Wait, did the Mac Win?

Just reading David Winer’s blog post saying it did.

In installations, that’s clearly not true. In business use, that’s clearly not true. In mindshare…it may well be true. And Apple is supreme in single-company computer sales…but are moving out of that business, it seems to me.

But I don’t want to talk about that.

What I want to talk about is his assertion that malware was a nightmare in the Windows world and in fact drove the Mac win.

That’s just not my experience. I’ve had Windows computers at work the whole time he discusses, and windows computers at home since I first installed Windows 1 (to develop an application). I’ve had full-time (broadband) connected computers at home since 1996. And in all that time, the only virus infection I’ve seen is one Word macro virus that I got in a document sent me by a fellow Minicon committee member. That’s it.

I’ve mostly run an anti-virus program — but it’s never stopped anything (it has, very very occasionally, suggested that something I downloaded from a site I considered dodgy probably contained something bad; but these were all cases where I had deliberately downloaded it, and manually scanned it to make sure because I felt there was cause to worry).

I don’t click things totally at random.

And I mostly don’t go to dodgy sites in the first place. I’m not often looking for illegal software or content, for example.

Malware has had essentially no impact on my user experience. It’s simply not a factor.

I have no idea why my experience is so completely different from Dave Winer’s.

But it’s a bit interesting.

One in a Billion Opportunity!

That’s how Malcolm Gladwell described Bill Gates walking into his highschool in 1969 and finding they had a computer terminal that students could use.   The quote is “He had this one in a billion chance to get good at programming in advance of every single member of his generation.” (In an interview played on National Public Radio’s All Things Considered this evening.)

This, of course, annoyed the hell out of me.

First, how many people were there at that highschool?  The opportunity was not unique, it was at the very least available to a few hundred or a few thousand students at that school.  And even considering the whole population of the world at the time, it’s a couple of orders of magnitude less rare than “one in a billion”.

Second, I had a very similar situation.  My highschool had a computer in 1968 (not just a terminal), and that’s where I started programming.  I also had access to a similar computer at Carleton college, and a few years later got access to a different computer at St. Olaf college.  And in fact in 1969 (when I was 15) I was hired to write computer software for Carleton.

So we need to include all the people in my highschool into that count. And I imagine there are several more worldwide; leading to a total of thousands or tens of thousands of students who had access to a computer in 1968 at about the same level Bill Gates and I did.

And very few of those people are titans of the software industry.  Most of them aren’t even in the software industry (I am).

Certainly opportunities are unevenly distributed, and play a major role in what people get to do with their lives. But very often, when you look carefully, the opportunity is not as rare as you might initially think; and the probable importance of hard work and choice by the individual thus becomes greater, not less.

Planning a New Computer

I’m starting to feel the urge to upgrade my computer at home (partly because the one they gave me at work is really pretty good). And of course I’m all out-of-date on what’s good.

I don’t play 3d-graphics-intensive games on this computer, and do very little video editing. The challenging application for this computer is photographic work, with Photo Mechanic, Bibble Pro, Photoshop, and Thumbs Plus being the main consumers of cycles. I do multi-task a lot, running Thunderbird and Firefox pretty much full-time, and often playing music through WinAmp while working on photos.

Photo Mechanic, Thumbs Plus, and Photoshop make Linux out of the question for this box (Bibble runs excellently on Linux, but there are no good alternatives for at least two of the other three). Dual-boot is not satisfactory. And besides, I’ve got a Solaris box with lots of spare cycles up on the same shelf (it’s primarily the file server), so stuff I preferred to run on Unix I could just run there. So vitualization doesn’t really seem to be a big win for this new box. One thing it might buy me, though, is living within the 32-bit Windows memory restrictions. If the Windows virtual slice *only* ran the photo programs, not any of the multi-tasking, having only 3.3GB memory available might be tolerable longer (I’m not doing stitching to produce gigapixel images, or any of the professional-level work that makes 500MB files normal; my PSD files rarely exceed 150MB), though, which might keep me from having to run Vista (64 Ultimate).

