Future of Web Apps – Delicious Things we’ve learnt

I’m sitting here with about 800 other people at the Carson Workshops summit on the Future of Web Apps. This will be a bad attempt to take notes – unedited or spellchcked

Joshua Schachter = things we’ve learnt.

Browsers – browser compatibility issues drives you nuts.
Header issues cause pain.

Scaling: don’t do it! Whatever you predict is not going to be the real problem. Design databases – really understand the ins and outs..set up a monitoring service. Use caching as much as possible – minimise hitting the database.

Abuse: wait to see what breaks before you fix it. Use a proxy – manage the resources. Figure out some kind of throttling

APIs..build. Make them easy to get into and out of – more people will use them

Identifiers: don’t expose internal identifier, expecially if sequential or conputatable – people will scrape the database. Lots of hits – have to build in defence mechansims.

Features: what you leave out is as important as waht you leave in. Don;t try and be all things within a app. Try and build features that people use, rather than what they ask for. Understand the why…what’s the real problem.

RSS: put them everywhere they can be – its a default way of getting info. Really need to understand the headers/caches etc to reduce hits to the database. RSS traffic is the largest on delicious

UR:S: make sure they follow the use of the site. They’ll be spread around, make them easily sharable.

Surprises: look for interesting behaviour – users can surprise you, do you amplify, ignore or damp user behaviour

Passion: solve a problem that you really have. Delicious was built to solve the problem of finding things in his link file

Release: get things out there – everyday you don;t have something in the world, yoou’re not picking up users and followers, you;re not picking the users brains. Get it out soon

Attention: give it, populise things. Works out well if population is small and biased in the saem way. Increase the population the bias drifts – and the overlal ‘popular’ is less popular to more people.

Spam: the opposite of attention. Undersntad what you are buildingo reduce attracting the spammers. Reduce the feedback to spammers – keeop them guessing about what is happening.

Tags: useful for recall, OK for discovery, awful at distribution. Not all metadata is tags. Don’t make it difficult. Beware librarians – ‘official’ tags – set heirachies. Impossible in a social network.

Motivation: why should people go there – what does it do for me. Expect the user to be selfish and build it for them. There’s no user 2 if user 1 does not have their needs met. (this has to be key for all technologies – why does it help me). If people are motivated to use it, that woill build evagelism.

Effort: make sure it’s in the right place. Targetthe effort.

Measurement: watch your system carefully. Intuition is guesswork backed by numbers. Look at everything. Measure behaviour rather than claims.

Testing – user testing is ‘pretty important’ make sure what happens with a user is what you expect. Ensure all the eam can watch the testing – get everyone to believe what is being said. Let people ‘play’ don’t set it up too formally.
Language: you have to speak the users languages. Bookmarks – if you used NEtscape or FF. IE it’s favourites. See what your audience use

Registration: don;t make users register before they can see your site. give them as much access as possible to see what the site does. registration is a big barrier..entice first. Show rather than tell.

Design Grammar: work out the best UI – understand how people use sites now and keep something similar. Don;t surprise people and do something diffiernet – understnad what wnet before

Morals – develop a sense of morals. It’s the users data…don’t abuse it. Dat is really purged from system and not kept in a log.

Infection: enable evangelism. Invade every communication stream you can find. Look for viral vectors.

Communities: understand community dynamics – but you can’t necessarily set up to drive a community. The community can be emergent. Many community dynamics suck…

FONT: the IRC channel seemed to be enthalled by the font in this presentation..this feels like a running theme today