This page describes Devon's thoughts (slightly out of date) on this subject. Warren's version is also available.
The game is new, and many good ideas haven't been tried yet. So aiming to write the best side yet is not unrealistic. Here's some of Devon's advice on writing good sides.
Combat is dangerous, so running away is usually better than fighting. Most of the best sides run when shot at. You can simply run away from whatever scared you (the most popular approach), or you can run to a well-defended location (like Commune's Sustenance).
Obviously you will get attacked, and you can't always run. It's not unreasonable to spend as much on defense as on growth.
Unless you have a long-range robot sensor, you should watch for shots and chase them back to their source so you aren't helpless against long-ranged attacks.
Many sides will defend themselves in force if attacked. This means taking potshots at them is probably a bad idea. Even if you can attack and beat them, the gain from this is unlikely to exceed the cost.
Intoxicated got away with this when most other sides are poorly defended. This situation didn't last; better-defended sides score better, so more were written. If you want to be agressive, try sending a cheap probe before attacking, to see how strong the defense is.
Productive's most important feature is not its missiles but its movement. Clever movement, or movement of any kind, isn't something usually associated with giant autotrophs, but running to the nearest corner isn't hard, and protects Productive from all sorts of depredations. Unfortunately it tends to run blindly into things and get killed. Trying to avoid enemy cells (with a long-range robot sensor) may work better.
The object of the game is to accumulate biomass, and sides which are not good at this won't score much. Plus they risk being killed and eaten by more numerous enemies.
Autotrophy is easiest - no movement and no risk of wandering into an enemy - but gathering manna can have higher returns.
Obviously this conflicts with defense - if you spend half your budget on arms, it's going to cut into growth. A clever side could try to determine when defense is needed and when it isn't.
The larger the area you can gather manna from, the faster you can grow. Sides like Microb and Bacteria build gatherers all over the place, so they can gather manna anywhere that isn't defended. So you might want to kill or chase away anything that tries to eat in your territory. (But be careful not to provoke enemies.) Business Cycle collects food with forcefields, so it can theoretically eat even from defended areas.
If your cells engage an enemy one at a time, they will die one at a time. Sides that use their whole force in one place are much more likely to win battles. This is pretty much automatic for stationary sides, but once you start doing anything mobile it's easy to forget about coordination. One of the reasons Intoxicated wins battles is that it stays in a group.
Grouping once an enemy is seen doesn't seem to work as well as already being in a group.
Taking full damage from attacks is not the way to win battles. Dodging (like Intoxicated or Active) or using decoys (which nobody has gotten to work well yet) can provide a big advantage.
Pregnant cells move slower, make big targets, and take more damage from explosions. They should not be in combat (so fighters probably shouldn't gestate) and may want to avoid other activities that involve lots of movement. Some sides store up lots of energy before starting to use the constructor, so less time is spent pregnant. Constructors and energy storage are cheap, so it's not hard to delay gestation.
We intend most cells to cost about 1000. If this is not the case, it's a sign that the game is misbalanced, so we'll change the rules to encourage cells to be the right size. This means that if you build giant cells, however well they work at the moment, you can expect them to break later.
If you can't squeeze the hardware you want into one cell, use several instead.
It's easiest to write types that simply respond to stimuli. This is not necessarily the best way. Spending a few frames thinking about a decision (e.g. fight or flight) can help a lot. Remembering (or communicating) information about the state of the world (what direction has the fewest enemies? where is there lots of food?) or your plans (where will we attack next? which types are we building?) is also useful.
Don't try to make a side with every possible feature and no plan. Pick one answer to each important question and make it do that. If you can make a side that grows and survives well, you'll have made a good side.
Grobots by Devon Schudy (dschudy@yahoo.com) and Warren Schudy (wschudy@wpi.edu)