[aragot]

Yes the most interesting revelation in Cringely's rumor was that IBM made software.

Let's be honest: An big IT product is about 25 programmers and 25 diverse people (incl marketers, designers, accountants) working for 3-10 years. With this you make a GitHub, a Word, an IE. IBM could be owning the planet today. But no, they're in the business of hotels, golf, and conferences.

[AlisdairO]

They do make many billions profit on software every single quarter. Who cares if the market is not very sexy?

If you think there's only 25 people working on Word, I think you're off by an extremely large margin.

[easytiger]

They make billions by corporate lobbying and political marketing

[AlisdairO]

I wasn't commenting on IBM's methods - although I think attributing the entirety of their profit to scammery is extremely far-fetched - but on the choice of market. My comment was simply to point out that I don't think there's any shame in targeting boring market segments, and that there's clearly plenty of money to be made.

[aragot]

If you planned on making a competitor of Word, would you copy all its features and throw 5,000 people at it? Of course not.

No, I don't think Word is 25 people. But a relationship of mine made an Excel in js with 3 people and sold enough copies.

Before you say I'm wrong again, no this Excel-in-js isn't isofunctional. It's just what the customer needs.

Point being: You can make great software with few people, and IBM could do it easily.

[AlisdairO]

> Before you say I'm wrong again, no this Excel-in-js isn't isofunctional. It's just what the customer needs.

isofunctionality is implied by saying you could make 'a Word'. I wouldn't dream of disputing that you could make a word processor that offers some limited subset of Word's functionality with a small team - but that's not making a Word-class product, it's making a simple word processor.

[aragot]

I agree we can't reproduce Microsoft Word with isofunctionality with 25 people (obviously). I agree my past comment made it sound like that.

I currently feel negatively from the readers, so I'm testing that again: I've written a comment below with figures and sources to back my claims - See comment https://news.ycombinator.com/item?id=8956544

[CHY872]

I think you have an incredibly short-sighted view of the difficulties associated with software engineering. In particular, your view of how hard it is to write large pieces of software seems to be way, way off.

[needusername]

Every other JavaEE application server is better than WebSphere.

[aragot]

Please provide your own estimates on how many people are required to make great software. Not talking about Word in particular.

I've worked in large companies, now I make a living off my own company, so I must know something about software teams, complexity and feature creep. I didn't provide these estimates in the void.

[jacquesm]

> I didn't provide these estimates in the void.

Actually, that's exactly what you did.

You made a specific claim about a specific piece of software and you failed to back that up with so much as a shred of evidence.

[aragot]

Let's take public figures I can talk about: According to Wikipedia [1], Atlassian has 1148 employees and 20 products. It's 57 employees by product. BUT Atlassian also grew by 44% in 2014, [1] and recruits intensively, so we can estimate that they had 797 employees in 2013. If correct, it means the products they have now have been built by 40ppl on average, including support people, accountants and marketing. Sorry I know some products are bigger than others and I can't talk about those specifics.

And Atlassian does make world-class products which certainly compete with IBM. Please downvote if you prefer IBM ClearQuest to Atlassian JIRA.

I apologize for making an unbounded claim. It was unprecise because I didn't want to be specific about previous workplaces I've worked at. I have felt a shred of hatred from the HN community in the present situation, with the downvotes and negative comments.

Internet Explorer 3 had 100 people and IE5 a thousand [4], which is superior to the figure I was once told, which was 25 people for the development team. I had never been surprised by that low number because complexity grows exponentially and brings problems. I admit downvoters were right that Word didn't take 25 people to build, because of course everyone knows Microsoft has 128k employees [6].

GitHub is 255 employees[3] - Half of which for Enterprise [unbounded claim] and they weren't so many when they got famous. Git itself has had 100 developers over its history [5]; If they were employees, they wouldn't have been simultaneous employees.

So let's change my proposal: Who thinks world-class products are built by rather medium teams (<100 employees) in a few years (about 3 years)? Who has estimates for well-known products?

[1] http://en.wikipedia.org/wiki/Atlassian

[2] https://www.atlassian.com/company/press/press-releases/atlas...

[3] https://github.com/about/team

[4] http://www.citeworld.com/article/2147006/consumerization/int...

[5] https://github.com/git/git/graphs/contributors

[6] http://en.wikipedia.org/wiki/Microsoft

[ibmthrowaway271]

But this kind of thing does exist throughout IBM.

I work on a product that's pulled in >$2bn in revenue over the last 20 years. All with an average of 25 developers working on it. It's still going strong.