Friday, August 26, 2005

Bruce Eckel on Consulting

Great post from Bruce Eckel on consulting. His definition:

"I think consulting is when you have some kind of special expertise -- come by through hard struggle and learning -- that you transfer to a group of people, in a relatively short period of time, and in a way that is unique for that group. I also think that consulting involves addressing particular issues faced by that group. "

He also addresses a very common industry phenomenon:

"It's tempting to make the analogy to a ponzi scheme, but it's not that bad. Both forms of a consulting firm (first-class consultants only vs. high-tech body shop) have value. The problem is the temptation to present the second form as the first, in order to apply the same high fees of the first-class consultant to each additional body added to the shop. The goal of the company becomes "how do we transfer the aura of authority from the high-image consultant(s) to anyone who works for us, so that we can charge the highest fees possible?" Or to simplify, at some point the bean-counter mentality takes over and the mission statement of the company goes from "how do we provide the greatest value to the customer?" to "how do we charge the highest fees possible?" (You can argue that this is the fundamental shift that any publicly-held company goes through. After all, a public company is legally beholden to maximize shareholder value, so how could it be otherwise?)"

I have noticed this myself, when I was working in Boston during the tail portion of the dot-com years. I was chatting to an Indian body shopper who was a sub-contractor for Andersen Consulting; While Andersen made $450 plus per hour, this sub-contractor got about $70 - You can imagine how much would have trickled down ultimately into the pockets of the programmers, who were paraded as Andersen consultants to clients.

Another nugget from the article:

But advice that appears to be rational does not necessarily solve real design problems, as we saw in the "structured design" movement of the 70's. This seems to happen over and over in our industry -- it's easy to focus on one aspect that seems to be "the solution" and miss the big picture, and to produce unintended consequences that eliminate the benefit of what may seem so clear in isolation.

Slash Wisdom

From an "Ask Slashdot" posting titled "Uneducated IT Managers, and How to Deal ?", I found this gem posted by Pope Benedict XVI ;-) :

"It's just about impossible to find a job working for someone whom you respect. "

To me, that about summarizes the crux of the problem with working in the IT industry. But strangely this nugget of wisdom has been marked as funny, instead of insightful, or some such. Setting the filter at four and reading reveals more insightful comments. Slashdot, for it's flaws, sometimes throws up insights that knock out your breath momentarily, or better still has the reader in splits - Some of the trolls are outrageously funny without being crude or offensive.

Wisdom of the commons, or shall we say, Slash Wisdom !

Monday, August 08, 2005

Joel on Software Writing

A few interesting points ( mostly paraphrased) from a forty minute interview with Joel Spolsky on software writing:

  • Most software writing is awful. There is a high correlation between successful software and writing.
  • Programmers are used to writing code for compilers; compilers don't care about order. Humans comprehend stories - can't give them a bunch of definitions and tie it together. Most programmers, who write for compilers , don't know how to write for human brains.
  • There is an awful lot of writing there about technology, even internal functional specifications that just doesn't get read, because regular people don't understand it.
  • At Fog Creek software (Joel's company) they ask for writing samples, not necessarily technical stuff from all candidates. Because we (Fog Creek) really do find that the ability to communicate clearly and effectively, is the crucial difference between an acceptable programmer and a great programmer that can work on teams, and get their ideas understood by the world.
  • Once you make software that mediates between humans, then the design decisions in the software have anthropological implications. In Joel's latest book, Danah Boyd talks about some websites that require a strict social relations - such requirements are pathological, not normal. People are actually creating software and websites that require you to conform to a set of behaviors that in the real world would get you incarcerated.
  • Paul Graham's assertions about great hackers - they use open source, they don't use Java etc flies in the face of facts. But his essays are beautifully written. Joel said his hands start shaking whenever there is a new post on Paul Graham's site.
  • Phil Windley asks an interesting question about whether it is opninions that make software writing interesting. Joel says that's where the value is being added (in opinions) and goes on to add that with great respect to the American journalistic tradition of objectivity, the only thing it results in is lot of cnet's news.com, that is very bland, that reproduces a lot of press releases , or they get someone from Sun to say something about Microsoft's announcement.
  • According to Joel, the number one rule of technical writing is, "Show, don't tell". Technical writers violate that rule so often. So much technical writing reads like mathematical proof - all the information is here, if you can decipher it, you can understand it. That kind of stuff doesn't get read unless it has to get read. The stuff that people read are the ones with stories in it. Human beings are story tellers. We have been telling stories around camp fires for hundreds of thousands of years - that's what out brains are good at and that's we enjoy, that's why TV is popular and that's how messages get into our head.
  • When people started using PC's there was lot of piracy, lot of the books were just shovelware. Companies were trying to make the books as fat as possible. At one time software could not be used without books. It was not until the GUI generation and much later, that software companies realized that they don't have to produce the manuals at all. These days some of the best books are the ones written to explain the domain in which software is used. Quick Books' manuals, for example, teach you a lot about the basics of book keeping.
  • Amazon ratings are not accurate and are not done by professional people ( I personally disagree with this sweeping characterization). Joel said that there were folks with axes to grind who wrote reviews on Amazon and gave the example of a Microsoft techie, whose book got bad ratings, because he used to be aggressive in some news groups.
  • If you are a computer programmer, put yourself in another person's mind - what do they know at any given time ? Try to answer the questions in their minds in the order in which they are having it. If you write a document the reader first with what the document is about, and start answering the questions that they might have in their minds.
  • Some of the great functional specs are written like a story.
  • Blogging is a good way to exercise the writing muscle. Joel noted that he had gotten a bit over defensive in his writing of late.

Saturday, August 06, 2005

RPC

Your brain does not use RPC to talk to your pancreas.
-Werner Vogels, CTO, Amazon.com

The talk titled E-commerce at the interplanetary scale is good, but a lot of it went over my head, especially the stuff about Epidemics; but he did admit that he was a recovering academic ;-) .