I have the general impression that memory throughput is the biggest single thing that can help Photoshop. Anybody have any clear information on that, or pointers to good benchmarks that would confirm or contradict this hypothesis?

Some Photoshop functions, and Photo Mechanic, and Bibble, and apparently the new version of Thumbs Plus, are pretty heavily multi-threaded (making use of more than two cores, I mean). Plus the multi-tasking. So I’m inclined to look at a quad-core rather than a slightly faster dual-core chip. It may lose a bit on some slow Photoshop filters, but it will win in the things I most commonly run 500 photos through.

I’m keeping all my photos (and the rest of my Windows Documents) on the file server, not on a local disk at all. I need a local disk for the basic software installation, some Windows pagefile (I expect I’ll start with 8GB of ram, so it probably won’t page all *that* much), and temp files (and the photoshop pagefile).

I started out thinking of maybe getting two of the 10kRPM SATA drives (small ones, 80GB or whatever the smallest size is), one to be the software install disk and the other to hold temp files and pagefiles. Then it occurred to me that small solid-state disks (flash memory packaged with an [S]ATA interface) were in that price range too, and they’re faster, or should be. I’m looking for clearer specs on this.

Also, the argument for splitting those uses across two disks is mostly access arm motion, and SSDs don’t have access arms. They still have bandwidth limitations, but they’re pretty good. So one 64GB SSD would probably perform better than *two* 10K magnetic drives, and would cost considerably less.

Then the really creeping wacko idea hit me: Two (or more) SSDs, striped (RAID0). Double the read and write bandwidth available for any large transfer, and no access arms so the competing uses would compete *only* for bandwidth, not for access arm position. I think perhaps this is too creeping wacko, but I do find it amusing to think about. Maybe I’ll benchmark it. (The lowered reliability is not a problem since it’s just installed software, doesn’t change that often, and the underlying drive reliability is much higher than with spinning platters.)

My impression is that Intel is the one a bit ahead at the moment in the chip race.

The graphic card doesn’t matter much any more; a third-tier mainstream card will do fine for Photoshop. I don’t have and won’t be getting this decade a full-Adobe-RGB 16-bit monitor or anything like that. I do need dual DVI out (well, only one this instant, but the old 17″ VGA tube is getting a bit long in the tooth).

Prices seem weird. ZipZoomfly will sell me 8GB of DDR2 800 memory for half what General Nanosystems will. Best Buy wants three times more for a DVD drive than General Nano does (I guess that’s not a total surprise). Motherboard availability is spotty overall. Hard drive prices, too. Best buy has a dual-DVI graphics board  a lot cheaper than General Nano.

I’m having trouble finding any motherboard reviews that really address anything I care about for current products. I’m largely looking at the Asus P5K family, and some at the P5Q. I do want IEEE-1394 (though in fact all my external devices are dual-mode), I don’t want WiFi. I want at least four RAM slots, but I don’t want to pay for a server-level motherboard, so I’m unlikely to get more. (Yeah, room for 4 dual-core fast chips would be attractive, and room for 64GB of RAM, but it’s also kinda expensive.)

The case doesn’t have to have room for lots of disk drives, finally, so I’m looking at a mini-tower for the first time in ages, and maybe even at one of the miniature (“shuttle”-type) cases. I do still care a lot about cooling, and a little about noise (there’s the file server with 8 spinning disks right next to it, and a window air-conditioner).

The current system is an Athlon XP “Thoroughbred” (family 6(7), model 8(8)). I think it was a “2800+” model. Single core, single thread. Currently with 2GB of PC3200 DDR ram. So, really, it’s not going to be hard for the new system to seem considerably better.

Still in research mode. Input welcome.

NS2 Taken Out of Service

For a few of you with geeky interests and long memories, this may mean something. NS2 was my second server, primarily a secondary nameserver, but also having various other functions loaded onto it now and then.

This was essentially the last of my servers providing public services; I relocated all that to Dreamhost after the second meltdown. Continue reading NS2 Taken Out of Service