I was curious, and last night I priced out a basic "stick pfSense on me" box with reasonable quality components. With the exception of Realtek NICs instead of Intel—which might be a problem as you go past 150Mbps, Realtek NICs don't have a terribly glorious reputation—you can assemble a Mini-ITX based system with mirrored drives for $360. Intel used to make some dual-NIC "corporate workstation" boards that worked really well, especially if you ponied up for a better CPU that supported vPro, so you could do remote IPMI console. Unfortunately, Intel got out of the motherboard business.

I haven't tried any of this equipment, so it may actually suck, but here's the bill of materials I came up with for "so you want to build your own router with commodity parts". Obviously, you could go with server-grade parts or with a ready-built box of various flavors too...

• BIOSTAR Hi-Fi B85N Mini-ITX motherboard
• 2x4GB DDR3 1600 (PC3 12800) DIMMs
• Cooler Master Elite 110 RC-110-KKN2 case
• COOLMAX CX-400B ATX power supply (but I'd spend a little extra on an Antec VP450 myself)
• Intel Celeron G1840 CPU (dual core 2.8GHz Haswell)
• Two Western Digital Blue WD2500AAKX 250GB disks

Something like that should be able to handle any reasonable real-world home network needs. RAM is pretty cheap; you could probably do fine with 4GB. SSDs are all the rage, but spinning rust is cheaper and disk speed isn't really a big factor for a router.

However, as a matter of common-sense security, I'd recommend keeping any such box limited to being a router/firewall. Sure, run DHCP and DNS services on it... perhaps OpenVPN... but resist the temptation to load it up with other services. You'll just bog down the performance and increase the potential attack surface, especially if you accidentally misconfigure the firewall.

Indeed. I only have 150Mbps service, and yet the cable guys are constantly amazed that I can achieve that throughput (or more, they don't hard cap it) consistently. The reason is that my router is a home-built UNIX PC with two Intel NICs and the cheapest Intel Celeron processor you can buy—which is massive overkill for a home router.

Also, a tool that is your best friend when installing new wall jacks that you're wiring down to the basement: There is a special drill bit that has a long (4+ foot) flexible shaft and an auger tip with a hole in it. It comes with a handhold that lets you insert it into the hole you made in the wall, twist it down, and have it bite into the base of the wall. It drills through into the basement, where you then attach the small wire basket that comes with it to the hole. Push the wire into the basket and tug to tighten it. Then, go upstairs and pull the drill back out, engaging reverse drive if needed (the basket has a swivel on it for this purpose). The wire comes with it. No additional fish tape needed. It's in the electrical-tools section of any big-box hardware store.

I would add:

• Do NOT buy your jacks at Home Depot, unless your local HD is still stocking Leviton parts. The Home Depot house brand "Commercial Electric/CETech" networking components are total crap. I have never had a jack I've punched down fail to work... until I bought the CETech parts and had all of them fail to work, or break in the process of assembling them. Leviton parts are good but expensive. I've had good luck with Shaxon parts from Amazon, but they're not quite cross-compatible with Leviton parts/faceplates.
• Invest in a decent punchdown tool. You CAN use the little plastic tool that comes with the parts you order at Home Depot. A good punchdown tool will work much better, and lets you use bulk bags of parts that are cheaper.
• Invest in a data cable stripper. This is a little tool that you squeeze open, slip over the Cat5 cable, and twist to cut the outer layer of insulation without nicking the wires. You CAN do this by hand with a pocketknife, but if you're wiring up a whole house, you will save so much time and aggravation that the slight cost of this tool is absolutely worth it.
• If you find you're going to be running more than one or two wires along a particular beam, buy some commercial-style "J hooks" for data cable. You can special-order them at Home Depot. Cable staples work, but they're unwieldy for running many wires in parallel. J-hooks make expansion easy.
• Buy plenty of Velcro. When making data cables neat, Velcro is your friend. Remember, the fuzzy side goes toward the cables; the hook side is more abrasive.
• Label your cables. You can use a regular label maker. If you have access to a cable label maker—one that makes labels that wrap around the wire—that's even better, but they're usually really expensive. Remember that eventually someone else will own your house, use labels they'll understand: "front bedroom", not "Jane's room".
• If you're crimping 8P8C Modular ("RJ-45") ends on your cables, invest in the little rubber strain relief boots that slide over the cable BEFORE you crimp on the end. They make for a better looking job and they protect the cable. You can get a lifetime supply bag of 'em on Amazon in your choice of colors pretty cheap.
• Don't forget that national and local electrical codes apply to data cable wiring. Check your local codes. They usually specify things for a reason. In particular, obey what they say about running data wiring near power wiring, and about sealing up any holes you drill that go between floors with an appropriate fire retardant caulk or foam.

Ah, but with the new reality of ownership, we are not the client. We are the product. The advertisers are the clients.

One wonders if the clients will still buy a product that ceases to be profitable once the product delivery system is broken in the name of progress.

I don't really see how the new design truly benefits the advertisers, other than giving DICE's ad execs newer, bigger, louder ad spaces to tout. The fact that it reduces the audience for those ads doesn't seem to enter into the equation.

The first thing you should ask the design team:

Do you understand that—while most of your readers don't really care about the design—those who do care are the sort of people who take one look at a site that mixes multiple sans-serif fonts in its interface and immediately have a visceral, intensively negative reaction that strips you of all design credibility and makes them look away just as surely as if they'd seen goatse?

