Shooting at clouds

It's not surprising that a man who doesn't use a browser doesn't like web apps.

Cloud computing's marketing appeal is that 95% of the populace would rather that computing was kept up in the cloud and off of their f'ing desks.

Utility computing is more descriptive and important, and I love your contrasting the two -- and your point that Oracle is growing (and investing in) their SAAS products.

Oracle is better than anybody at selling to their install base. When Oracle acquires companies, such as Peoplesoft, it does so as much to acquire customers as products and technologies. Old technologies are profitable in a number of ways, especially in software where upgrades are a gold mine.

Utility computing threatens Oracle's business model in several ways. The first is by eliminating profits from software upgrades. The software upgrade process is much easier for customers because they don't have to do it - and it's free in most cases.

The second is that the technology powering a utility application is hidden from the customer. Customers don't care much if there is an Intel (or Oracle) inside the applications of their SAAS provider. The technology becomes secondary, just as the manufacturer of a disk drive became unimportant to buyers of PC systems. Oracle stands to lose a lot of its revenue potential.

Most computers have an autoupdate feature enabled by default. That includes a lot of Linuxes. (And there have been accidents involving this...) Then there's the BIOS underneath the OS... etc. Trusting Amazon isn't much more to ask for. Most business under 200 people will be 100% virtual for their core servers in 5 years.

Why would I want "computation off my f*cking desk"? My desk is very convenient for me. I might want my *data* distributed to my desk, and across every other device I have. 95% of people don't want a dumb terminal.

Cloud computing is a marketing term and like all marketing terms it takes work to define it in such a way that people understand and accept. Up until now neither Amazon nor Google, nor any of the early adopters of the term have managed to make it mean anything. As a result, every company out there sees an opportunity to either stretch the term to their liking or try to declare it dead. Both Richard and Larry have reasons to hope "cloud computing" as a marketing term fails. Richard because he sees it as promoting proprietary software, and Oracle because they are heavily invested in the term "Grid" which they would prefer the industry to use.

Updating desktop software is one thing - updating enterprise applications is something else. There is no auto-update for those applications and they tend to be expensive and have unpleasant risk exposures.

I agree with marc. But the laptop of the CFO who has full admin rights to the enterprise application might have a copy of an RSS reader on it, which will autoupdate, giving some small s/ware firm in Berlin potential access... There are so many potential points of control, open-source or not, that Stallman's just being paranoid by fixating on physical ownership.

The problem with people like you, Nick, is that you don't understand software. Not really. I mean that respectfully. Stallman is dead-on right and here is why:

What people "like" -- what they've voted for with their feet -- are various useful details of the user experience they get. They like programs that link them in communication with others. They like programs that are conveniently available (with their data) from any server. etc.

Because you don't understand software you think that means they like centralized proprietary computing. You fail to recognize that nothing in those charming attributes of the applications in any engineering sense (technical, economic, etc) justifies or necessitates the centralized service, the EULAs, the inability of user's to change the server-side software they themselves use, the price the user pays, etc. "Putty in their hands" is right and you've illustrated it.

Hey man, hackers just crashed the world financial system. You want us to do the cloud next?

-t

I think Stallman is right, but there are Cloud Computing / Platform-as-a-Service companies today who are offering "Open Cloud" solutions. Meaning, you are not locked in and there is no proprietary implementation required. You can move on and off these platforms freely. So, you're not locked in. Two examples I know of today are HostedLABS and Joyent. Probably more to come....

Tom Lord et al.,

I agree that I was too dismissive of Stallman's overarching fears - lock-in, exploitation of personal data - in this post. I discuss the drawbacks of the centralized utility model extensively in The Big Switch (and elsewhere). But Stallman's paranoia blinds him to the market reality, which is this: when it comes to personal software, people are more than happy to transfer control to outside providers in exchange for convenience, low cost, and - this is also important - the benefits of data and application sharing that come with a multitenant model (there's a reason that social networks are popular and a reason that their rise coincided with the rise of cloud or utility computing). Yes, at a software level, I'm sure there are ways to run a Facebook or a Google search engine while ceding less control over personal data to the corporations running the systems, but the lack of commercial incentives to create such systems doom them to either ugly, kludgy, or nonexistent. (Stallman may be willing to "send mail to a demon which runs wget and mails the page back to me" in lieu of surfing the web as normal people do, but he's an odd duck - even if I, personally, have considerable sympathy for his fears.)

Corporate software is indeed a different animal than personal software, but here too the benefits - cost, flexibility, collaboration - of multitenant systems are real and compelling to businesses and increasingly (if more slowly than the hypesters want us to believe) will push companies toward the utility model. (Hopefully, companies will be concerned enough with the dangers of lock-in, that they'll demand data-portability, standards, openness, etc., as this transition proceeds.)

I agree with Thomas Barker that "Stallman's just being paranoid by fixating on physical ownership." We're all networked now, like it or not.

Nick

I fully understand your concerns, but I do believe a significant privacy crisis could encourage both users, on one hand, to exert a little more control, and, on the other hand, data-gatherers and other cloud-farmers to try to limit their responsibility. Aren't people in the UK a little more concerned now? I certainly do not which it happened, but it might, as setting up a proper menu for privacy settings is so hard. After all, judges decided that (in Microsoft case) there was such a thing as an unintelligible manual.

