In the past year, Red Hat has gotten deeply involved in two projects- mugshot and OLPC- that get them outside their comfort zone (enterprise Unix-y operating systems) and into new territory. Here is my free advice on the next outside-the-enterprise-box project they should take on- one that would make the world a better place, keep them close to their core competencies, make them some money, and get their foot in the door in a critical market, all at one time.
I think that RH should go into voting machines.
Eh? Voting machines are not sexy. They’re dull. They’re duller than dull. But they are very important to democracy (a growing market! ;) and hence the world, and they are going electronic- with currently embarassing-to-criminal results.
So why should RH get involved? First off, electronic voting matches up well with open source’s traditional strengths.
- security- As Ed Felten demonstrated spectacularly yesterday, the current generation of electronic voting machines are painfully insecure. Go watch the video. Open souce security auditing can do much better than that. (Diebold’s defense, by the way, is that Felten should have asked them for more information. That would not be a problem in an open source context.)
- cost- Governments are fairly price sensitive, especially in low-profile areas like voting. Open source is traditionally very cost competitive, and in this particular case, the closed-source systems have to license components like WinCE, so they are definitely at a disadvantage.
- pre-existing community- Corporate-sponsored open source work does best when it works in hand with existing bodies of volunteers and expertise. Such groups already exist in open source voting; open voting consortium is the first hit on google but I believe there are others as well.
- political motivation: one of the most tried and true ways to motivate open source contributors is to give them a bad guy. Voting fraud is replete with bad guys on all sides; if a project got enough backing (i.e., RH) to make it look like it might get actually used in an actual election, people would come out of the woodwork to audit and patch it.
Secondly, voting plays specifically off RH’s strengths:
- ‘enterprise’ understanding and sales- governments are enterprises; they like being told about how enterprise-y their software is. RH is very good at this; telling them that Fedora E-Voting (or whatever) was Enterprise Quality shouldn’t be hard.
- hardware support- as demonstrated by OLPC, and historically by RH’s long tradition of working with server vendors, RH is good at getting Linux going on new hardware. If RH could leverage the OLPC platform it could conceivably lead to very nice economies of scale, with a platform RH is already an expert on.
- complex systems- RH is good at pulling lots of disparate pieces together. Obviously a comprehensive e-voting system is complicated and would require lots of engineering talent- which RH has in spades.
Finally, a voting project would fit needs that RH should be seeking to fill:
- High profile in a critical market: practically every government in the world wants secure e-voting; what better way to get your foot in the door for later sales of servers, desktops, etc.? Combined with OLPC, this could be a double-whammy for RH marketing to governments.
- Continued expansion of the Linux footprint. Anything that benefits Linux benefits RH, and making Linux the secure standard for one thing that every citizen should do would make a huge mindshare impact for Linux in general.
- Revenue: after the various voting fiascos in 2000 and 2004, lots of governments are spending money on electronic voting right now. RH should get a piece of that pie.
So, yeah. E-voting. Have a blast, RH, and drop me a note when it is making you piles of cash and we’re all better off because of it :)
[Tangentially, it may be worth noting that OLPC and Mugshot don't necessarily match up well with all these things that I've noted should be important to RH, or that RH is good at. In particular, mugshot is not something at all within RH's comfort zone; instead, it is a deliberate attempt to grow RH's skill base. They are experimenting with new design techniques; they are experimenting with a new business model (server-side services); and they are experimenting with a category (social software) that open source has never really tried to do before. It isn't bad to be experimental- I do hope they succeed- but given how far they are from RH's traditional strengths, it will be an uphill battle. OLPC, by contrast, plays off a lot of these strengths, even if the glamour part- the UI- is fairly different from what RH has done in the past.]