j_random_hacker

15 lat temudiscuss

I love the concept of sharding counters to trade write performance for read performance! As so often happens, my understanding advances by looking at something I thought I knew the only possible answer to ("Of course a counter goes in a single variable!") and seeing that other alternatives are possible.

j_random_hacker

15 lat temudiscuss

You are worrying about correctness of concurrency. Yes, this should always be worried about.

The parent poster is worrying about performance of concurrency. This doesn't need to be worried about unless you are one of about 5-10 tech companies whose name is recognisable to people on the street.

j_random_hacker

15 lat temudiscuss

I didn't like this idea at first, since it struck me that if for some reason you had to create many records in quick succession (e.g. some kind of bulk upload) they would essentially be placed in random order. But thinking about it, the only scenarios where you would need to do this and want to preserve the order exactly would be if all the records were coming from the same source -- in which case you could just choose a single UUID for the batch and append increasing values of a local counter to this.

TL;DR: This would actually work really well I think!

j_random_hacker

15 lat temudiscuss

My take on that comment was that it was not a subtle attack at all -- it was merely an observation that it's not a logical necessity for the leaders of a large country to be extroverts.

j_random_hacker

15 lat temudiscuss

Thanks, you're quite right. Having now skimmed the paper "Does an interface with less assistance provoke more thoughtful behavior?", I would say that the measures of how many moves were made or remade, and how long was taken before the first move, are basically irrelevant, as the only quantities of genuine interest are how long the solution takes, and how well the subjects did on the questions afterwards. As you say, there was no significant difference in average solution time -- even though the low-NFC group did on average 19s better on the externalised version. Nor was there a statistically significant difference in either of the question types (although one type was close to favouring the internalised approach at p = 0.06). Maybe the other papers are more persuasive.

j_random_hacker

15 lat temudiscuss

Totally agree. I didn't read the paper, but I doubt they have modeled the fact that people have a finite amount of time that they can invest in internalising stuff. IOW, (I presume) they have not looked at the opportunity cost of internalising UI details, and often that time is better invested in other pursuits.

comments