Although data storage is cheap, a good programmer by himself might be able to program a proper Distributed Social Network (DiSo) with a reasonable business model — and such a feat would probably drive the attention of decent designers and other UI-capable people. I actually have feasible ideas for such a strange contraption: Skype, a distributed service, does work, and FireFox (an open-source project) makes money; it doesn't make much sense now, no more then a mainstream open-source browser was a legitimate goal back in 2001. Look were we are now. . .

Wow, a famous writer agrees with me - yey internet! :)

But (the other) Tom is right, a lot of commentators, including Nick, don't really understand the mechanics of software. See my trains comment on the Chrome post :-) Centralisation often isn't helpful to making the systems users want and need. Better control over IT ought to see functions smeared across the network, rather than just concentrated down to a few points.

Perhaps we ought to resurrect the idea of "distributed systems" as an alternative to talking about The Cloud? That implies practical choice over where functions are and how they're inter-linked.

On a side-note, I designed (but didn't really implement) a DiSo system at uni. [http://thomasbarker.com/files/Distributed_Web_of_Trust_0.pdf], which I've started to think about again after a session at BarCampLondon on "The Social Graph". Lots of arguing went on. Someone even suggested a "credit score" for creepy people! It made me realise that the only sensible way to govern these things is for each person to have their own personal "truths" about their relationships. They figure out their wider social position/graph through dialogue. And a big part of that is lying, being ambiguous, inferring relationships between different part of people's lives, etc.

Nick,

Stallman has had *some* success over the years of changing the demand-side of the equation. I'll return to this point in moment but keep it in mind.

You say that the benefits of the multi-tenant model are observably compelling since so many people go for it in the role of user. You're right -- lots of people are signing up for this crap. That much is trivially true.

You say that there is little business incentive to offer such features in another form. Well, you're too quick. First, what if the demand side becomes better informed as to what's going on? Your view that people "just do" go for it might not hold. RMS has some history at influencing the demand side. Second, along with the downsides for users of this tech comes, hand in hand, big risks for the providers. Screwing the users comes, in the long term, in a society that progresses along the path of enlightenment, with big risks. At some point, that mode of business can well flip from buy and hold to sell.

Your position is tricky because professionally your bread and butter audience is more likely to be on the "pro" side of these business models that screw users. In glibly stating, as if it were a permanently done deal, that these relatively recent models are inevitable and irresistible we have to question the line between where you are reaching a logical conclusion vs. where you are pandering to an audience of oppressors.

Your point about RMS' odd way of using the web (namely via email) is, um, kind of an ad hominem. It has no logical connection to your thesis. It mainly has to do with (a) his travel habits and (b) his bandwidth. Sure, he is not "of the body" of the average web user but that's for practical and principled reasons and not because he doesn't get what's going on.

In the corporate world the concerns of free individuals morph slightly but are basically in tact: the current form of centralization creates single points of failure and critical points of irresistible surveillance. Once a few more enterprises get badly bitten, they'll be demanding decentralized computing, too.

-t

Robert Stallman has a peculiar way of surfing, and I doubt it has to do with his bandwidth—he reckoned himself this was an inconvenient way to stick to his values, which should justify why his claims are to be taken with a grain of salt. I fully understand his concerns, but the legalese of Free Software as well as the design of a safe DiSo, that's two major pains in the back—needed, but harsh.

Although this as obvious link to software control, his turf, I would just like to point that he is not the only one to have the expertise and the capability to defend such positions, nor the first, and I was personally far more influenced by people such as Christophe Aguitton and Ian Brown: arguing that a for-profit company (or a government) is not the best repository of exhaustive, centralised information was a concern put to law in the 70's — and I remember taking a oath to protect ‘statistical secret’.

"When it comes to personal software, people are more than happy to transfer control to outside providers in exchange for convenience, low cost, and - this is also important - the benefits of data and application sharing that come with a multitenant model"

The whole point that underlines both of Mr. Stallman's interests is the relationship of contemporary technologies to fundamental rights.

As much as he might be considered an outsider, his principled stand is a welcome counterweight. Happy people who give up control and define the market are still, in the long run, the worst enemy of anyone who values a pre-surveillance/security fractured culture, odd duck or not!

The vast majority of the happy customers have very little interest or idea about the ultimate loss of privacy than will come down the pike!

Alan

I would expect a major driver of privacy-friendly technologies to be a desire to avoid subpoenas, rather than protecting users. GMail's is going to get a lot of lawyer's letters over the next few years!

We are at the moment, constantly reducing the set-up costs of a totalitarian society. However distant the possibility, being able to run a few queries across the mobile phone position database and Facebook to find all the foreigners/liberals/royalists, etc, is not a great thing for human freedom. But that's not going to stop 14-yr old girls using Bebo, or me carrying my mobile around...

FYI... be careful dropping RMS's name. You can be sure to get some heated disagreement in any ensuing replies.

Nick,
Cloud computing is just yet another misused phrase. People now try to use the word "cloud" wherever they used "utility","saas","paas" or "hosted". But fundamentally, all these are manifestations of disruptive software engineering trends like virtualization, SOA, OpenSource etc..

Thanks,
Arun.PC

Well, you're too quick. First, what if the demand side becomes better informed as to what's going on? Your view that people "just do" go for it might not hold. güzel sözler

Thanks for signing in, . Now you can comment. (sign out)

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

URL:
Remember me? YesNo

Comments:

Rough Type is:

Written and published by
Nicholas Carr

Designed by

JavaScript must be enabled to display this email address.

What?