HP misses the spot

There's always been a beguiling modesty about HP's research and development. The company has a global army of very smart people doing amazing things, but they've always preferred to let their inventions speak for themselves. So when a press release appears trailing a "revolutionary" new idea — in this case, Memory Spot — it demands attention.

What the company has done is take a fast radio, half a megabyte of memory and some RFID smarts and built them into a chip. The cleverest part is that it doesn't need any power. It gets what it needs from the radio signal used to send or receive the data — technically interesting, but revolutionary?

RFID hasn't lived up to its early promise, because the technology was answering the wrong question — "How can my baked bean tin identify itself to a passing reader?" The right question is "How can I know what's where?". RFID is just a part of that — the easy part. The difficult part is how you codify all the complexities and human messiness of a physical distribution system and manage it usefully on a computer. That is being worked out and RFID is playing its part, but reality has refused to comply to the marketing dream.

HP's Memory Spot, for all its media-friendly gee-whizz nature, makes the same mistake. Smart objects that carry their own data around with them are nice to think about — but peculiarly useless in the real world. Data that doesn't have a permanent connection to the Net is barely data at all: it's invisible, because you can't search for it, and when you do get it it's already out of date.

An object's name is the only data about it that doesn't go out of date — and you don't need megabytes of data for that. You need a handful of bits for one unique number and a way to look up that number on a database. If you'd made a million widgets with information attached and needed to update that information, would you rather go out to each with an RFID writer and update them — or change the one copy of the information in your central database?

It matters immensely that we can associate our physical world with the shadow universe of data we're creating: we know when we're doing it well, because things tend to disappear. That's another bad sign for Memory Spot: HP's own list of potential uses centres around physical objects such as photographs, documents, medical records and postcards, all things that are already rapidly vanishing from the real world.

HP's inventions will succeed if they fill a real need better than the alternatives. With Memory Spot, there's almost no real need — and where there is, the alternatives are better. Beguiling modesty is always better than feverish hype: one piece of data that deserves a permanent spot in the memory banks of HP Marketing.