The problem Broadcom has in comparison with other SoC makers is they never open source their drivers except under extreme duress. The practical impact is that you can never fix problems in the firmware and you can never upgrade the kernel. It looks like they're building this thing on top of the Marvell Armada 385.

I don't know of any 802.11ac WiFi radios with open firmware, but the Qualcomm 9880 at least has an open driver. It looks like this Turris router will have Qualcomm radios.

tldr; You hipsters don’t appreciate the effort that went into good design. Hint: Flat Design is not an example of good design.

Reductio ad absurdum. Norman and Tog did not say all of it was impossible to read. They said it was difficult or impossible, and it is true that it is difficult. Some of Apple’s advertising images even featured skinny white text on a white background. For all of computing history, we’ve had gradually increasing resolutions and easier to read fonts, and then suddenly, bam. Everybody is doing skinny white text on a purple gradient or something.

Great. So where is the science? Norman and Tog actually did studies. They got people who were not infatuated with their vision to test the interface and see if it worked. They watched how people used the device, and they iterated. Where are the studies showing that this new way is better? Who is more productive? And why is the easier to use way not the default?

Bad example. The mobile devices I know still have a type of scroll bar: The contacts app would be unusable otherwise.

On the web browser, there is no scroll bar. There might be a scroll indicator, but there’s nothing that I can grab. I just swipe, and swipe, and sometimes I give up and just not consume the content that I want. This is bad design in practice.

Also, I am not putting content all the way to the edge of the screen. I have like 200 ppi in that blank space. I can afford a few pixels to at least show a persistent indicator of where I am in the content. Not like when I was trying to find out what Apple Mail on MacOS X was doing, and the activity window had no indication that I could scroll down to see what the other connections were doing.

Sure, it’s easy to criticize. Making a good design would be a better proof. But it shouldn’t be so easy to just point at what they did in the past, and say, that was better for this reason. You shouldn’t have thrown that away.

Forget autonomous car development, Toyota is woefully behind in computer-controlled car development. Random relevant article: Toyota's killer firmware: Bad design and its consequences.

What are you talking about? If you are in a construction zone and the lane changes suddenly, I think you will find yourself in a traffic jam. For that matter, I don’t think you will be driving “60mph+” in a construction zone; I think you will already be in a traffic jam.

That's because DD-WRT is not "full open source, etc." It's open source Linux, but closed source device drivers. It comes from Linus's "pragmatic" desire for Linux to be used, with no interest in the political reasons for the GPL.

I don't know the process at OpenWRT exactly, but they tend to use open source drivers more than DD-WRT does. On the minus side, this means it supports much fewer devices. On the plus side, this means "supported" devices really are supported and have updates available to them, while DD-WRT is just throw a firmware at a device and never give it a feature or security update ever again even if there are open-source drivers for it.

4k seems like it should be obviously useless. 2k already makes it really hard to see individual pixels, and more pixels take more power to push.

Where 4k makes sense is for VR displays. The Samsung Gear VR and the Google Cardboard use the phone for the display. And the accelerometer. Those do need higher resolution to look good.

In my opinion, OpenWRT is better than DD-WRT because OpenWRT is under pretty active development and has features that matter for making a better Internet.

DD-WRT is very difficult to compile, so in practice when a device comes out, you have one guy making a firmware stuffed with like 4 hotspots and 4 VPNs and 2 VoIP switches and DynDNS, or as many of those things as he can fit, and there’s no space for your own programs on the router. IPv6 is not a top priority at DD-WRT. And then nobody makes a new firmware for that device ever again, no matter how many security holes appear over the years.

In contrast, the latest OpenWRT comes with FQ-CoDel, IPv6, and DNSSEC. The default web-based administration these days is not bad, and the package system allows you to add interesting stuff, if your device has enough space. The Kconfig build system and the plain text configuration files make customization pretty easy.

The main downside is that OpenWRT is more picky about hardware. For DD-WRT, you have an ancient WRT54G, that’s fine, just install an equally ancient firmware. Ignore the problems; everybody else ignores the problems. Current releases of OpenWRT insist on a device that can run a modern kernel, with at least 4MB of flash and 32MB of RAM.

Right, that’s why I said, “develop” something better, and not just look into alternatives. If you want to leak secrets to journalists right now, or if you want to sign your distribution releases right now, then OpenPGP is the best alternative. If you want to communicate securely with family, probably S/MIME is better, because it’s way more compatible and less hassle to use. Already, security is bifurcated into incompatible solutions.

For practical use, probably the biggest improvement in people’s security has been Gmail. Sure, it’s entirely the plaything of the Borg, and vulnerable to the rubber stamp of FISA, but the actual connection to the server is protected by TLS, with pinned certificates in the major clients. I don’t remember who it was, but somebody said switching to Gmail has been the biggest improvement in actual email security, because they have good technology and a legal team that actually puts up a fight.

The biggest problem with OpenPGP is that it doesn’t protect the metadata. It has to be backwards-compatible with the existing mail system, and that mail system needs to be replaced. The most interesting alternative that I have noticed is Dark Mail, but that is only an incredibly complicated RFC right now.

No, rjh is correct. The problem with cryptography is that it is incredibly tricky. If you don’t do it just right, then you compromise your security. If you get it really wrong, then it’s as good as not having cryptography at all. GnuPG has to be complicated to be compatible with the design-by-committee OpenPGP standards, including its clunky manual key system.

rhj is probably also correct that OpenPGP gives the best security for email. My problem is that no security system is useful if nobody uses it. OpenPGP is used a lot for verifying that open source contributions come from particular developers, so their public communications can be authenticated. So far, so boring. It’s not used at all for communicating with friends and family. S/MIME at least has transparent key distribution, but that is not used by anybody, either.

The most interesting alternative to me is Dark Mail, but so far that is nothing but an incredibly complicated RFC. The proposed architecture looks interesting, though, so I want it to succeed.

The fundamental problem is that SMTP was not designed for security, and there's no provision to change over everybody at some point.

Also, privacy adds significant usability problems. You need to generate and copy your private key manually, instead of having your machines provision it among themselves. And privacy means webmail providers can't monetize the contents of your email, and you can't search through past emails efficiently.

When activists like Moxie Marlinspike are calling for the end of PGP, it's probably time to look into alternatives.

PGP's problems are endemic to its design. It cannot be fixed, and increased adoption won't help.

But RAM? Hell, 4Gb is fine for basically any business task unless it's a server. ... The clients, however, need next to nothing.

Coming from the 640k days, I still react with astonishment when somebody calls 4 million kB "next to nothing."

Because you can’t.

The computer I built in 2003 has 1.5GB of RAM and 200GB of storage. This is enough to run Windows 7, though not Windows 8 or 10 because it doesn’t have SSE2 nor NX. With Linux and BSD and various strange options, I have endless choices of what OS I get to run on it.

The typical phone of 2007 has less than 128MB of RAM. You are not running a modern OS on less than 128MB of RAM. Furthermore, while the PC was relatively open, most phones have depressingly closed drivers. Just look at CyanogenMod releases: No gyroscope, or no GPS, or no sound, or very likely most of the hardware not working. That’s if you can get it to boot, or even accept and install a third-party ROM.

Just forget running anything other than iOS on iPhones. Not gonna happen.

First, he complained about the download. I anticipated this problem, downloaded the ISO on Windows 7 with Microsoft’s stupid downloading program, and burned a DVD/USB. Problem solved. Also, you can buy Windows 10 OEM media in stores.

Then, he complained about the updater not having a clean install option. It’s not obvious, but there’s an option somewhere in the installer to “Keep nothing.” This does a clean install.

He did not complain about tying the Windows account to a Microsoft account. It’s possible to make a local account not connected to a Live.com, and it’s more obvious how to do so than in Windows 8.

Then, he complained about the hybrid Start menu. That can be resized.

Other than that, I guess the review was okay. I liked the part about the Hi-DPI experience.