Ping ([info]zestyping) wrote,
@ 2006-05-22 17:21:00
Previous Entry  Add to memories!  Tell a Friend!  Next Entry
“Performance” considered harmful.
Computer science research papers, industry literature, and even end-user interfaces are littered with the term “performance.”

In all these cases, what is really meant by “performance” is speed. “High performance” is just a high-fallutin’ way of saying “fast.” But being fast is certainly not the only way in which a computer can be said to perform. In fact, a quick look at the primary dictionary definition shows that it’s a rather poor choice of word:

performance, n.
1. a. The accomplishment or carrying out of something commanded or undertaken; the doing of an action or operation.
Oxford English Dictionary

I don't see AS FAST AS POSSIBLE in there anywhere.

I suggest to you that there is something fundamentally misleading about using a general term such as “performance” to refer to a single specific factor such as speed.

Why does this matter? As long as we continue to abuse the word “performance” like this, it reinforces the ingrained idea that speed is everything. That usage is a relic of the 1970s; other concerns have been more important than speed for quite some time now. Corrupting the word “performance” hampers our ability to talk about actual performance — as in, computers doing what they’re supposed to do and doing it well. Being dependable. Operating smoothly and correctly. Recovering from problems. Not falling over suddenly.

In other contexts, that’s what “performance” usually means. When you go to see a musical, a “good performance” includes the whole experience — the acting, the singing, the music, the engagement of the audience. When we’re talking about cars, “performance” refers to reliability and control as well as speed. If you hit the brakes and the car doesn’t stop, that is poor performance.

So, here is my humble request of you. Don’t use the word “performance” when you are really talking about speed. If you mean speed, say “speed.” If you feel weird about using the word “speed” because it is a less important-sounding word than “performance” or because it sounds like less of an overarching design priority, that’s because it is. That’s all the more reason to be candid about it.

Go ahead — take this little badge and pass it on. And the next time you write a paper or an article about computers, please think twice before you write “performance.” It’s easier to break the habit than you might think.

Update: Here's a PDF if you want to print it out or make other designs.

Another update: It’s a good thing most people don’t apply the computer science definition of “performance” to sexual performance. Phew!



Page 1 of 2
<<[1] [2] >>

(Post a new comment)


[info]boredatheist
2006-05-23 12:33 am UTC (link)
Awesome. I'm going to print that and tack it to my cubicle. I take your words to heart, as I feel like my current project is essentially broken at the time of deployment due to constant "performance" tweaks and rewrites that really just break shit. Fear micro-optimization. Love simplicity and correctness.

*hugs Ping*

(Reply to this) (Thread)


[info]zestyping
2006-05-23 12:43 am UTC (link)
Sweet.

(Reply to this) (Parent)


[info]tedesson
2006-05-23 01:04 am UTC (link)
The Python sprinters have it right, with the "Need for Speed" sprint.

Nice analysis. Right up there with the question "Exactly what are you optimizing?"

(Reply to this) (Thread)(Expand)


[info]zestyping
2006-05-23 06:11 am UTC (link)
Great question.

(Reply to this) (Parent)

(no subject) - [info]akuchling, 2006-05-23 12:31 pm UTC (Expand)

[info]aklikins
2006-05-23 02:33 am UTC (link)
Even worse is the growing use of performant to mean "fast".

(Reply to this) (Thread)


[info]zestyping
2006-05-23 05:57 am UTC (link)
Ugh. Fight it!

(Reply to this) (Parent)


[info]androgy8
2006-05-23 06:00 am UTC (link)
I think "performance" is one of those key words like "leverage" and "interface" that get thrown around a lot because they supposedly sound more impressive than other words.

(Reply to this)


[info]hukuma
2006-05-23 07:32 am UTC (link)
When we’re talking about cars, “performance” refers to reliability and control as well as speed. If you hit the brakes and the car doesn’t stop, that is poor performance.

I disagree; any car enthusiasts who "improved the performance" of their car aren't talking about tuning up the brakes or getting better mileage — they're talking about making the car go faster. In computer science, where entire subfields have the word "performance" in their title meaning speed, I think your campaign is a losing battle.

(Reply to this) (Thread)(Expand)


[info]zestyping
2006-05-23 08:17 am UTC (link)
In computer science, where entire subfields have the word "performance" in their title meaning speed, I think your campaign is a losing battle.

That others use a poor word is no reason not to choose better words in your own writing. It doesn't matter whether it's a winning or losing battle, because saying "speed" instead of "performance" costs you nothing — no one is going to misunderstand what you mean if you say "speed" when you mean speed.

