Observations on Ownership
We are clearly moving to a software focused world — this conclusion is almost as inevitable and natural as taking your next breath (or eating that next Little Bits burger — but don’t get the big one unless you’re really hungry).
But, as with all things, there is a flip side to the world going to software. It could actually turn out that the IT world is on the path to becoming our own worst enemies. This, by the way, is what caught my eye this week, and what causes me to rant a little.
The cost and hassle of repairing modern tractors has soured a lot of farmers on computerized systems altogether. In a September issue of Farm Journal, farm auction expert Greg Peterson noted that demand for newer tractors was falling. Tellingly, the price of and demand for older tractors (without all the digital bells and whistles) has picked up. “As for the simplicity, you’ve all heard the chatter,” Machinery Pete wrote. “There’s an increasing number of farmers placing greater value on acquiring older simpler machines that don’t require a computer to fix.”
The issue at stake, at least in the United States, is the Digital Millennium Copyright Act, which makes it illegal for someone who has purchased a license to a specific piece of software to copy that software in any way. In order to protect their software, makers of farm tractors (such as John Deere — I just sold my old John Deere last year, by the way) prevent anyone but a “factory authorized technician” from accessing the software guts of their new tractors. This might not seem like a big deal, but farming is tied to the weather and seasons. If you miss a harvest before a heavy rain because some engine sensor won’t work due to a software flaw, the farmer could be forced to stand there and watch an entire crop be ruined — with no recourse. Sure, you can call a repairman (you’re not easily going to have one of these towed to the nearest service station), but it could take days of waiting, and a lot of expense in getting the authorized technician out to a farm. Imagine you’re a farmer in Sydney, Nebraska, a little more than two hours driving time from Denver. What if the nearest technician available is in Texas? Hence the drive for mechanical equipment that doesn’t involve software.
The argument is as old as electronic devices. Certainly you own the hardware, but hardware is patented, not copyright. Software, on the other hand, is copyrighted, and so the end user only really ever gets a license. You don’t ever really own the software you “buy,” whether it’s music, an office package, or the system guts of a farm tractor. But what use is the hunk of hardware you “own” if you can’t use it without the software you only have a “license” for? How long will the world suffer with “too big for their britches” companies playing the game of, “sure, you can buy a $100,000 tractor, but I’ll keep the software, thankyouverymuch?”
In fact, this is what’s driving the entire white box revolution — we in the networking world are getting tired of the vendor dictating our software and our hardware in one big glob. Instead, we want to handle the software and the hardware as separate “things,” with (potentially) different vendors, their own lifecycles, and all the rest. We want our investment in the tools and ideas we have around software to be separate from the vendor’s marketing cycle.
How ironic is it that the IT industry is driving a revolution in taxi service by arguing the software is separate from the hardware, while at the same time we see big companies using copyright to prevent the hardware from, well, being separated from the software? Maybe we need to rethink how we do business in some fundamental way.
If we really want computing to be ubiquitous, then we need to have a little more respect for the users who are paying our bills. We’re going to need to separate the case of software embedded in hardware, without which the hardware isn’t useful, and software that’s, well, just software. The only other option is to lose the trust of those same people, and relegate ourselves to a niche — “that’s a nice toy, but to get real work done you need a car without a computer.”
Something to think about.
Thoughtful, interesting, valid observation. Thank you!
One thought: a tractor’s value is fundamentally as a piece of hardware. The software is an implementation choice.
A modern network switch, acknowledging the value of the ASIC, is fundamentally a software product. If it were about hardware speeds and feeds, the data center would be dominated by InfiniBand switches, whose software is relatively simple and primarily open source.
With all due respect for decades of invention by a lot of smart people in Redmond, the creation of Windows NT (think Windows 3.51) accomplished disaggregation, and managed to make server operating systems much less expensive by amortizing their cost of development and support over 10-100x as many units, but did absolutely nothing to address the question of openness raised above. Similarly, if Cisco tomorrow packaged the Nexus software to run on white boxes, it would be just as closed as it is today.
So I would argue that disaggregation and openness / evolution of software from an embedded proprietary model to an open, commons model (Linux) are related but cannot be conflated.
Oh, and the tractor customers as described are exhibiting capitalism at its finest, voting with their checkbooks. Vendors listen!
Steve — I would actually agree with your analysis. In fact, I have a post coming up in a couple of weeks diving into thoughts around what “vendor agnostic” really means, and the difference between “open” and “open.” It’s going to take a couple of weeks to get there, though, as I have a couple of other posts scheduled before then. Thanks for the comment!
Russ, this revolution was absolutely a phased one. I appreciate most facts you called out here but this sounds like ignoring phase 2 and prescribing phase 1.5 regardless of the audience; which could be an ISV, start-up, enterprise, SP, telco, public sector etc. I am sure you’re well aware that the battle of whitebox-vs-nonwhiteboxes is beyond the banal asic-vs-ethchip conversation. Sad but true; falling for the vibe; I can see some people with little tiny and “affordable” tractors who must throw out their vehicle the moment they get flat tire! if you have 5 small hot-spare tractors stashed away in your basement you should be fine but again it comes down to who you are and at what scale are you talking to your customers? Some people do store stuff and this model works well for them. (please let me skip over “the code” you’d need to be running on your “affordable” tractors no matter how tiny they happens to be; reminds me of the old adage of “nothing’s free even the free ones”).
The point I was looking for but failed to see in this article is the favorite statement of “it all depends”.
Going back to the phases; if phase 1 was me owning my own cage in a DC playing around with my notsowhiteboxes, paying for the support, paying for the licensing and connecting up my 100 servers, phase 2 is NOT me replacing my network gear with whiteboxes which I am being told are (quite arguably) cheaper. Phase 2 would be me, moving everything to the cloud, sending 80% of my ops and network engineers home and vacuuming my DC cage and handing its keys back to the security guard. Don’t get me wrong, there indeed still is a phase 1.5 where I go commodity but only IF the cloud paradigm terribly fails me. Otherwise, I’d rather the cloud owner deal with commodity folks, fight for the code, fight for the compliance certs, fight for the space/power/cooling and fight for the logistics while my engineers are fostering my own business which certainly was and is not building and optimizing data centers in the first place.
Two points — first, I don’t agree that the cloud is the solution to everything data center. What you’ve done there is to say, “it depends whether or not I’ll move to white box” (not an argument I made in the first place) — “because everyone is moving to the cloud anyway.” This is one absolute with another, neither of which I asserted. Cloud isn’t the right solution for everything, any more than “pure” white box, or open source, etc. Second — if everyone, in fact, moves to the cloud, that’s an implicit endorsement of white box.
As for the “tiny little tractors” — if a farmer could use 100 small, cheap tractors that were remotely programmable to cover the same space, then yes, I’m certain they’d rather scale out rather than up. At this point that’s not an option being presented to them.