HP And Bruce Perens

After Bruce Perens' brief stint as a venture capitalist (which followed his stint with Debian and OSI among other organizations), he has moved on to work with HP in a sort of consulting role for all things Open Source inside and outside of the company. The article talks about HPs questionable history (including the recent printer driver debacle among other things) and what sort of things Bruce will be up to.

Comments

My personal experience with HP and a suggestion

[Flavio]

Not that I think my suggestion is worth anything to a company like HP, but there are some things I'd like to state.

First, I have a lot of HP products. From calculators to printers, everything I've ever bought from them was excellent.

Now they'll have to either take the first step embracing Linux drivers or face the competition that does.

Speaking realistically, not many companies are a threat to HP's desktop market. There's Epson, Canon, Lexmark and others, but HP is large enough to dictate tendencies.

However, consider that Linux users tend to be influential in the computer world. Let's suppose, for example, that I, as a network admin, have got to install a print server and a box for digitizing images in a small office. Linux would be the perfect choice *if* I had printer support for it.

With cheap printers getting 8+ ppm in black, one deskjet can be more than enough for a small office. I'd use this computer as a mail gateway as well, and maybe for NFS and other things.

I'm NOT willing to get a new box just to run Windows on it and use it as a print server, but as things are today, I have no choice. I refuse to buy a 2880x1400 dpi printer and use it in 300x300 mode under Linux.

Now if some company starts shipping a printer with decent Linux drivers, I'd buy it. I don't care if it only prints with half the deskjet-in-quetion's resolution and at half the speed. I'll get it!

Ditto for other devices with flaky Linux support.

So what I'm saying is that in some situations Linux support can be crucial. Perhaps not for the normal joe that runs office on his desktop at home, but that's going to change as well.

I, as a desktop user, find it irrational to reboot into Windows just to print a document that has a color photo in it.

As Linux takes over other shares of the corporate (and gasp! home user) market, HP will be forced to change.

Corporations don't care much about ideologies per se, but they will care when money's the issue. I'd do it early while I can if I were HP.

Flavio

Re:My personal experience with HP and a suggestion

[Bruce Perens]

Hi Flavio,

The lack of printer drivers (and by extension, other hardware drivers) is the number one thing I'm hearing about from the community. I have now collected a mandate from one division to deal with it, and will work on another division tomorrow. Right now you can look at hp.sourceforge.com and a second effort on sourceforge that deals with HP printers, but possibly not the printers you want. I will probably have to visit the printer divisions in Ohio and Washington state to talk with people. My desire is that all HP hardware interfaces be open and documented. Obviously, I will have to evangelize that within the company. I will have something to say, and something to show, at LinuxWorld in NY, but will not have finished with the issue by then. It could take much longer. In some cases documentation doesn't exist, in some cases you may be able to command the device to destroy itself, in some cases the existing Windows driver contains other people's proprietary IP, not just HPs, etc. So, this will be a pain to deal with, but it'll get done.

Thanks

Bruce

Re:My personal experience with HP and a suggestion

[Trepidity]

Hmm. While generally the HP hardware has been fine and I haven't had any problems with it, the drivers have been complete crap. Forget shaky Linux support; they have shaky Windows support. The printer driver takes 80-90% CPU on my Pentium II 266 just to print a normal page. My old Canon took around 5-10% CPU to do the same thing. The scanner's drivers are utter crap - half the time you have to reinstall them before scanning will work, and if you use multiple Windows users you have to install the drivers/software from each user's login because of the retarded way it stores info in the registry. The CD burner had this minor little bug in that it didn't recognize CD-R's, whether blank or already burned that luckily was fixed in a firmware update by the time I had bought it. You'd think recognizing CD-R's would be a somewhat important feature to have in a CD burner.

So in short: HP has some quality hardware but need to hire some programmers who can write drivers.

Brief stint?

