Reason #8009 to opensource code

I'm in North Carolina, so I'm staying in a nice hotel. My body is still on Israel time so I wake up at 2AM, and enjoy some complimentaru tea in the lobby every morning. The night desk guy and I talk- there isn't much to do in Carrboro at such hours.

I went over some of my computer programs with the night desk guy. This morning he said I was " inventing Startrek like stuff" . I told him, my work is really not so complicated, and if he learned to code even at the level I do, he could do similar work too. He said he would think about it, but then he sighed 'it's hard right?'. I warned him that increasingly computer systems run things- and if you don't understand the code behind these systems, you are at a disadvantage.

What will be even harder than computer programming will be having your life run by things you don't understand. I explained to my new friend how machines can make decisions based on data. We went through some examples..and some 'lightbulbs' went off in his head. He realized that the computer program his hotel uses to allocate rooms was a good example of a certain kind of algorithm. He also had noticed that the program often registers a guest who is disabled, he inputs the information including disability, and the computer program still puts them in a totally inappropriate room. "You can fix the algorithm" I told the guy.

Then I thought about it a bit more. I could fix the algorithm, so certainly to people who made the system could have. This system or some variation, I presume is used at all of the hotels of this particular group/corporation. That means hotels on multiple continents that range from mid-range to luxurious are probably running similar code. Something tells me no guest in the 5 star hotels ever gets allocated to the room that doesn't fit them- disability or not. So clearly, the chain of command for this corporation made some kind of calculation it simply wasn't worth fixing in the mid-range hotels in all regions.

And here is the scary thing- I can see the same thing happening in terms of computer programs in medicine. What do we expect if we have black-box algorithms, and varied populations? Are they really going to re-tool an algorithm for some small population of people with limited resources so it's fair?

I recently read an old paper which stated that ALL medical software should be open-source. I actually don't agree. There are cases where proprietary software may be the way to go...let's face it, medical people need a lot of support once they have implemented a software. But if we are talking about an algorithm that deals in diagnosis or clinical decision support, perhaps at least some explanation of the algorithm should be available to the patients.