The excessive whitespace and awkward layout compounds the problem. There is a "right" amount of line spacing; it's a very well understood thing in the publishing industry. Any nerd understands that nerds tend to favor information density over "right" line-spacing. Therefore, incorporating excessive line spacing on a News For Nerds website simply shouts "incompetence" to the world.

Seriously, the visual aspect of the redesign is as if your web designers saw the style and popularity of Apple's Jonathan Ives' school of design (but not the typical Slashdot readers' reaction to such), and then hired the sort of people whose work is featured on Cake Wrecks to implement their own version.

Someone needs to be put in charge of this effort who has the understanding and the authority to say "our 'audience' does not want a custom, trendy font; they know that webfonts have to be loaded and will slow things down. They want whatever they've chosen as the body font in their browser's options."

Me, I'd start with the idea that Slashdot has to be minimally usable even if no CSS or JavaScript is loaded. It needs to have well-structured HTML that is content-based, not design-based. Then you can start layering design on top of that, in ways that allow for customization. If your guys want to have a CSS option that looks like a marketing MBA's WordPress wet dream (Beta), that's fine, as long as it's not the only choice—or, indeed, the default choice—and doesn't drive the bones of the site.

Scrap the Beta. It's a dead end. Start over after you draw up realistic specifications that include honest user research, not just advertising optimization hacks.

You thought computer viruses were bad, wait until you have to deal with computer diabetes...

I agree. For an inkjet all-in-one, I'd recommend the Canon MX882 or its follow-on models. The printer is fast and high-quality, and has a bypass input slot and a duplexer. The scanner is as good as any standalone consumer-grade photo scanner you can find nowadays—which is not a given in the multifunction machines—and it has an automatic document feeder with duplexer. It has wired and WiFi networking, and it generally just works.

My place of work insists that I have a Brother MFC-J5910DW as a home-office printer. Next to the Canon, it's a piece of crap. The print quality is atrocious. The paper tray was designed by a sadist. It jams far too often—I don't think I've ever had a paper jam in the Canon. While it can duplex print, the ADF cannot duplex scan. Scans are washed out with poor color fidelity. The front-panel interface has a strong affinity for fax mode, even when there's no phone line connected: if the thing's been idle for any period of time, it's in fax mode the next time you try to use it... and if you push a different mode button to wake it up, it give you error beeps until it finishes waking up and starting fax mode. At least once every 48 hours, it startles you by entering a loud self-cleaning cycle that purges a little more ink from the system.

My first laser printer had the AppleTalk name "%%PostScript Error" for much the same reason.

Given that Thunderbolt carries not only the equivalent of a PCIe x4 connection, but also a DisplayPort connection... and that the new Mac Pro has six Thunderbolt 2 connections... it's obvious that the HDMI port is there as a convenience for those who would otherwise bitch about having to buy a Mini DisplayPort to DisplayPort/DVI/HDMI/VGA cable. Since Apple has advertised the unit as supporting three 4K displays out of the box, obviously at least three of those Thunderbolt 2 ports can be used for DisplayPort video.

Yes, Time Warner's top-tier 50Mbps is priced beyond the reach of most customers. At $100/month, it's a luxury.

But there's another issue. Right now, the biggest reason to get big bandwidth at home is to support multiple users with diverse interests. There are a lot of potential uses where the upstream bandwidth just isn't there to justify a fatter pipe. Netflix may have a content-delivery network to support higher speeds... but TWC hasn't signed on for it. For most people who work from home, their employer doesn't have enough bandwidth to make a bigger pipe useful. If your employer has only a 45Mbps connection shared by all business needs, you're going to saturate any remaining bandwidth with a 50Mbps connection at home; why would you need gigabit to work from home? In that scenario, 50Mbps is only useful so the kids can Netflix without crimping your VPN speeds... And to get the higher return-path speeds that come with it.

Netflix and its rivals don't come close to using 50Mbps bandwidth per stream. They usually stream closer to 3Mbps. If they offered hire quality streams, or if there was a lot of 4K-resolution content out there, there'd be more demand.

The uses for ultra wideband bandwidth will come, but they're not here yet for most people... And especially not at those prices.

I had a better solution for long-distance sales calls for a while before Do Not Call. I worked for a telecom company. When the marketer would call and assure me that he could save me money and beat my current rates, I could truthfully reply "Well, I work for XYZ Telecom, and so I get free long distance. So how much are you willing to pay me to use your service?" This would reliably end the call...

Oh, and unless you have a minimum of three able-bodied people on site at all times, some sort of server lift is a must. Preferably one that is electric, has no fluids to leak, and has a shelf that slides to ease insertion and removal of servers. They are expensive, but they turn three-man jobs into one-man jobs... And prevent worker's-comp cases.

A label maker designed to make cable labels. That means it's designed to use wide tape and print on it sideways, and it will take flexible vinyl tape. The best ones print on "self-laminating" labels that are opaque where the label is printed, but clear at the end, so the overlap protects the printing.

At least one, and preferably two, USB 2.0 to IDE/SATA converters. There are plenty of ways in which you can find yourself with a bare drive you need data from, and no good way to plug it in. Also, in a pinch, a bare CD-ROM can become an external drive for a server with no drive. These things are cheap, and when you need one, you REALLY need one right now.