Killer App

Killer App is short for KillerApplication - both deserve a refactorization.

KillerApps are those applications which suddenly change the paradigm of computing as we know it.

KillerApps in temporal order (FastestFingers may rearrange)

Compilers: we could use our language to talk to computers.
[Email actually became available over ARPAnet about this time, I believe. - SH]
VisiCalc invented the spreadsheet: Launched the personal computer as something of value to non-nerds.
WordProcessors: we had electronic editors who could check spelling and grammar.
ContactManagers?: keeping track of contacts, phone numbers, appointments, sales...
Email: quickly interconnected many networks, esp: Bitnet, Internet, Decnets, Usenet, etc.
Gopher: The big TCP/IP application that described the Internet as a huge treelike structure of pages.
WWW: Better than Gopher, it made the internet look like a bunch of interconnected documents.
VNC: VirtualNetworkComputing: makes the distinction of "What OS do you use?" vs "What OS is on your desk?" a real question, in fact the answer to the second question is "It no longer matters!"

           * MIT Athena did this with X much earlier

Lossy Compressed Media: Napster, iPod, WMP, iTunes, mp3, DivX, etc, DVD, etc. Every one of those were a little seizure in the computer industry, all stemming from the moment people started ripping MP3s since they realized that "perfect to my ears" and "perfect copy" were two very different things.

I would take issue with the description of Gopher and the Web as "making the Internet look like" something, because both these services actually make a subset of the Internet's content available. Unless you believe that TheMediumIsTheMessage? as far as the Internet is concerned, there are many services, some not content-based (e.g. Network Time Protocol) that neither the web nor Gopher really integrate.

I'm not taking issue with gopher clients and browsers as killer apps, just trying to correct what I feel is a perception skew. The Web != The Internet.

Gopher doesn't even register to me. It was really only ever a rather limited prototype for the web. And VNC can't be a killer app because hardly anyone's ever heard of it. -- CharlesMiller Agreed. A KillerApp is something that is useful to common people, not just to the techno-elite.

I don't like the above at all. It just seems wrong to me in so many ways. Here's an attempted rewrite.

KillerApp: an application the use of which is so compelling that it is in itself sufficient reason to use a particular computing platform.

It is "accepted wisdom" that no platform will achieve widespread success without such an application. A platform can not in itself be a Killer App, it first needs something to run on it that makes people want to change the habits of a lifetime. For all its good points, BeOS never had a killer app, and DaveWiner suggested in a recent DaveNet? article that JeanLouisGassee may even have tried to avoid finding one because he was afraid of becoming subject to it, as Apple was beholden to Adobe in the early days of the Mac.

The term is most commonly used in the sphere of personal computing. Here are the most famously recognised killer apps in personal computing, by platform:

Apple II: VisiCalc
IBM Compatible PC: Lotus 1-2-3
Apple Macintosh: Adobe PostScript & PageMaker

There is some controversy over what the KillerApp of the Internet is. If we had to point at a single program, then it would be NetscapeNavigator, but I think that's more a case of good timing. The net was already growing exponentially before Navigator was released, it just hit the right spot on the curve.

Email is, and always has been a far more compelling reason for the adoption of the Internet than the web, and email has never been dominated by a single application - even in the height of Netscape's browser market share, they had to share the Windows email client space with applications such as Eudora.

  (or, conversely...)

KillerApp: an application the use of which is so compelling that its absence is in itself sufficient reason to abandon the use of an otherwise hardware-capable computing platform.

In the long term, it's usually easier to port any particular 'killer' application (or provide a suitable clone) to a platform that is missing it than it is to make a new platform and get people to use it by providing a killer application. However, when there are many killer applications in a particular field, the barrier for introducing a new computing platform to that field is very high, as it requires porting and cloning many existing killer apps. Meanwhile, if the new computing platform attempts to compete by providing its own new KillerApp, the cost is much lower to port or clone it away from the newly competing platform.

While the earliest KillerApps did an outstanding job of promoting new OperatingSystems, today's abundance of KillerApps have a mostly negative influence upon the development of new systems in many fields. The barriers for entry are rising.

Don't forget the role of video games as killer applications. They have served the role on both consoles (which are a sort of computing platform) and personal computers, and have driven the need for larger memories, faster processing units, graphics processing units, and aided in driving the CD drive for data. Individual games that might be considered Killer Apps (or KillerGames?) include Myst, various Zeldas, Final Fantasies, Tetris, Frogger, Blocks, Sim Cities / Sims/ Railroad Tycoon, Civilization, Halo, Freecell, Spider and Second Life. (I can't count on two hands in unary the number of people who have told me they'll buy <system X> just for <game Y> and hope that a few other games come along, too.) Beyond games, other sorts of entertainment apps provide a great deal of intrinsic value -- such things as Windows Media Player and Quicktime, legitimate DVD playback, and other multimedia have driven the adoption of more powerful computers and later OperatingSystems, and also hindered the adoption of those lacking it. (Microsoft, among others, sells versions of its OS based on such things.) Games in general hinder the adoption of Linux among the younger crowds.

I don't think it makes sense to try to single out a particular e-mail client or web browser as "killer apps". E-mail and WWW browsing are KillerApps? of the Internet - without them, the general public would still not know nor care about the Internet, but with them, more and more people are finding Internet access to be indispensable.

"Real people" use the Internet primarily for:

E-mail. The identity of the client software isn't important - the fact that anyone on the Internet can send e-mail to anyone else is what makes it so compelling.
Online shopping.
News and information. (Although most people do tend to go back to TV/newspapers when big stories come up.)
Entertainment. (Downloadable porn is, for better or worse, a KillerApp of the Internet.)
Chat, bulletin boards, and other "community" forums.
Software downloads and upgrades.

The race at the moment (August 2001) is to become the KillerApp for what is hyped to be the next platform, WebServices.

Another potential next-killer-app would be videophone/videoconferencing. Right now, we aren't quite there because the infrastructure doesn't support the bandwidth and there is no standard protocol embraced by all. But when the Internet becomes a reasonable replacement for the telephone, with video and real-time document-sharing, everyone will use it.

Q: Will WikiWikiWeb ever be a KillerApp? -- MikeSmith

A: Perhaps, many of us are already captivated. Imagine the future: since wiki is almost entirely text based, software, i.e. IntelligentAgents will be able to understand it - we are all just processing the SameStreamOfBytes. As soon as software begins to grok anything, if it comes across the PatternPattern page, we had better just step aside and watch the oncoming ParadigmShift.

It's useful to note that the Wikipedia is already a staple site for people to visit when a question needs answering (about anything from precipitation in Antarctica to the premise of the last episode of House).

I don't believe IntelligentAgents will make a significant contribution until text is quite a bit more structured than a 'stream of bytes'