ChuckyBox said:
Okay, let's have a show of hands: How many of you think the next production run of the U.S. Series 2 boxes will include an ethernet port? Faster processor? More memory?
Somebody needs to pop one of the Taiwanese boxes open, because that's going to tell us where we are going.
Tivo is software.
From a software perspective, you want to have as homogeneous a hardware platform as possible. Then when you have a new software release, you can easily propagate it to all devices. Look how long it took TTG to come to the DVD burners. For the future, the reduction in variant platforms means that you have fewer sources of bugs due to unanticipated subtle hardware interactions in legacy devices. My feeling is that existing SA Tivos are going to be around for a long long time to come. The parts that break can be fixed, so until the cost of replacement comes down to the cost of a repair, you are going to have a hefty number of legacy devices.
But this sort of conservative approach to hardware updgrades is related to whatever Tivo's software development philosophy is. I'm a big fan of simultaneus builds and testing for all platforms and a single unified code base. Do a weekly build to all platforms both single and double byte character versions, big and little endian. You catch problems with poor coding practice before they pollute too much of your code base. But the software for each Tvo model may well not come from the same code base and might actually be a fork due to complexity of squeezing the last drop of performance out of proprietary hardware which used barely enough computing power to get the job done.. The only stuff I did of that was very old school- all assembler and direct manipulation of memory mapped memory io locations with critical timing dependencies. Any change in hardware meant that all timing dependent interactions would have to be retuned and retested regardless how distant they were from the hardware change. Anyway, the point of this ramble into unified software bases is that it is hard to do at first when you are pushing the hardware to the limit, but it becomes easier as hardware perfomance is sufficient to be more forgiving and smarter so that you can upgrade to new software releases rapidly. The programming staff can spend much more time on new features than porting and tweaking for all the different variants (140, 240, 540, "560" (dvd burner).
So I think it would be a mistake to expect the kind of incremental hardware variations you see added each year to general purpose computing platforms where it doesn't matter because software is insulated by drivers, or added each year to totally hardware oriented CE devices where it doesn't matter becase software is an afterthought.
That's bad news for tech whores like us, but good news for Everyman (aka joe average user).
With GP computing it is easy where the hardware is virtualized and all you have to do is write a new driver (which usually is provided by the device vendor). With hardware oriented CE devices, the hardware engineers don't give a rat's *ss about the software implications and will break the software to get the latest and greatest chip in. There are no legacy issues because as you see, software updates maybe last the first year (and then only to fix bugs), then there is nothing.
So Tivo is in that middle ground, and something has to give. ONe thing you can do is try not to dribble out incremental hardware improvements, and instead put more of them in fewer models
If dvrs were already well known in the Taiwan market, and there was a lot of demand there for high end features, I think you could have seen a major new experimental platform in Taiwan. But the huge market for greater China is in the Tivo for Everyman. The SA architecture machine.
When we get a shot of the TGC motherboard, I think you are going to see the same broadcom chips. I think there is a chance for variations using chips that are even cheaper or using chips manufactured in PRC (for political, not technical advantages).
On the other hand, the US machines won't always lead with new features. Some features are easier to do in Greater china markets. For example, whenever the high end market emerges and you see HDTV broadcasts there, you probably will see analog encoding of HDTV (components) inputs first in China. Obvsiously, it may take a while for the demand for HDTV devices to grow in China, but the opposition to HDTV encoding shows no signs of ever slackening in the US.
This is all just armchair engineering opinion, but I think some of the considerations I touched on play a part in the decisions Tivo is making on minor upgrades to the platforms.