[Bruce Perens]

CT, I spent more than a year as a VC, not a brief stint. You know the market has been terrible, there's no point in doing that now. Also, why the heck was this story rejected twice when I submitted it?

Thanks

Bruce

Careful

[American AC in Paris]

Are we sure this is the real HP? It could be an impostor...

(no disrespect intended)

$ man reality

HP Open Source History

[Baldrson]

While he was at HP, Joe Ellsworth convinced me that IBM was going to support Linux well in advance of the Forbes magazine cover story on open source. That's what got me to create the LibmUX claim at Idea Futures back in July of 1998 predicting that IBM would ship a Linux server before July of 1999. That turned out to be a month or two optimistic, but you have to understand that at the time, none of the suits really believed open source, let alone Linux, would be supported by the big server companies.

In August of 1998, while at the first open source conference, I briefly talked with Tim O'Reilly about approaching Paul Allen's Interval Research concerning open source strategies. I had a few well placed contacts at Interval and I figured if Linus would go work for Allen, maybe it was appropriate that Allen's think tank get in the act. However, it turned out that my contact with Joe was more important than my contact with Interval.

Joe Ellsworth's foresignt at HP turned out to be critical to HP's participation with open source -- something I think he should have received more credit for initiating. Joe knew it would be very difficult if not impossible to get Idea Futures set up as an executive decision support system within HP, so predictions like my (his) LibmUX claim weren't enough to establish priority for open source ideas within HP.

Nevertheless, we did discuss the idea of setting up prize awards for achievement of various open source objectives and after the first open source conference, Joe took that idea and ran with it within HP management, as well as contacting O'Reilly. The end result of his effort was a meeting with representatives of O'Reilly Associates on the same day that I departed for Russia. In fact, I walked Joe to the first meeting with Brian Behlendorf on my way out to catch Aeroflot. Joe thought he had convinced key managers of the HP-UX division to put up almost $10 million in a variety of open source awards that would have systematically converted all of HP-UX's administrative utilities to Linux as a way of channeling the growing base of Apache servers into the HP family of large servers. It was a great positive sum vision that I still think would have worked. In fact, I was convinced enough of its merit that I was traveling to Russia, on my own nickle, to discover what the impediments might be from the perspective of the Russian Academy of Sciences, to distributing prize awards in Russia for open source projects should HP actually come through with some major award money. The RAS desperately needed (and still needs) hard cash for their programming teams. That meeting with O'Reilly went well and my meeting with the RAS folks got their interest up and exposed some of the pragmatics of distributing such prize awards in Russia.

Fortunately, I presented the Russians with a lot of caveats, knowing how often they have been let down by Americans before. I say "fortunately" because support within HP with O'Reilly quickly went a fairly different direction than Joe (or I) had envisioned. For some reason, HP decided not to fund prizes for the massive translation of HP-UX utilities to Linux, and what money was available for prize awards was limited to US participants. Also, for some reason, Joe was not kept as the lead representative in the relationship with O'Reilly Associates and the rules governing the Open Awards program were substantially altered from the original internal white paper on the concept.

I don't know the status of all of this, lo these 2 years later, but its pretty clear to me the entire open source community could benefit from a way to set up objective prize awards, with provision for second and third place contenders. That way programming teams in developing (or recovering) economies can eat and (in the case of Russia) keep from freezing in the winter as they bring their manifest skills to bear on open source.

PPA printers

[Dr. Tom]

PPA printers are like Winmodems; they offload processing to the host CPU, so are cheaper to manufacture, and cost (a tiny bit; $$) less to buy. HP has repeatedly, consistently, and officially refused to release the info required to write PPA drivers, because the advanced color correction done by the driver contains valuable HP intellectual property that they don't want to release.

I can see their point. If they did release the information, and a competitor started using the same color correction algorithm, HP would have no way to know that that competitor had stolen the code and violated HP's copyright (since the competitor wouldn't open the source either). The assurances of large companies that they do not violate licenses like the GPL apparently are not enough for HP.

