Is cloud computing rubbish?

Here’s a pretty light-on-analysis piece quoting Richard Stallman and Larry Ellison’s skepticism and warnings about “cloud computing”. What’s going on here? Are these guys just too behind the times to “get it”?

As with anything, the devil is in the details; and this article is too lightweight to give the whole picture. A more nuanced analysis and report of what these two people are talking about would point out that their caution is all about trust – and mostly rooted in the warning not to store data in the “cloud”. Giving your data away to a third party is convenient, but is too likely to lead to long term grief; you have to trust that party to keep your data integral (backed up and accessible at all times), safe (good security from the outside), and secure (no access to your data from the inside, and no use of it for any purposes other than what you explicitly authorize). In truth, no existing web facility can claim to be better at this than keeping your own data on your own machines.

Richard Stallman is additionally making another point about cloud computing: as far as I understand him, it’s okay for you to use the cloud to perform computations if you’ve installed your own software there, but not reasonable to trust a cloud computer that someone else can put software on. In the end, that means really that any cloud computer can’t be trusted, even Amazon’s EC2, since Amazon’s emulator could be designed to allow them access to your running instance at any time. However, I’d guess that on a sliding scale of trust, their setup is more trustworthy than Google Apps (or Sun’s Project Caroline) where you create applications which are written to their libraries and APIs, without knowing exactly how those are implemented (and how they’re exposing your information to the world or to the application hosts). Google Apps is open source code, so you can run a version on your own computer; in this way, you could create your own cloud and trust it after auditing the code. However, you can’t trust Google to be running a version which is the same as what you could run at home, so even having the source code is not sufficient for trust.

I realize this is a lot more subtle than what the linked-to article is expressing. But even if my analysis is wrong, I hope I’ve inspired you to go read the source material and draw your own conclusion!

The (stupid) future of the Electric Car

Wired Magazine, in their Sep. 2008 issue (on the newsstands, but I can’t link to the article yet) has a cover story about an Israeli company called Better Place which is trying to develop an electric car ecosystem “to eliminate oil” as a transport fuel.

First things first. The young visionary who founded the company, Shai Agassi, is impressive; his goal is ambitious, and he’s clearly got the enthusiasm and credibility to bring important partners on board to realize his vision (he’s purportedly already raised $200 million to fund the startup). The plan is to create an area-wide (country-wide in Israel and small countries) grid of electrical charging stations and battery-replacement shops, charging car owners for recharging and battery replacements. This new company would in essence control the new fuel for their automobiles – stored electricity. They would not create or sell the cars themselves (and that would be insane anyway); the goal is to convince auto makers to use their battery and charging technology in new auto lines. To make things easy on car owners (and to hopefully drive adoption), they plan to have on-board software to calculate trips and battery capacity and to locate nearby charging stations and battery replacement shops, and charging stations will be easy to use with swing-out arms to automagically plug your car in for a charge.

This is attractive to a number of different constituencies. People concerned about the environment like the idea of removing oil entirely from transportation; hybrids (whether the 2-mode systems like the Prius or the superior “range extender” technology of the Chevy Volt) don’t do that and will still require gas stations and a gas creation and distribution ecology into the foreseeable future. Car manufacturers might find the idea of having a single standardized battery technology appealing, although to counter that, it may constrain their ability to optimize and/or differentiate their cars sufficiently. Electrical utilities might decide that this approach would bring them an expanded market for electricity much faster than hybrids will; hard to tell. Governments may find the idea of potentially being independent from foreign oil compelling and may subsidize the effort for that reason.

Here are some reasons why this scheme could be a bad idea for governments and consumers:

1. Presented as a complete ecosystem for purely electric cars, the idea appears air-tight and compelling. However, after any modest reflection, it doesn’t seem to provide much of an advantage over the hybrid technologies which are already farther along in production and which will not put any one company in a monopoly position on batteries, software, or charging technology. Because batteries are large, dense, and expensive, electric cars will always have a limited range; so how you extend that range is still open to debate. The vision of Better Place works fine in a small dense area like Israel or Denmark (perhaps even for much of Europe) – when you run out of range (or are about to) you find a local place to plug in or get a new battery. Now imagine doing that for driving across the US Midwest or Southwest, or really across any sparsely populated area. The hybrid technology answer to running out of charge is to bring the extension with you; and, when that runs out, you can still take advantage of the already deployed gasoline infrastructure until you get to a charging place.
2. Government subsidies (direct or in the form of tax breaks) should not be used to create monopolies. If any government subsidies are extended to Better Place they should extract a number of firm concessions in the business model (and to be fair, for all we know, these may already be in the business plan, but they should be conditions anyway). First, the charging infrastructure should be treated much like a deregulated residential power grid, phone system, or pipeline; Better Place may be allowed to install the chargers and extract a fee to cover their use and maintenance, but car owners should be able to buy their power from any operator. Similarly, Better Place should not be allowed to operate the battery-replacement shops directly or indirectly, but should instead be required to franchise them out (they are likely to prefer that anyway as it shifts the bulk of the capital investment to the franchise owners). Finally, the batteries themselves should have a number of different suppliers; Better Place should not the sole source or broker for the battery technology (and they might not want to be anyway).
3. Privacy is a big issue; you can forget about it with these cars. Every time you plug in to charge your car you are announcing your location to a central server; and that’s only if the on-board software in the car isn’t already reporting that on a constant basis so it can find nearby charging stations and swap-shops anyway. Compare that to fueling your car and charging it at home; if you use cash at the gas station (and Citgo even encourages that by giving a 3-4% discount over the credit price) then you aren’t being tracked at all by anyone.
4. Allowing Better Place too much control over the specification of the battery technology would potentially stifle innovation in the design of the actual automobiles. Because of all the design issues for high-energy-density auto batteries (heat, safety, discharge, operating temperatures, size, longevity, etc.), auto makers need to be able to design all kinds of trade-offs when creating electric cars. I think that in the end, auto batteries will be a lot like lead-acid car batteries or even more varied like laptop batteries; there’s standardization of the input (charging), output (12V/5V, etc), and the macro units themselves are built from mostly standardized cells. But you typically can’t use your SUV’s battery in your sedan, or your Toshiba laptop battery in your Dell or Lenovo, and in fact, you might not even be able to use the same battery across two laptops even from the same vendor. For different models, they make different trade-offs; and this is good. Trying to force a standardized battery profile just to facilitate the “battery bay” swapping would be a bad idea.
5. What’s the profit model in automated battery swapping facilities? Will people pay a huge amount of money to do that ever, periodically, on a regular basis? Is there any analogue in the sealed-lead-acid battery world? Of course the charge lasts a lot longer in SLA batteries, and they don’t get swapped out as often. How many swaps per day/month/year would have to make this profitable for a shop, at what cost per swap? Put another way, what kind of capital investment would a shop require, and how long would a shop take to see a return on their investment?

