This is an epic tweet storm about Apple’s development process by Steven Sinofsky. If you don’t know him, he ran the Microsoft Office business unit for over a decade. This rant touches on how to balance quality, launch dates and features, IBM, iPhone dominance and more…
1/ Apple has a software problem. Here’s how it plans to fix it. https://t.co/dJaikfRhs7 via @markgurman // Let’s take a step back and talk about the broader context and product development at scale. Lots follows…
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
3/ Scanning the landscape, it is important to recognize that in total the work Apple has been doing across hardware, software, services, and even AI/ML — in total — is breathtaking and unprecedented in scope, scale, and quality. Not saying that lightly or trolling. It just is.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
5/ The pace of change has been remarkable. In the 10 years from when Apple acquired NeXT OS X was reinvented in a completely modern architecture. And in the next 10 years the iPhone went from that code to where we are today.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
7/ Microsoft Office released ever 18-30 months from about 1990-2010 but had a shaky start so you could say that from 1995 it kept that cadence but today puts out mostly modest visible changes to be SaaS like.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
9/ The only comparable project would be IBM System/360—the creation of IBM 360 hardware and software. The PC of course, but the scale (even at the time) was much less. Windows NT clearly had the scope/scale of software but had been done before (VMS) and built on PC h/w.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
11/ What is lost in all of this recent discussion is the nuance between features, schedule, and quality. It is like having a discussion with a financial advisor over income, risk, and growth. You don’t just show up and say you want all three and get a “sure”.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
13/ In practice when building Office (and later Windows) whenever someone on the team would panic and ask “are we date driven, feature driven, or quality driven” we would just roll our eyes and pull up a chair…This was so common we just called it conversation #37 and move on.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
15/ Customers don’t care about any of that and that’s ok. They just look for what they care about. Each evaluates through their own lens. Apple’s brilliance is in focusing mostly on two audiences—end-users and developers—tending to de-emphasize the whole “techie” crowd, even IT.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
17/ This approach is rather unique compared to other tech companies that tend to develop new things almost independent of everything else. So new things show up and look bolted on the side of what already exists. (Sure Apple can do that to, but not usually).
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
19/ There’s nothing magic about this. It goes back to a balancing act. Mature orgs just manage this the whole time. There are processes and approaches that you use so you never face the absurd notion that this is a zero sum trade off between quality, schedule, features.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
21/ What I think it happening at Apple now is not more dramatic than that. What they had been doing got to a point where it needs an adjustment. Reality is that for many at Apple it feels dramatic b/c it might be first time they have gone through a substantial “systems” change.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
23/ In my view the ‘moment’ is being manufactured a bit right now because of the perception that the Apple products have become less stable or…”buggy”. This is where the “signals” about the state of the world can get confusing.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
25/ How does that explain general “buggy” feeling w/ so many super smart/skilled people saying products are suffering? It’s because of the depth and scale of usage that comes w/success. A responsibility.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
27/ I can’t prove this but I’ve also worked on some really big projects where people said the same thing and we had tons of data. Apple has the same data. What is different is that at scale a bug that happens to 0.01% of people is a lot of people. A stadium full or more.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
29/ The more a product is used the more hyper-sensitive people get to how it works. The human brain is extraordinary in how it recognizes even the slightest changes in responsiveness, performance, and sequencing of operations.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
31/ But what happens to a team as complexity evolves is simply the challenge of coordination and more importantly consistency or leveling of decisions across a complex system. This is particularly acute if the bulk of the team has only known the previous few years of success.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
33/ They have more data and understanding to make adjustments than anyone. The only thing I think is fair to say from the outside is that this is not nearly as dramatic as it is getting made out to be…
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
35/ The idea though that this is some massive shift to focus on one dimension of the overall product process: quality OR features OR date is just **nonsense**. Nothing of scale is thought of or executed that way.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
37/ Big projects run poorly are “date driven” or “we’re getting this whole thing done (famously “second system syndrome”). Lame projects are “we’re fixing bugs” (used to call this “re-indenting all the source code”.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
39/ Ultimately at MS used to have Conversation 37:
• Eng wants to do nothing but fix code // BUG BUG
• Sales wants new product every year w/new quotas
• Press would like a new thing each month
• Techies —revisit core UI, add options on demand
• IT—no change, ever 🙂— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
41/ Some people say “oh consumer products need yearly releases” or “enterprise need constant value for SaaS”. The only thing you need is to do good products when you have them. In the scheme of things market timing except for seasonal-only products isn’t how to scale for 1B.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
43/ Growth hacking or “move fast break things” sounded great until it wasn’t. This especially doesn’t/never worked in enterprise. Again, adopting a methodology absent building a great product *always* fails. “Internet time” was kind of a bust the first time around.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
END/ So to me on Apple, even as an outsider, I feel confident saying that this isn’t reactionary/crisis or a response to externalities. Importantly it isn’t a massive pivot/“student body left”. It’s a methodical and predictable evolution of an extremely robust and proven system.
— ??♂️Steven Sinofsky ॐ (@stevesi) February 12, 2018
