The Old Joel on Software Forum: Part 3 (of 5)

Standards Schmandards

For some time now I've been getting a little disenchanted with all the standard's worshipping going on in this industry. Every time we turn around there is a new standard telling me how to do x, y or z, or that I shouldn't do x and y together. I'm sick of the standards bandwagon. I think these two articles hit it on the head.

http://www.artima.com/weblogs/viewpost.jsp?thread=4840

http://www.artima.com/weblogs/viewpost.jsp?thread=4892

Mike
Monday, December 15, 2003

I think the fact that this comes from a senior engineer at Sun does not bode well for them.

christopher baus (www.baus.net)
Monday, December 15, 2003

Especially since many "standards" are really just recommendations.

Matthew Lock
Tuesday, December 16, 2003

>> Especially since many "standards" are really just recommendations

...Or requests for comments.

JT
Tuesday, December 16, 2003

I think that, for the most part, standards have helped move technology forward. For the most part, a standard gets adopted when useful and ignored when not. Both proprietary and open standards have come and gone or completely been ignored based on usefulness.

Standards and politics? Hmmph. Blog, schmog. Controversy about nothing.

hoser
Tuesday, December 16, 2003

Ever imagined being an IT manager in an organization with more that 40 different applications? Ever had some that could not import or export any data and the users had to keep retyping data where a simple import/export operation would have worked? Ever bought some expensive software from a vendor that soon went under?

Ling
Tuesday, December 16, 2003

The thing I like about standards is that there are so many to choose from.

Gord Schmidt
Tuesday, December 16, 2003

The author is absolutely right.

Standards work best when they are codifying an existing and working implemention that has become a de facto standard.

If you look over the history of standards, the VAST majority of those that are actually good and useful (eg. TCP/IP) were built first by small groups and then released as standards to the wider world. Contrast this with the way most standards are created now -- by having a group of companies (usual competitors and oftentimes with differing interests in getting certain techniques their company has a patent on in the standard) hold meetings for 2-5 years, producing a document that describes a system so byzantine that the system they are standardizing is never implemented in full by anyone, because it is just too ridiculously complex and in some areas nonsensical.

In short, I'm all for standards, but the standarization process as practiced these days is highly flawed and more often than not results in big, ugly standards that try to be everything to everyone and in the process wind up being next to worthless for everyone.

Mr. Fancypants
Tuesday, December 16, 2003

Yeah, anyone seen SOAP lately? That standard and all the related standards basically make it so that you have to buy into a full-blown tool-kit if you want to implement them in your application. And not just a library, a whole framework is needed.

Saruman
Tuesday, December 16, 2003

The nice thing about standards is that there are so
many to choose from...

foobarista
Tuesday, December 16, 2003

It's the standard for there to be multiple conflicting standards. You should refer to RFC 345457578567345 section Q:

If something can be done one way that is obviously superior, there should be two competing standards - neither of which is the obviously correct way.

www.MarkTAW.com
Wednesday, December 17, 2003

In the event of there being an obviously technically superior standard it should only be introduced after a techically inferior standard has become the de facto standard and is practically impossible to displace thus fueling an unwinable quasi-religious war.

John Ridout
Wednesday, December 17, 2003

The nice thing about posts that say "The nice thing about standards is that there are so many to choose from... " is that there are so many to choose from.

Ged Byrne
Wednesday, December 17, 2003

This is just another volley in Java vs .Net war. MS pushed a lot of .Net pieces, such as C# and CLR, through some standardization body (ISO?). This way they can claim it's not proprietary. Ha. On the other hand, Sun has steadfastly refused to move Java to a standards body, a la C++, instead sticking with their JSR process.

In general, there are two type of standards. The first kind codify things that actually work. The other is just a marketing tool - something to claim on your "features and benits" sheet or "marketecture" to freeze out a competitor. Sometimes, a version of the latter kind of standard will turn into a real standard. Most of the time, though, it's just fluff.

igor
Wednesday, December 17, 2003

Radical thought:

Some standards are useful, some not. Use the useful ones.

Jim Rankin
Wednesday, December 17, 2003

(...which seems to be the point of these articles)

Also, a hypothesis:

A standard's usefulness is inversely proportional to its size.

Jim Rankin
Wednesday, December 17, 2003