The question is, what do you want to express — do you want to promote the idea that speed is everything or do you want to be clear that you're only talking about speed when you talk about speed?

(Reply to this) (Parent)

(no subject) - [info]jamiemccarthy, 2006-05-24 12:39 pm UTC (Expand)
I think you are wrong hukuma. - [info]kel57, 2008-03-24 10:57 pm UTC (Expand)

(Anonymous)
2006-05-23 01:09 pm UTC (link)
Also, performance can be good or bad. It can't be high or low.

(Reply to this)


[info]eqe
2006-05-23 03:38 pm UTC (link)
Ah, but if they advertise "speed", then they might be liable if their product isn't faster than the competitor's. If they advertise "performance" there is more weasel room.

(Reply to this) (Thread)


[info]zestyping
2006-05-23 09:14 pm UTC (link)
Good point.

Okay, perhaps marketroids aren't a good audience for this rant. It would be nice if academics and journalists thought about this, though.

(Reply to this) (Parent)


[info]don_rodent
2006-05-23 07:57 pm UTC (link)
Hello!

It sounds like you have quite the "issue" with this word.

Not a ruggie!
Goodbye...

(Reply to this) (Thread)


[info]zestyping
2006-05-23 09:13 pm UTC (link)
I have an issue with a particular use of this word. There are contexts where "performance" is a perfectly good word to use.

What's a ruggie?

(Reply to this) (Parent)(Thread)(Expand)

(no subject) - [info]don_rodent, 2006-05-24 03:45 am UTC (Expand)
Performance
[info]r6
2006-05-23 11:42 pm UTC (link)
I think performance is a perfectly reasonable word. More performance means accomplishing more tasks. With computers this can be done by increasing speed, or it can be done by increasing parallelism. Adding a second CPU increases performance without increasing speed. In fact, it seems to make more sense to compare performance between computers rather than comparing speed.

(Reply to this) (Thread)

Re: Performance
[info]zestyping
2006-05-23 11:52 pm UTC (link)
What you are talking about is called "throughput."

(Reply to this) (Parent)(Thread)(Expand)

Re: Performance - [info]r6, 2006-05-24 01:22 am UTC (Expand)
Re: Performance - (Anonymous), 2006-05-24 09:25 am UTC (Expand)
Re: Performance - [info]zestyping, 2006-05-24 09:33 am UTC (Expand)
Re: Performance - (Anonymous), 2006-05-24 01:23 pm UTC (Expand)

(Anonymous)
2006-05-24 01:57 pm UTC (link)
Plus, "speed" is fewer characters to type than "performance", so switching to it increases writing speed. :)

-anders

(Reply to this)

Could be worse
(Anonymous)
2006-05-24 02:59 pm UTC (link)
What really grates with me is the "word" used to describe something with good performance by some people: "performant". Ugh.

(Reply to this)


[info]amoken
2006-05-25 09:32 pm UTC (link)
I spent a lot of time a couple weeks ago working on memory performance.

(Reply to this) (Thread)


(Anonymous)
2006-06-27 12:05 pm UTC (link)
Hi - I need a better word for leverage - I am trying to say "we use what we have better/ more effectively/ cleverer/ and we took what we had and made it work better too"

Any ideas?

Liked the “performance” chat so thought this was a good place to ask.

Cheers

(Reply to this) (Parent)(Thread)(Expand)

(no subject) - [info]amoken, 2006-06-27 03:25 pm UTC (Expand)
(no subject) - [info]gowylie, 2007-01-18 02:51 pm UTC (Expand)
gelinlik - (Anonymous), 2009-02-05 09:05 pm UTC (Expand)

(Reply from suspended user)

(Reply from suspended user)

(Reply from suspended user)

(Anonymous)
2008-01-11 11:39 am UTC (link)
hiiiiiiiiiiiiiiiiiiiiiii

(Reply to this)

Nice article
[info]zuluboy
2008-02-11 03:21 pm UTC (link)
Nice article. What you say is true.Speed not equal performance.

regards

Ideaman

(Reply to this)

(Reply from suspended user)

(Reply from suspended user)

(Reply from suspended user)

[info]tatarlarda
2008-05-16 06:49 pm UTC (link)
Tatil Yerleri
nice share. thanks for the information.

(Reply to this)

(Reply from suspended user)

(Reply from suspended user)

(Reply from suspended user)

(Reply from suspended user)

(Reply from suspended user)

Page 1 of 2
<<[1] [2] >>

Create an Account
Forgot your login?
Login w/ OpenID
English • Español • Deutsch • Русский…