In the end, although I like this guy’s vision (a future without oil for transport) and his dedication to the concept, I really think this is not a good way to go about it. I’m fine with getting a range-extender hybrid, since I don’t typically anticipate needing range extension during the work week. With that technology, I’m planning to get as close to an infinite number of miles per gallon as possible, and that’s good enough for me.

Using GPL Software in for-profit companies (a response)

This post is a response to Edmund J. Walsh, who writes at http://www.law.com/jsp/legaltechnology/pubArticleLT.jsp?id=1202421869652 :

Mr. Walsh, I believe the tone of your article is misleading, as it at best makes some incorrect assertions.

First, as you are probably aware, the GPL does not preclude the use of open software by for-profit companies. Nor does the license interfere with the distribution of free software. The important point which your article initially glosses over (although it comes back to this later) is that the problem comes when a for-profit company modifies free software, distributes the modified version, and refuses to distribute the modifications. This is precisely the behaviour for which Verizon and others have been accused. Except for the AfferoGPL, all other uses (including modification) are fine even by for-profit companies. So in fact these freedoms do accrue to the users not the software. In particular, this assertion is misleading at best: “Any activity that leverages software for business advantage is likely to restrict the software’s freedom, and the growing use of open source software by for-profit companies has been a growing irritant for free software advocates”, since the opening phrase is not true (“Any activity…”).

Second, your assertion that “Running commercial Web services using open source software without releasing source code has also caused consternation in some quarters” is tangential at best. While there may be “consternation in some quarters”, this is not an issue with the GPL and is a red herring. It adds to the negative tone without improving the reader’s understanding of how for-profit companies may be affected by the GPL.

Finally, you make an unsubstantiated assertion with “The next legal fight could be an attempt to force release of proprietary server code due to some part of the output of the server constituting a ‘work’ generated by open source components on the server”. The cases that I know of which involve GPL code embedding parts of itself in derivative “works” have been analyzed and resolved in favor of the user; the user is free to distribute those derivative works without being required to license them in any particular way (GCC, Bison, Flex, etc.). Unless you can provide an example to back up your assertion, you’re at best using this statement to sow fear, uncertainty, and doubt about the ability of for-profit companies to use free software.

Perhaps the best counter-argument to your article would be to simply point to Google and their use (and extension, maintenance, and in some cases distribution) of GPL-licensed and other free software. Clearly their own legal department have determined that they can use free software to create huge business value. I think it’s unclear that your article provides any new or relevant information which would cause their counsel to re-think their model or their use of free software.

OLPC and Intel

I’ve got nothing more useful than opinions on the OLPC/Intel spat. That said, here are some observations based on the article at news.com:

1. If what Nicholas Negroponte says is true, that Intel is selling Classmate PCs to countries at a loss (i.e., “dumping”), then this is at best antisocial behaviour. Dumping your goods on a market at a loss is a tactic to drive off the competition in anticipation of recouping those losses in some other way or by higher prices later on. By dumping computers on a market, with whom is Intel competing except OLPC?
2. No commercial board of directors would allow a representative from a competitor. Why would OLPC be different? I would guess that Intel was given the choice to stop competing directly or leave the board. But how would I know?
3. The story makes a more serious allegation: that Negroponte and OLPC demanded that Intel stop supplying chips to other low-cost PC makers. If that’s true, it’s really an absurd thing for OLPC to ask for. At best, OLPC should demand that Intel supply them with chips on terms no less favorable than other PC manufacturers. In that case, OLPC would ensure that as a charitable organization, it’s getting a deal at least as good as for-profit ventures; and that Intel is not trying to undercut sales of the XO laptop by providing chips to competitors on better terms.

I don’t personally believe that there should only be one solution for getting technology to children in the developing world. If the OLPC foundation truly can foster a market in which their XO laptop competes fairly, then the beneficiaries will truly be the constituents they hope to serve. If however the market is being distorted by Intel selling laptops and chips below cost to countries and select manufacturers then in the end, Negroponte is correct to call them on it and boot them from the project.

But again, what do I know?

* Footnote: OLPC stands for One Laptop Per Child.