Hacker News Reader

Comment by apgwoz | Hacker News Reader

If you're assuming this based on some microbenchmarks, you should probably instead reevaluate your evaluation techniques.

(This isn't meant as a personal jab, but too many people focus on silly things like microbenchmarks when choosing tools, and not on other more important things like support avenues, tooling, libraries, etc.)

View on HN

I don't think it's fair to assume that TheMagicHorsey was basing his evaluation on the microbenchmarks themselves.

The real story here is that people are working on a number of new libraries/modules/tools for doing web development in CL — the benchmarks (as you correctly point out) are a minor detail.

Perhaps you should read more carefully. "If you're assuming ..."

It's just not clear from the context we have. But, caution! Microbenchmarks should only be Microtrusted.

I think this is truly sagely advice.

My grand takeaway from Aphyr's Jepsen series (http://aphyr.com/tags/jepsen) is that the more sophisticated software becomes in the never-ending quest for features and performance, the more likely the fundamentals underlying it have been overlooked, glossed over, or ignored completely.

When you compare the time fukamachi's spent on these libraries to the time spent on Clack, Caveman, etc., it's pretty clear there's obsession with performance.

As aroman said, "The real story here is that people are working on a number of new libraries/modules/tools for doing web development in CL".

I didn't mean to imply that the original post was not worthy in anyway whatsoever, in fact I think all of the work he is doing in the CL ecosystem is awesome and I look forward to playing around with it soon in my next CL project.

But apgwoz's general point about how we evaluate software is spot on. In general, I think benchmarks and the like have a sort of "woo"ing effect on people (in the James Randi sense), and we should probably take a few moments to reflect on that before making any major architectural decisions. :)

I agree entirely, but at the same time, I think too often people make social rather than technical decisions. Lots of stars on GitHub and a well-designed website are usually indicative of a good project -- but people put too much faith on CSS, the same way they put too much faith on microbenchmarks taken in laboratory conditions.

This -> people make social rather than technical decisions

Leo Meyerovich did a study on the social factors that affect language adoption[0]. Vivek Haldar has a highlights the salient points[1]. More info at his site: [2]

[0]: http://lmeyerov.github.io/projects/socioplt/papers/oopsla201...

[1]: http://blog.vivekhaldar.com/post/56972066006/empirical-analy...

[2]: http://lmeyerov.github.io/

I think too often people make social rather than technical decisions

Indeed. It's sad that as engineers (who theoretically espouse logic, reason, and evidence) we still end up making so many decisions based on herd behavior.

I would posit that engineers are human creatures rather than robots, and basing decisions on social factors is a natural part of being human.

For my anecdotal experience, apart from JS libs (where style over substance is frequently the norm) I find a well-designed website is usually a sign the project cares about its users, and is more likely to provide good documentation & a space to ask questions.

Reply on news.ycombinator.com