By the way, the reason I used an SDHC instead of a USB stick was mainly because the SD slot fits inside the laptop body, while a USB drive would have stuck out and I'd have had to remove it every time I put the laptop into its bag, plus I often needed both USB slots. For a desktop box the USB would win, or an internal SSD hanging off a SATA port, which most motherboards seem to have spares of.

A hybrid drive builds the caching function into the drive, as opposed to having the OS keep track of disk-vs-flash-vs-RAM and doing its own caching. So yeah, it's possible that if the hybrid drive is badly designed, a failure in the flash could result in losing access to the data on the spinning drive as well; you'd hope they'd avoid that.

And yeah, of course you need backups. A bullet-proof hybrid drive design or OS caching algorithm isn't going to stop you from scribbling the wrong stuff onto the drive.

30 years ago, disk failures were a serious problem. 20 years ago they were still a problem. By 10 years ago, Moore's Law was cranking disk price/performance so fast that your disk had a 99.99% chance of being boringly small and obsolete long before it failed, and it's still the same. Yes, that 8GB SSD cache for your 1TB drive may fail in 5 years, but 5 years from now you'll have junked that wimpy drive and replaced it with a cheaper 128GB cache and 8TB drive.

Yes, SSDs are faster and more expensive. The obvious approach is to have an SSD that's big enough for the stuff you use all the time, and a spinny drive to handle the bulk storage. Hybrid drives (which were the original topic here) have a spinny drive with a small flash cache, so you can do most of your I/O from flash and it can keep the mechanical drive updated at its leisure, so your CPU isn't stuck waiting for rotation latency.

But you could do just as well by having an operating system doing the caching in separate storage; hybrid drives are mainly a win for people with lame OS's. (Oh, wait, most OS's are a bit lame about this.) MS's Readyboost stuff was supposed to be a big win; I've only tried it using an SDHC flash card, which might be on a slow bus, instead of an SSD that probably has a faster connection, but I haven't seen a significant speedup when I use it.

Guest networks are socially important, and friendly, no matter how much the cable modem companies dislike sharing. Unfortunately the WiFi encryption standards assume that if you want privacy you also want to limit who can use the network, so if you want encryption on a guest network you need to resort to approaches like your SSID "passwordispassword".

For a long time, most of my neighbors and I ran unencrypted networks, so if my DSL was out for a day, I could borrow a nearby "linksys". (When 802.11g came out, most of those routers got set up with passwords, so I'd have to go over to Starbucks if my DSL was down.) The only problem I ever had was when one of my neighbors got a virus and her PC used my Wifi to spam. Fortunately I have a friendly ISP, so they just called me and said "BTW, we've blocked half a million spams from your line today, can you see if it's your PC or your Wifi?" One of my friends considered it a civic benefit to run Wifi so his neighbors' teenage kids could have uncensored Internet if they wanted - was fine until the kid discovered file-sharing and flooded the airwaves.

My guess is the opposite - you'll probably notice the difference in resolution more for black on white text than for photos.

My first laser printer was an Imagen wet-process printer (based on the printer mechanism of a Canon copier engine.) 240dpi, with a liquid toner that was pretty much carbon black dissolved in kerosene. It didn't need anti-aliasing to smooth out jaggies :-)

Until we see a lot of 4K TVs out there, we aren't going to get cheap 4K PC monitors. As far as high enough quality for work goes, color resolution doesn't matter to me; pixel count and size do. (And as for responsiveness for gaming goes, as long as the screen isn't actually flickering, Nethack isn't bothered by 50ms latency.)

Having finally acquired an HDTV this year, I've found that it's nice to have a monitor that is in the same aspect ratio as movies so I don't need to black-box or have the sides cut off, but there aren't that many movies that really need the extra resolution as opposed to the correct aspect ratio. Regular TV programming cares even less about resolution; either the writing's good or it's not, and HDTV won't fix bad writing, and talking heads are either saying something sensible or blathering. Sports and nature programming are exceptions to that - being able to see a moving hockey puck or tennis ball better helps a lot, and wildlife pictures on National Geographic do look a lot better if you've got hi-def.

Yeah, for a 24" monitor I'd be happy to have 300dpi, because I want to be able to read my computer as easily as I read paper. 200x200 is a standard-mode Group 3 Fax machine resolution, and while it's a lot less ugly than the 100x200 mode, it's still ugly (though of course your monitor doesn't have the vertical sloppiness that mechanically-driven paper printer rollers have and the fuzzy pixels of thermal paper.)

If you're watching movies, or even still pictures, resolution doesn't matter as much, because your eyes will fix that stuff. But when I'm using a computer monitor, I usually want to read lots of text. For that you need actual pixels, even if most of them are running in 1-bit color, or 4-bit color so you can do better navigation clues.

