Summary of Steve Jobs' talk in 1992 at MIT
This is a summary of
• There are two kinds of apps: standard apps like Excel that everyone uses, and custom software developed for a particular company's needs. Some of this custom software is mission-critical, and therefore no less important than standard apps. Unsexy but important.
• Object-oriented programming will the future. It leads to 5x faster application development: you can build your app in a quarter instead of one year + one quarter. This makes all the difference. On Windows, you still build apps without OO. This path has no future.
• The number of lines of code an engineer can write per day is same across languages. Higher level languages win by letting each line do more work, not by letting you write each line faster.
• The operating environment (like OPENSTEP) matters a lot, not just the operating system (like NeXTSTEP). The operating environment consists of the development environment (like IDE) and runtime environment (like the runtime for your language). That's why NeXT is porting their OE to Windows NT.
• A tech company can either have few customers each paying a lot, or lot of customers each paying a less.
• When each customer pays little, like $10001, you can't employ a highly skilled salesforce. You'll end up only with unskilled salespeople who say, "You want a 17-inch monitor? Here it is." The industry will focus on racing to the bottom and become very efficient at fulfilling demand for the lowest cost, but lose the ability to innovate and create demand for the innovative product. When an innovative product comes along, they won't be able to sell it. They'll say, "Why don't you buy this Windows PC?" They don't understand what's different about NeXT. Innovators like NeXT need to build their own sales and distribution channels for that reason.
• Hiring a great person can take years. They may turn you down, and over the next few years, you’ll talk to others to see if they’re a fit. And they won’t measure up. So you have to keep courting the star for years. When they join, it will all be worth it.
• Sometimes technology progress makes something feasible that wasn't until then. It takes 5 years for people to notice that the window has opened, and another 5 years after that for a commercial product. 5 years after that, the commercial product has become mainstream and a new frontier has become technically feasible, repeating the cycle.
• If you try to build before the technological window fully opens, you’re trying to force the window open.
• Tech innovation like NeXT or the original Mac costs $200m.
• Marketing a product is very costly. If you build a revolutionary new spreadsheet, for example, it will take you $100m in marketing to rise above the noise level on mainstream platforms.
• The first product you build to do that is often the wrong one. This is a mistake that both Apple and NeXT made (with the Lisa and the Cube respectively).
• It's a good strategy to sell to academia, heavily discounted. You'll get good feedback, and when the students start working, they'll recommend your software to their employers.
• As of today, laptops are underpowered and meant to be used only when you're outside. They're not a substitute for the desktop for use as your primary computer. But that is going to change: desktop technology is going to be made lighter and smaller and find its way into laptops, resulting in desktop replacement laptops. On the other hand, power users are demanding even more than what desktops offer, let alone laptops: photographic quality displays, which don't exist. Big, high-resolution displays. 100 mbps wired networking, not the 19 kbps wireless that exists in laptops. They want high performance, which will drain a laptop battery in 3 minutes. They want a lot of storage. So the market is going to be bifurcated into: a) laptops becoming more and more powerful and working as desktop replacements for people who aren't too demanding b) desktops become even more powerful for power users.
• When new technology comes along, like the Mac when DOS was popular, the old tech can do from day one some of the stuff the new tech can do, but some of the other stuff requires years of catching up.
• Keeping inventory locks up working capital. Besides, if a supplier is sending defective parts, you want to catch the first one and stop the shipments, not accept thousands of items and then realise you have a warehouse full of defective stuff.
• Typically there's an incoming warehouse (for parts) and an outgoing warehouse (for completed products). JIT manufacturing eliminates both in favor of daily or more frequent deliveries.
• When researchers build successful technology, they're sometimes tempted to commercialise it within an academic environment. But some company notices it and clones them, doing it better than the academics. This happened 100 times.
• The future will be object marketplaces where you can buy objects or components from vendors and embed them into your OO app. Some companies will sell only objects, not apps.
I've doubled the costs Steve mentioned to account for inflation.