If only there were a way to enforce their copyright without resorting to expensive reverse-engineering and legal battles, that would clear the way for HP (and many other companies) to release Open Source products. Are there any technical solutions? How can you know if somebody is using your code in violation of the GPL?

Meanwhile, fortunately for PPA owners, a rather good reverse-engineering effort has resulted in a working Linux driver that has been included in several distributions:

http://sourceforge.net/projects/pnm2ppa/

Keep up the good work!

Re:PPA printers

[Bruce Perens]

We need interface documentation first. Given interface documentation anyone can make a binary driver. There is no reason you can't have Open Source drivers, they would just not have the same algorithms HP uses for color correction and resolution enhancement, but what we can do is good enough.

Thanks

Bruce

Re:PPA printers

[Bruce Perens]

The Free Software community is well able to come up with its own color correction. This is a well-known art, with lots of technical papers published on it. You might not get color correction as good as what HP does, but you'd get good enough and knowing you folks, you could improve on the state of the art. What we need is to be able to send raw pixels to the printers, something that does not require the disclosure of HP's color correction algorithm. So, I don't really consider this a reason to keep from releasing a printer driver.

Thanks

Bruce

A fine thing, I suppose

[grant12345]

So I guess this is good news, although it seems to me to be an example of the whole "executives listen to consultants before they'll listen to employees (or g*d forbid customers)" syndrome. Clearly only progress can come from this, so it's a good thing.

In any case, until said progress arrives, people ought to take stock of those companies that offer better support for free software users, and buy products from them. HP makes all sorts of things (reasonable mid-range LAN euipment, workstations, etc) but as I mainly know printers, here's what I know about the industry as things stand right now:

• Epson provides publically available developer information for their printers, scanners, etc. Epson inkjets, therefore, work extraordinarily well using free software drivers.
• Lexmark provides binary RH Linux-x86-only drivers for two representative inkjets. They're klunky, and they're nonfree, but they do work, and they do represent a gimmer of actual Linux support.
• Of the reverse-engineering-reuired inkjets, HP is the best understood, followed by Canon, followed by Lexmark. None, of course, is at all well understood in absolute terms, although the gimp-print folks are rapidly absorbing support and developers from other projects.

As always, if you want to know anything about the state of free software printer support, consult www.linuxprinting.org. Particularly apropos are my vendor scorecards and suggested printers pages.

Good

[Anonymous Coward]

"He criticized HP for holding on to the source code for its printer drivers, and for not releasing printer interface specifications, thus hindering development of drivers ported to other operating systems, namely Linux and the BSDs.

Not only that, but he asked HP either to kill its HP-UX operating system and replace it with Linux, or just Open Source the Unix splinter. He finished up the letter with this warning: "You'll also find that we're rather cynical about ringing endorsements; we've heard those before without result, and they won't earn you a lot of cred by themselves without actions and commitments that back them up."

He's focusing on opening up options for users. Will he have any kind of authority/ear of senior managment? Somebody there must have grabbed him because they have ideas for making open source profitable for HP

Re:HP is doomed, with or without linux

[Ars-Fartsica]

But I'm surrounded by people at work who print everything.

Okay, I'll say it again so you have it clear - its a dead business because margins are incredibly thin. A company like HP can't stay afloat long with printer margins.

Re:Don't forget the other free software OSes

[Bruce Perens]

I am more interested in documenting the interfaces, because I think that the Free Software folks do best writing their own drivers. You will, however, see a lot of Free Software written by HP people on salary.

Thanks

Bruce

Information from the mouth of Bruce

[Livn4Golf]

This appeared on Bruce's site Technocrat.net yesterday. It also links to a Cnet article on the topic.

Bruce sez: "There are two parts to the job. I get to be an activist in the Linux community, on company time, and speak for myself when necessary. And I get to advise top management. There are three people between Carly (the chairman) and I. So, I'll be a pretty effective bridge between the Open Source community and HP management." Here's the link.

