Especially with the new smaller size restrictions, making weight is not nearly as challenging as it used to be. We aren't building giant double jointed arms, hanging mechanisms that extend 10 feet in the air, or double ramp robots anymore, and as a result it's a lot easier to make weight.
Here's what 2791's done to easily make weight every year:
First, build a simple robot. A simple robot has far fewer problems making weight than a complex robot. For the past several years, if we're planning a design that is going to have trouble making weight, it is probably too complex for our team to do effectively. Dozens of pure cycling robots in 2013 were under 100 pounds. Some of this year's low catapults were nearly as light.
Second, we try to build a robot so that the robot does multiple tasks with single mechanisms (i.e. combination arm and shooter in 2014, a shooter-hanger in 2013, bridge tipper and drop down intake in 2012). This isn't a universal rule as sometimes it is simpler and lighter to just use two separate mechanisms (2013), but it's a good way to keep part count and weight down.
Third, use close to the minimum amount of structure needed to do the job. I've seen tons of robots have trouble making weight with large square tubing frames with tons of redundancy. If you really think about it a lot of superstructure frames can be built with very few members. Our "superstructure" in 2013 was
a set of 4 1/4" plates. Even counting the shooter, it was just four pieces of 2x1 and some 1/16th sheet. No problems.
One of the coolest examples I've seen of minimal structure is
233's 2012 robot. They recognized that the ONLY superstructure needed other than the intake was to mount the shooter high and near the back of the robot, so that's all they built. The conveyor belt just ran in open space, not surrounded by frame the whole way down.
Finally, use the lightest parts you can get. Vex gears are a huge weight savings over AndyMark and can be used in all but the end of some of the highest load mechanisms. Build with more 1/16 wall, especially above the drivetrain. Don't go to steel or thick walls until you have to.
With all of these ideas combined together, we haven't had to really actively think about weight for the last half decade. If we started going down the path of a design that required complex structure or a high number of gearboxes / motors, we would probably start budgeting weight in CAD, but we try to avoid those designs if at all possible.