Hacker News Reader

Comment by projectramo | Hacker News Reader

Here is the issue: software is not a product, it is a service.

It has to be modified, supported, updated, debugged, improved, and so on.

I don't think a "flash" organization can produce a "serious" product. And by serious I don't mean one that is complicated or large -- it can do that -- by serious I mean one that is intended to be used for more than a few months.

View on HN

I know companies in the video game space that do this. They either are hired to write something or produce something in house. They staff up, write the title, ship it, do one or two patches, then essentially disband the team while they look for the next project.

Yeah, and it's total hell, I worked in that industry.

Imagine having to uproot your entire life and move at the end of the project because there aren't local jobs or that pay(which is 30-70% lower than standard SE roles) means you can't afford stay around in a high CoL. Many people I knew in the game industry lived month to month(present company included).

If there's any SE industry that could do from unionization it's gamedev.

I mean, you took the job. You know what you were getting. I have never done anything like that, I would turn it down.

Video games are show business, though, and follow the same gig-oriented pattern.

Unfortunately without the protections that most show business workers enjoy.

Sure, but it's still just software.

Way more manhours go into art, level creation etc. than writing software.

I wouldn't be surprised if a lot of maintenance (perhaps even the majority of maintenance) for games is done by a different team to the one that wrote it originally; especially any patch releases that come out many months after the original release.

The majority of this sort of thing is usually platform ports, these have a long history of being outsourced. Bug fixing will tend to involve at least a subset of the original team as shipping code in a game can be err... interesting by the end. Otherwise it's mostly bringing in specialists to write specific things that will then be maintained by the original team.

With the rise of games as a service it's also becoming more common to outsource the development of the game and develop the platform layer in house. Or visa versa depending on the expertise of the studio.

Platform ports was the main line of business for the company I was talking about. It was very much a feast or famine situation.

The domain knowledge involved with developing games is... "developing games."

The domain knowledge involved with developing an industry-specific tool or product is not always applicable outside that single context.

The recent shift towards DLC has changed this, though. The advent of day-one DLC and frequent releases of additional DLC have done a lot for job stability in the video game industry

Here's an infographic that explains it: https://static.giantbomb.com/uploads/original/0/5370/2147534...

Here's an interesting article about this: https://www.wsj.com/articles/why-a-75-billion-business-is-ge...

(I believe posted here on HN a little bit ago)

Sounds like the right use case.

The lifecycle of a game is similar to that of a movie, or book etc.

The lifecycle of most games, but far from all. Subscription based games (mainly WoW and other MMORPGs), F2P cosmetic games (e.g. MOBAs and Valve games) and DLC-heavy games (e.g. Paradox games) are all more similar to SaaS than movies or books.

I suspect that these games are developed by employees rather than temporary contractors.

And don't forget the way Lucas kept upgrading Star Wars years after release. Some things are never done.

It can produce a serious product.

Startup gets in touch with one-or-two programmers. They do a POC. Startup gets the funds. Funds must be utilized. Those one-two-programmers now create a FlashOrg1. In just two months its a huge 30 people team (including non-programmers). They deliver the product fast. FlashOrg1 is costly. Startup brings Org2 into picture. Org2 politicizes and hinders output from FlashOrg1. After a few months CFOs decide to cut ties with FlashOrg1. So this flash org was created and destroyed in a matter of 8 months. But the product continued

Have worked for one. It is fun to work with and stressful too. And such orgs are definitely not for employees - but for independent consultants/freelancers.

_Bad_ software is a "service". Good software works for 15 years before you need to do anything.

I used to think so, but not anymore.

Consider: you write perfect code, and you write the perfect web and mobile app.

1. Does it have a server?

2. What happens when everyone upgrades their browser?

3. What if everyone upgrades their iOS?

It is not your fault that it may not work in 2 years, let alone 15.

In my book Craigslist is good software. Looks the same as it did 10 years ago, does the job, works in the shittiest browsers imaginable.

Is there any chance in the world that Craigslist has not been continuously updated and maintained behind the scenes, regardless of the stability of the UI look and feel?

I agree with the points you made:

1. It is good software

2. It looks the same as it did 10 years ago

Although I don't know for sure, I suspect it has been updated, modified, debugged and so on quite actively over the last 10 years.

A pretty basic example, but one which illustrates your earlier point pretty well.

Even if we assume that it were a perfect piece of software 10 years ago, craigslist, at some point within the past few years, updated its CSS to include @media rules to display properly on mobile browsers. This update needed to be made in order to support an entire class of devices that didn't exist 10 years ago.

their business practices and model has changed some over the past 15+ years as well - some posts are charged for, more opportunities to report spam, etc. if it was 100% the same system it was 15 years ago, it wouldn't survive.

It takes a large technical team to pull this off:

https://www.craigslist.org/about/craigslist_is_hiring

Are you trying to suggest that Craigslist hasn't been updated in that 10 years? The UI might not have, but I guarantee you that various changes in business direction and in regulation have required Craigslist to make changes to the software running their servers.

Craigslist has surely had to keep up with browser updates, new browsers, new mobile devices, etc. There's no way the Craigslist codebase of today is the same as 10 years ago.

It depends. For software that interacts with services, most certainly.

Otherwise, consider emulators - games or applications written for platforms long gone can continue to run today, because that software was written (mostly) to some sort of standard.

My only experience with emulators is trying to run Commodore 64 games, and in my experience they end in heart break usually when you beat one level and are about to go to the next one.

Which games, and which emulator(s)? I've done a fair bit of C64 emulation; setup is sometimes a pain, but I've never had any technical problems once the game loads.

Sounds like C64 emulation isn't particularly robust or mature. For the opposite extreme of things, look at something like DOSBox or ZSNES.

But modification, support, debugging, improvements and so on take less than 1 full-time person.

It depends on the product.

I suppose we should now talk in ratios of FTE:PTE, and where we in the lifecycle and so on.

Note that bringing in part-time help, contractors and consultants is not new. The article seems to be making a point about the degree to which it is used.

Building a product from scratch is very different from maintaining a product.

It stands to reason that those that make the product don't need to be the same guys who maintain and extend it.

Do you really need a software architect after the major architecture has been built? Do you need a senior automation engineer when everything has been automated? Its really a waste of the company's money and the workers' time to have highly experienced and perhaps specialized doing maintenance work - it also leads to unnecessary projects in order to keep those people interested and engaged.

It's obviously a product, but it has to be modified and updated all the time because:

a) the industry on the whole doesn't care about quality, so we ship crap that needs to be fixed later.

b) selling a subscription seems to be more profitable than getting paid once, so companies tend to force customers into subscriptions.

Reply on news.ycombinator.com