
I’m sure I’m not the only one who has watched the sad, embarrassing
decline of “The Learning Channel”, as it moved from a sort of 24-hour
feed of videos that a substitute teacher would show, to plumbing the
depths of the public’s appetite for softcore exploitation porn,
“educational only in the way that a bacon double cheeseburger can be
described as part of a healthy, balanced diet” according to
If you’re morbidly curious what happened, Audrey Watters has a great
article (written back in 2015 but evergreen): “What Happened to
Educational Television: The Story of ‘The Learning Channel’”.
It’s an interesting case study of a good idea leading to an
organization that decides to perpetuate itself at the expense of its
mission. What began as a well-meaning experiment in satellite
delivery of teacher training to inaccessible Appalachian regions in
the mid-1970s, later billed as a way of democratizing
education, became merely a vehicle for a for-profit company to sell
advertising by the late 1990s.
While today we have the Internet – probably a better vehicle for
educational content than one-way TV broadcasts – it’s painful to
consider what opportunities for actual educational programming were
wasted. As someone who grew up in a rural area (with cable TV, but
painfully slow Internet to this day), I was theoretically the target
audience; it would have been pretty neat if some of The Learning
Channel’s original instructional programming – cooking shows that
actually taught you how to cook, college credit courses on business
management, even shows covering basic computer programming – had
remained on the air.
It’s entirely possible that the public would have been better served
if the Appalachian Community Service Network, onetime operator of The
Learning Channel, had just closed its doors when they could no longer
succeed as an independent non-profit, rather than selling out. By
doing so, they created PR cover for growing cable companies desperate
to convince the public of the value of (frequently subsidized)
commercial Cable TV, and they likely suppressed the emergence of other
competitors in the bandwidth-limited, channelized world of cable.
A cautionary tale, if ever you see a public-benefit organization
beginning to “pivot” in order to stay afloat.
Via The History of Teaching Machines, via Hacker News.
Read more »
16 Jan 2019
Malcolm Gladwell has weighed in on, of all things, cannabis
legalization. And people are not impressed.
Not that anyone should really be surprised that Gladwell is a
shameless abuser of statistics; he basically said so
himself to Michael Lewis back in 2017. But this one is
especially egregious, and doesn’t even have an entertaining just-so
story attached to it.
This Metafilter thread has some good jabs, with the
knockout punch delivered by “Mayor West”, digging into the data to
show that Gladwell seems to have cherrypicked his start and end
years in order to make Washington State seem like it’s on the road to
ruin since legalizing cannabis.
Congratulations, Mr. Gladwell, you’ve learned that favored trick of
the desperate undergraduate: how to cherry-pick your data, by
conveniently measuring change between a low statistical outlier on
one end and a high outlier on the other end.
If you skipped reading Gladwell’s article, the most headline-grabbing
part of his fearmongering is an implied relationship between cannabis
legalization and an apparent increase in homicides in Washington
state, between 2013 and 2017. I say ‘apparent’ increase, because 2013
was a low outlier, with 160 homicides statewide; 2017 was a high
outlier, with 266. This makes it seem as if the homicide rate is on a
terrible upward trajectory. But if he had looked at 2012 to
2016—the same window width, just shifted back one year—he would
instead have seen a decrease, because 2012 had 217 deaths while 2016
It could be an example straight out of How to Lie With
Statistics, although that book at least has the honest goal of
enlightening its readers, not decieving them.
But why would Gladwell bother to pull his trademark statistical
shenanigans about cannabis legalization in particular? That’s where
things get… weird.
As it turns out, Gladwell seems to have a bit of a history in this
department: not anti-cannabis, but carrying water for industries of
ill repute generally. Enough so the “SHAME Project” even has a
profile of him. (Although the SHAME Project probably
shouldn’t be regarded as a totally agenda-free source, it points to
primary source materials which can be evaluated on their own merit.)
Most of it focuses on Gladwell’s apparent friendliness with tobacco
companies in the 90s and early 2000s, but he also seems to have a
thing for defending price-gouging pharmaceutical companies
and the Credit Crunch-era financial industry.
Cigarettes, overpriced drugs, mortgage-backed securities—it’s
looking a bit like a pattern: he sure does love to be on the wrong
side of history. Perhaps we should consider that Gladwell has a bit
of a thing for defending the indefensible? That, far from being a
statistical hitman available to the highest bidder, he’s actually
something much more common and far less interesting? Specifically:
that he’s just a troll? Nothing, really, but a faux-intellectual
shit-stirrer willing to say ridiculous stuff for clicks, enamored with
himself as devil’s advocate? Mistaking, in the way that trolls do,
having everyone shouting at them for bravery and originality?
I’m not saying that he is, just—in the spirit of his latest
Read more »