Re:HP history and pnm2ppa - from a core developer

[Bruce Perens]

I will be in touch with you folks, and will try to do a bit better than you've seen so far.

Thanks

Bruce

Re:Raph's patents on color halftoning

[grant12345]

Indeed. It's particularly interesting when you observe that the gimp-print framework will be able to print--rather well--on inkjets from various competing manufacturers using the same dithering/color code. When free software has leveled the "software-driven color quality" playing field (which really is pretty level as is for the big players, just with lots of duplicated effort to keep up with the Joneses) it will be interesting to see where manufacturers begin to differentiate instead. Mechanism quality? Operating cost? They certainly seem to be running out of room for improvement in the raw dot size and placement arena; the best HP has managed recently is the Detroit route of thirteen different trims for each actual model (most of the others have only half-heartedly followed thus far, but Epson's got a card-reading Stylus, and further daft SKUs are probably going to show up from the rest soon).

I'd love to see a new mechanism arrive without the various disadvantages of current low-cost printing, but heck if I can find anything likely on the horizon. It's just a hard problem to do laserlike black, dye-sub-like photos, and sharp, accurate spot color on plain paper from the same printer... The industry has done wonderful (heck, almsot miraculous!) things by spinning on the inkjet concept, but as a techie, I'd love to see the spinning repeated on another technology.

Good for Bruce

[dragonfly_blue]

I'm personally very happy to see one of the brightest, most motivated, active, and honest people I know find himself in such high demand. Congrats, Bruce!

I have been working with QT lately, and it simply would not have been possible to develop free/GPL'ed software using QT without Bruce's beautifully diplomatic persuasion of Trolltech. He is a true scholar and an eloquent statesman of the first caliber.

HP history and pnm2ppa - from a core developer

[ajv]

I'm one of the leads on pnm2ppa. We have had several leads with HP since I joined the project last year. This is the most comprehensive reply from HP on their position wrt PPA printers.

I've asked some of the original protocol developers and they don't have access to the documentation anymore. I've asked some of my friends who work at HP, and their access to the places where this doco is stored came up empty.

I've asked maddog via his Linux International link (of which HP is also a primary sponsor) to talk to HP for us, but never received a reply. He's a busy dude, so I didn't mind too much.

PPA printers are well supported using pnm2ppa 1.0.4. Usuable versions are in most of the distributions now, and we are FreeBSD/NetBSD/BeOS compatible (and for that matter, cygwin and simple to make under Visual C++). I develop under NetBSD on the alpha, and it's 64 bit clean.

About the last thing I'm going to work on is ghostscript integration. We need some help from the ghostscript dudes as we must calibrate our printers, so that should be fun.

PPA printers do use a lot of CPU time. We feed the printer data that is ready for the print head - there is nearly nothing in the three families of PPA printers. The sheer amount of data is uneconomical from the point of view of how fast you can send data down, and the level of compression we can achieve in the protocol is only moderate in comparison to PS or PCL3e (which is what the other HP deskjets use).

Sure, brief

[Galvatron]

Of course, when you use a term like "brief," with no objective meaning, anything can be considered brief depending on what it's compared to. In this case, I don't think it's much of a stretch at all. "More than a year" is a heck of a lot less than the decades that most career VCs put in.

I think what CmdrTaco was saying was that you did some VC work, but you didn't make a career out of it. I mean heck, I've had temp jobs that have lasted for as much as four months, a year really isn't that long a time to spend in a profession.

As for your comment about the story being rejected when you submitted it, I assume you just wrote that because you are frustrated? Depending on who reads the submission, and what else they've read that day, and what their mood is, anything can get accepted or rejected. For that matter, how long ago did you submit? Maybe they'd already decided once to post this story before your submission came through. I had a story wait in limbo for over a week before being posted once, and another time I had a story get rejected less than ten minutes after being submitted. Besides, don't you think it looks a little less like self-promotion if the article is submitted by someone else?