Friday, August 8, 2008

TiVo the Olympics with semi-open, semi-compatible standards

The Olympics begin today, and NBC hopes to earn back its nearly $1 billion payment by broadcasting and webcasting thousands of hours of Olympics coverage. In Thursday’s USA Today, technology columnist Edward Baig wrote about how this coverage is going to tax any TiVo’s storage capabilities — particularly for viewers who go for HDTV (which is 10x as bulky as NTSC).

A sidebar by Baig notes that the TiVo HD and other recent models are expandable with off-the-shelf commodity hard disks that use the eSata interface. TiVo used to require you to buy a larger TiVo to get more storage, but not any more (perhaps due to competition?)

This remind me of a favorite among my own papers, a 2006 book chapter that attempts to spell out what is an “open” standard. The term had previous been abused — either as a marketing slogan, i.e. magic pixie dust to be sprinkled on platform technologies (cf. “OpenVMS”) — or a binary yes/no classification that grossly oversimplifies a wide range of alternatives. Instead, the paper notes that there are both degrees of openness (like completely to not at all) and dimensions of openness (e.g. for competitors or complementors).

Normally platform owners want complements, but sometimes (as with TiVo selling expanded disk drives) they try to lock them out so they can get all the add-on sales themselves.

This was a big deal in the 1970s and 1980s when IBM and DEC were charging obscene margins for commodity disk platters because they had an IBM- or DEC-compatible controller. Entrepreneur (turned congressman turned professor) Ed Zchau made a fortune working around these problems so he could sell cheaper disk drives to minicomputer owners. During my computer job in high school, we figured out how to write our own disk driver software so we could use a cheap 3rd party drive (20mb if I recall correctly) on our HP minicomputer.

For owners of the newer TiVo models, Baig reviews $200, half-terabyte spindles from two of the big three (Western Digital, Iomega) and notes Seagate has promised its own model Real Soon Now. These same drives also work with the Scientific Atlanta DVR that’s bundled by all the cable TV companies. Interestingly, TiVo recommends the WD but not the Iomega, even though Iomega claims theirs will work. I don’t know what’s going on, but there’s two good possibilities.

One possibility is that TiVo doesn’t really want to be open to all third party complements, because it has some sort of marketing deal or strategic alliance so that TiVo recommends WD in exchange for something of value (cash, royalties, support help, comarketing).

The other possibility is that TiVo has found a real incompatibility and doesn’t want to burn the slim margins with thousands of $50 support calls. In this latter explanation, the eSata “standard” is not actually all that complete, and so various host and drive implementations are mostly but not quite entirely compatible.

Normally I’d suspect the marketing explanation, but my luck with hard disks the last few years supports the incompetence argument. I buy about two hard disks a year for backup (bigger every time) and use them with one of the four Macs we have at home or my personal Mac at work. Between FireWire 400 and USB 2.0, every drive is either flakey out of the box or fails to work reliably (i.e. be visible to the host) within a year or two. I even switched to name brand drives, but to no avail.

Once upon a time — more than 30 years ago — I could debug RS-232 serial connections with an oscilloscope to check waveforms and timings. Today, the complexity of some of these interfaces requires an equally complex debugging instrument that may be no better at implementing the host protocol than the host device.

As with the incompatible modems of the market, time to market is causing firms (presumably on both sides of the cable, not to mention their component suppliers) to ship implementations that mostly but not entirely conform to the standard. I’m sure there are buyers out there that would pay an extra 10% ($20 on a $200 disk drive) for something that conformed to the standard — and perhaps even was tolerant of slightly nonconforming implementations.

But none of the trades — InfoWorld, CNET or the like — are providing this information. Is it because they don’t have the equipment, because they don’t pay their testing lab directors well enough to get a good EE, or because they don’t want to piss off advertisers? I dunno, but it seems like

No comments: