(LinuxWorld) — Here it is, just as promised. This is the conclusion of the Perens extravaganza begun last week. This week's column is made up of excerpts from my conversation with Perens two weeks ago. The conversation followed no set course. My questions were not pre-ordained; they wandered over the Linux terrain like it was a cow-path in Texas. But Perens' responses were lucid, well-informed and very interesting. I have trimmed and edited the conversation a bit in order to pack as much of it as I could into this week's space.

Lobbying for change

Perens: You know, that is an interesting question. We have jackbooted federal marshals accompanying the BSA into your company to audit them. But first of all, contracts are still governed by state and federal law, and there is still an assumption of innocence until proven guilty.

What is happening is that through various legal "gotcha's" that are incorporated in the EULA, or in things that are deceptively sent to companies, they are getting a company to waive a legal right: the right to privacy.

They send you a postcard and ask you if you want to know about licensing. If anyone in your company signs it and returns it, it actually gives them authority to come in and audit your company. I don't have direct experience with it, and if you go look into the press on BSA you will find out about this one. They target low-level employees in your company, and it sounds on the postcard like it's a seminar on software licensing. But if someone checks "Yes, I'm interested", then somewhere in the fine print it's actually an invitation to come and audit the software licensing for the company.

So that is the dodge they are using, and as a hot-headed journalist, I would suggest that you study that particular thing more and bring it up with your Texas politicians. Because it seems to me that there is some degree of deception — or may I say fraud — going on there.

It helps a lot to get very specific about these things, rather than just say contracts should be subordinate to the constitution rather than vice versa, because if you were to say that to a politician he would just say "Well, I thought they were." So you really need examples of where the law is being bent. You have to get extremely specific.

But I think that what you were suggesting may not always be the best thing for us. If you go back and look at UCITA, UCITA was actually an effort to do just that, to pretty much spell out certain of the terms that could be offered in software licensing, where a number of parties felt that there was a need for reform.

The Bitkeeper debate on LKML

LW: Have you kept up with the Bitkeeper debate on the Linux kernel mailing list? (Richard) Stallman and Larry McVoy have been going at it lately.

Perens: I sent you something about Larry a year or so ago, I think I just shocked you with it. Every time I talk with Larry he threatens to sue me, so I just don't talk with Larry any more. I think that he is just the perfect person to be in an argument with RMS. I actually have known Larry since he was seven. I used to live down the street from him.

I think that [the Bitkeeper debate] is an interesting situation because Linus seemed to need the technical help, and the technical help seems to be working. Larry very definitely feels that he cannot support his family with this product as an open-source product. In fact, he's hardly able to support his family with it as a "not-open-source" product. He claims that making payroll every week is a real pull. It's said to be good; I've obviously never tried it.

There are people working on open-source alternatives. One would be subversion. There was a fellow named Tom Lord who made a program called arch. Tom, unfortunately, has not been able to continue with that due to personal reasons — not due to the product, which is still a really great product. I haven't really heard from anyone who has picked it up. Had he been able to continue, I think that would have been something capable of competing with Bitmover.

So I am disappointed. I think that the Linux kernel team should get away from Bitkeeper as soon as there is something else that can do the job — in fact, even if it doesn't do the job quite as well. If I were a business person... I'm very sorry, but my previous interaction with Larry has been just so crazy that he is not the person I would pick to do business with.

Generally, I like the open-source world to not have to fall back on proprietary software. I think that something is wrong when we do.

I think that the Linux kernel project, a very visible open-source project, [can't] just go to the public and say "Here, open-source development can't even support itself" [and] we have to go rely on some non-open-source tool to do it. So I find this all a very distressing situation.

In every case I am asked by people who want to make their proprietary tools available for free use by the open-source community, my usual answer is "Don't bother." I've said that to some rather big companies that make compilers. I've said that it's a mistake for the open-source community to make itself dependent on a tool that we cannot necessarily maintain, do not know if we can continue with its use, and cannot develop by ourselves.

I discourage open-source projects from incorporating any proprietary tool into the development chain, and I discourage producers of proprietary development tools from approaching the open-source community on anything less than open-source terms. If they find the way to make open-source terms available — and, of course, I evangelize them pretty hard about it — obviously I encourage them and will help them as part of my business if they wish.

What do you know about TCO?

LW: Do you have any comments on the recent marketing activity on TCO "studies" showing Windows to be cheaper than open source?

Perens: I was really glad to see Grant's study (See the link to Grant Gross's piece for Tech Republic in Resources below). There is [an IT advisory agency] in New York called the Robert Frances Group, and they have a person there named Chad Robinson. He has a number of industry clients currently using [Linux]. First of all, he determined their salaries and it turned out that the Linux sysadmins made an average of $71,000 and the Windows sysadmins made an average of $68,000, which seems to me to be almost insignificant. Second, he found that the Linux sysadmins managed an average of two or three hundred systems whereas the Windows sysadmins managed 30 or 40. So this is a study that says Linux really is cheaper.

LW: A Windows sysadmin friend said the study was totally bogus because it's all about the skill of the sysadmin, not the product.

Perens: I think that is very interesting because obviously, when you go to pay for this, the fact that the competence rests in the person rather than the product is irrelevant to the money you are spending. The point here is that in general, Windows sysadmins do not script as much as we do. It is actually a product feature because we have made scripting incredibly easy; there are a hundred different ways to do it, there are 20 different languages. You'll notice that one of the Microsoft directions now is to provide a top-quality scripting and command-line environment.

So they realize that "Yes, we actually do have them beat in areas." What you're saying is that Linux sysadmins are in general more competent, which brings up emotional issues. It may be that the competent people simply do not choose to work on other platforms. I would not criticize Chad's study on that basis.

It's nice that we have our own study now on the TCO issues, but I want to bring up that Windows TCO issues are not entirely bogus. Think of the kind of person who just doesnt have a sysadmin. We're talking about the SOHO who said "I didnt have a sysadmin, so I got a Mac."

Although Windows is reasonably easy for the not-entirely-technically-competent person to install and administer, I think Macintosh leads it in that space. Put that person in front of a Linux system and it is still appreciably more difficult. We have been making progress as far as easing the UI installation. Obviously, disk-partitioning is one of the areas of development, but it's not the only one. You know, for example, that Debian is still using an installer that is patterned on one that I wrote — in 1996, I think. And I was copying something that Ian Murdock had written before. So it's not the same installer that I wrote in 1996, but it looks like it.

In the long term, installation cannot stay a differentiator because it is an easy problem to solve. This means that we need to choose one of these installers that is open-source and all pretty much pool our resources on it. Certainly on the Debian side we do. I think we could even do it in common with Red Hat if we wanted to.

Here come the preloads

LW: Who do you think will be the first major PC OEM to preload and sell Linux desktops in the United States?

Perens: I think it is almost a year away. I think they are waiting for things like a better control panel in GNOME that can actually administer the system. I suspect that the first might be Sun Microsystems for both SPARC and IA86. Sun is not retail level, you know. Sun sells to business. I think that Sun has actually been putting significant effort into its Linux. That could be totally out in left field; this is pure speculation. They have not briefed me on this.

LW: IBM was rumored recently to be preparing a Linux preload for the Indian market. Then, just a few days ago, HP announced a Presario model that they are going to offer preloaded with Linux on the Indian market.

Perens: HP has actually been doing Linux preloads in China for a long time. HP is quiet about it, and I think that one of the reasons they are quiet about it is that they feel that people put bootleg Windows on some of their systems after they buy them.

That is a problem with some of the Linux preloads: we can't say, just as we can't say that all the Windows systems that are sold run Windows. We know that some of these Linux servers come with Windows preloaded. The same thing happens with Linux, to some extent. Microsoft, I think, has been getting a little more aggressive about instituting product activation. I think it is a good step for Microsoft. If you want to have their model, that's a good way to do it.

I think it only helps us Linux folks, because it means that some of their systems will actually run Linux rather than bootleg Windows. My feeling is that the Linux preload actually works better for appliances than for general-purpose computing.

Think of what the typical office worker does with their system, and think of the open-source tools to provide those things: Web-browsing, an office suite and we have a very nice drag-and-drop file-manager in Nautilus. I'm sure there is something similar on the KDE side. I think we have 80 percent of what office workers need to do their job. I think that you could put that on the desks of 80 percent of the office workers in the world.