PC timing problem

Lasse Langwadt Christensen wrote on 6/27/2017 2:44 PM:

Who said anything about reaction times? We don't know what they are measuring really. It could be something that doesn't require the person to push a button. I can't imagine that hasn't been studied to death.

--

Rick C

Yes, I understand that. But 1 ms accuracy is noise here - both for reaction time measurements, and for sprinters. And I know that in top-level sprint races, people take timings with 1 ms resolution - but it is pure noise - pure luck. Even at the 10 ms level it is noise - there are tens of milliseconds differences in the times that each competitor hears the starting gun.

For measuring reaction times to visual stimulus, 10 ms resolution is more than accurate enough.

(Having said that, now that you have mentioned neurons - if the customer here is measuring electrical activity in neurons as well as the time taken for a reaction, then 1 ms resolution is appropriate.)

1 ms is maybe not a bad place to start, I agree. But if a customer came to me and said "I want a 1 ms accurate timing of people's reaction to a picture on a screen", I would suggest that they lower their requirements to a more realistic 10 ms.

Even before you asked how large their budget was? I don't try to change my customer's expectations unless there is a clear reason. This is not one that can't be met, so I would just give them what they asked for. They are in the business of being customer, not me.

--

Rick C

Clearly this is not something that can be boiled down to a few fixed rules that can be described in a paragraph on a Usenet post - how you handle customer requests is going to vary greatly between customers.

In general, however, I would say that my aim with a customer is to have them end up with the most appropriate solution for their needs. It is rare that this matches 100% with what they initially ask for - customers generally have a rough idea of what they are look for, but are missing many of the details and subtleties. Sometimes they walk out of the initial brainstorming meetings with a completely different solution from the own they specified when they came in.

In this case, I would be likely to say that 1 ms timing is not possible on a Windows machine with a normal screen, but 10 ms timing is achievable with a quality gamers screen. If they respond with "we have lots of money, we want 1 ms" then I would /still/ say it is impossible on a Windows machine with a normal screen. But I would be happy to do them a Linux laptop (using the backlight trick discussed elsewhere) as an alternative. If they respond with "we have lots and /lots/ of money, and we want 1 ms timing on any Windows system with normal screens" - then I would tell them we can't do it and I don't expect anyone else to do it either.

Some development companies work on the basis of getting a detailed specification from a customer, giving them an offer for the time and cost of implementing it, then do the job. For us, and for our "typical" customer, the contact usually starts a lot earlier during idea, prototyping, and specifications stages. My customers expect me to tell them "your idea sucks" (but politely, of course!) if necessary.