(March, 2011 update: A couple of people have taken my comments about the acquisition of Cilk as being critical of the company Cilk Arts. They are not. Cilk was a great place to work. Intel was not.)

I used to work for Cilk Arts, a startup formed to commercialize the Cilk programming language from MIT.

Cilk Arts was acquired by Intel in July, 2009. While financial terms were not disclosed, this was not one of those everybody-gets-rich deals. I'm not complaining about that -- worthless stock options are a customary risk of startups -- but don't think I am sitting on a mountain of money feeling sorry for myself.

I left Intel in May, 2010. Seven ex-Cilk Arts employees were acquired in 2009. I was the fourth to resign. I was very nearly the first.

This is part of the reason why I quit. Two missing sequence numbers are internal Intel matters that can remain that way.

1. The acquisition of Cilk Arts was super-secret, don't-even-tell-your-family secret, for no reason. If the only options were super-secret or public then Intel should have issued a press release once the preliminary deal was signed a month before close. I learned later that people inside Intel with need to know were not informed either.

2. Intel wasted my summer. I spent July and August in the office doing nothing. There was no meaningful work to do but I had to be around all the time in case something came up. I was not able to take a summer vacation like I planned and I started off with no accumulated vacation at Intel.

3. Intel's acquisition team told us all to sign a blatantly illegal noncompete agreement. It was so grossly in excess of what the law allows that it shows bad faith on their part. I was very close to walking away. I negotiated it down instead. I should have walked away.

4. At Intel's first briefing to the team shortly before the deal closed we learned that the business plan was to cancel the product that we had spent two years building and had started to ship for revenue. Intel wanted to keep Cilk technology off the market for a year or two until Intel could come out with its own product. Again, it was tempting to walk away but I didn't.

5. Intel moved us to DEC's former ZKO facility in Nashua, New Hampshire. I had worked there in 1999 and knew the commute was too long. I gave myself six months to decide to move to New Hampshire or find a new job.

6. Remember all the secrecy? It takes two days to create an Intel network account. Somebody started the process three days before our start date, and when the account was ready one day early it was automatically deactivated as not belonging to an active employee. This was a sign of things to come.

7. Remember all the secrecy? Nobody was ready for us. We didn't have access to information we needed to do our jobs. We were functionally part of two organizations but administratively part of only one. If there was a plan to integrate us it was not even close to adequate.

8. We had the same problem as when I was acquired by Cisco: Too many cooks. Intel had several groups all wanting to be in the parallel programming language business. The West Coast compiler team claims right of precedence in language issues. For example, we were recently scolded for using the term "Cilk team" to mean the people who worked at Cilk Arts. If not the Cilk team, what are the people who created the language? Better seen and not heard, apparently.

10. There is a project planning spreadsheet. We were not on it in any meaningful way. Frederick Brooks pointed out in The Mythical Man Month that you typically find out two weeks in advance that a deadline will be missed. You need intermediate milestones on a long project to detect schedule slips early. We had none. We found out two weeks before Intel's Cilk implementation was supposed to be ready that it would not be, because all the features with milestones or with support inside the compiler team had priority.

12. See item 4. Intel buys things in order to break them or keep them away from competitors. Intel bought the former DEC compiler team so nobody else could have it, then set out to destroy the team to make sure it didn't reform. Intel has a large work force in Russia. They are former Soviet weapons people and their main value is being off the global job market. We moved into an environment where we did not feel valued by the company.

13. See item 4. Intel's compiler may get design wins but it won't get mindshare. A few big customers will buy it. The vast majority of the programming world will have no idea it exists. Cilk Arts was working to change the way people program.

14. I had an idea for a project and had to give it up when I saw how much of a barrier Intel's legal and management process presents to innovation. It took three months for the Cilk team to get approval to release Cilk Arts' product as a free "what if" technology demo download under the Intel label. That was with support from a few levels of management. Anything I worked on would be left to rot within Intel and, worse, my ideas would be Intel intellectual property preventing me from working on similar things in the future.

15. Intel wants to be a monopoly. I don't like Intel. I don't want Intel to destroy everything that might be innovative in the computer world to shove the x86 architecture down our throats.

16. One of Intel's core values is following the "letter and spirit" of the law. After watching the antitrust cases I do not believe that company practice follows company policy.

17. One of Intel's core values is "a great place to work." It is not. I have some reason to believe the environment at some other sites was better, but we were in Nashua.

18. I don't like Microsoft Windows. I was warned before I joined that that attitude was "career-limiting."

19. Intel's computer security policy gets in the way of doing my job, and gets in the way of other things too. For example, I had trouble working at home because I use Unix and can't use laptop keyboards. All good corporate citizens love their Windows laptops which have VPN software. I could come up with a long list of little complaints too. Firewalls, email retention policy, etc.

20. Last on this list, but certainly not least, is the essence of work at a megacorporation. Intel's internal employee web site linked with approval to the company's top 100 best companies to work for listing. So I will hold the company accountable for the reason Intel is supposed to be a good company to work for: "It is a magnificent feeling to be a cog in such an important machine." I do not want to be a tiny cog in a big machine.

Despite all that, Intel did not kill Cilk. There is a beta version of Intel's compiler with Cilk support. Interesting as the underlying technology may be, I will not be a tiny cog in the massive machine that sometimes grinds out a new version.

Some people have suggested that various of these things are inevitable when working for a big company, or when being acquired by one. That is not an excuse. I do not have to work for a big company. Small teams get things done. On June 1^st I will join one.