At $DAYJOB, we're always getting lots of company propaganda about "Don't Text While Driving". Not a problem for me - I can't do texting without my reading glasses, and I can't drive with my reading glasses on. And the [expletive-deleted] HTC-flavored Android texting application may look pretty, but it doesn't let me change the font size, and doesn't let me do the two-finger stretch thing, and doesn't even do a decent job of adapting to landscape-more screen orientation. Yes, the Google-provided keyboard thing with it has a little button you can press to do speech-to-text, if you've got your reading glasses on so you can not only press the [expletive-deleted] little button and then read the results it got back to be sure they're not hopelessly garbled, but that doesn't really help.

Yes, a higher resolution screen helps make the text clearer. But what I really want is a phone that can be controlled entirely by voice for most common applications, like texting. The Bluetooth connection to my car supports voice dialing; why can't the phone support at least voice-controlled reception of text messages?

Skills don't show up by magic; they're something you learn. You need both talent and practice to get them, and if you've got less talent you generally need more practice (or better teaching materials to practice with.)

If Apple's actions are being described correctly, that's time that clearly belongs to be on the clock.

Disclaimer: At $DAYJOB, I work on managed security services using Sourcefire, but this is my own personal commentary, not that of my employer.

Sourcefire's primary product line takes Snort, wraps it in hardware appliances, and adds a lot of management tools that you can use in an enterprise or managed services environment. This past year, they've added a firewall capability to compete with Palo Alto* and the UTM vendors like Fortinet - in addition to basic firewall support they've got application identification, so you can do things like allow users to read Facebook but block Facebook games, and you can also do things like URL censorship and known-bad-site blacklisting. They've also been buying up other companies like ClamAV and Immunet, so they've got feeds of malware site identification, and are starting to integrate that with the firewall/IDS as well as continuing the host-based versions.

Cisco's IDS/IPS offers have been pretty lame the past few years, but they've got decent firewalls, so we'll see how those product lines play against each other. (I don't know what Cisco's doing in Anti-virus and cloud malware detection these days.)

Sourcefire's hardware at the low end is basically Linux box appliances, and at the high end they're doing a bunch of hardware acceleration. Their largest single box will handle 10 Gbps of inspection, and they can cluster up to four of those to support 40 Gbps. There's not much competition up at the high end - McAfee may have come out with a 10 Gbps follower to their previous 5 Gbps box, and Juniper has some boxes that are bigger but are mainly firewalls with some limited IPS capability. If you've got existing Snort on Linux, Sourcefire does also sell connection tools to integrate with their management systems.

*The term "Next Generation Firewall" means "whatever Palo Alto's marketing says it means", but is at least firewall plus application identification. I've heard that Cisco tried to buy Palo Alto last year.

It was easy to have a responsible attitude about it, because I actually *was* renting. And my landlords were trying to sell the place (which I think they managed to do before winter.)

The first computer I was allowed to hack was an IBM 403 printer, which I used for a Boy Scout mailing list. We weren't allowed to mess with the wires on the plugboard, so it may not count, but we could do anything we wanted with the paper tape, punch cards, and card sorter. It had very little internal memory, but it could fit a big stack of cards; 1000 cards * 80 columns = 80K bytes, and I think the character set had 48 values (so you could call them 5.5-bit bytes if you wanted.) And the Model 026 keypunch ran on vacuum tubes.

The PDP-11 probably had 24KB of RAM, maybe 32KB. I think they upgraded from an 11/20 to an 11/44 after a year. It was based at the local university, and a dozen or so high schools time-shared on it, using RSTS-11 and programming in BASIC, with Model 33 ASR teletypes.

The first computer I owned was an HP programmable calculator (I forget if it's HP-21 or HP-25. 49 words of program memory, 4 words of stack.) The next computer I owned was a retired 386, because there was no point in owning a home PC when I had a terminal into the machines at work.

The first computer I ran myself was a VAX 11/780, which had a huge 4MB RAM that required two cabinets. Our application really needed 12MB, so I played a lot with virtual memory for a few years; after a few years chip densities improved and we were able to afford to upgrade it to 16MB, and suddenly the application ran in an hour instead of a week. (We could have probably done that upgrade a year sooner, saving a lot of work and getting better results, if the bean-counters hadn't thought that capital budgets and labor costs were entirely different kinds of money.)

My current work computer has 4GB of RAM, and [grumble] 32-bit Windows on it, which is the current annoying-640K-equivalent. The hardware would be fine with a 64-bit OS, but the IT department isn't. I am connected into a larger VMware server, which probably has 48GB, and most of my VMs are 1-2GB. And I've also got a window connected to Google - I don't really know how much memory it has :-)