After some discussion in the discord, the topic of variance was brought up last week. Codex is a very low-variance card game compared to many of the other games on the market like Magic or Hearthstone. Magic primarily achieves variance by having a resource system with uncertain resource growth (as well as rather large starting decks). Magic’s resource system is a flawed system because there will be non-games when the correct resources aren’t drawn in a timely fashion or too many resources are drawn. Hearthstone has a stable resource growth system - players linearly increase by 1 every turn, regardless of their other actions. To increase variance in Hearthstone, they added a variety of cards with uncertain outcomes - spells deal damage to random things, creatures come into play and have random bonuses, etc. This was especially important because Hearthstone has relatively low deck sizes (30 cards). Another way variance is increased is by having 9 factions that can’t be mixed (Magic only has 5).
Codex has a superb resource growth system where players pay two costs (1g and 1 card) every time they want to grow their resources, further encouraged by needing to meet resource thresholds to play stronger cards. Back to variance, Codex’s resource system ensures a near-perfect mix of stability and uncertainty by putting the choice in the hands of the player. Codex has variance through its deck construction via teching cards. However, there is a relatively low amount of variance - a player will almost always draw a card that they tech by turn 3 or 4, which adds a huge amount of consistency to the game. Perhaps too much consistency?
So from the vantage point of wanting to increase variance, what can be done to change Codex while keeping all the other excellent card designs, various polished systems (Hero, Tech, and Card Discard/Draw)? I have some ideas and they involve the starter decks. The goal being to make the first two turns higher variance, which would cause a cascading effect of adding variance to the rest of the game. These ideas would probably need significant balancing in their wake.
I. Bigger Starting Decks
Imagine if each starting deck had 12 cards rather than 10. Before the game starts, each player would remove 2 cards at random from their starting deck and place those cards in their Codex. This would be hidden information and those cards could still be tech’d later. The inspiration for this idea comes from 7 Wonders: Duels which has players remove 3 random cards each age, adding variance to make games feel different.
Pros: Increases variance! Even playing several games from the same positions (p1 and p2), this would make the early game noticeably different because some games a player would have far different options available. New cards! It would be really fun to have new cards in Codex 
Cons: Starter decks are designed to ensure your hand will always have a playable unit. This proposal would probably only work if the new cards for each deck were units. Codex has achieved a really nice balanced early game and new cards would undoubtedly upset that balance to some extent.
II. Smaller Starter Decks
Imagine if each starting deck was actually just a starting hand! This proposal can be tested in conjunction with the previous one (strangely) if a player put 5 cards from their starter deck at random into their Codex. The upside is that you would have each player tech 5 cards on turn 1 (rather than 0). Players could tech any of their starter cards not in their initial hands as normal. Then on turn 2 they begin teching 2 cards/turn as normal. The upside is that players could tech in non-starter cards like spells or tech I cards (big buff for p1).
Pros: This would add a significant amount of variance and make turn 2 of Codex much less predictable! This variant is very easy to test - no new cards required. The ability to build so much of your deck would undoubtedly lead to new strategies that could be explored.
Cons: This would undoubtedly force a rebalancing of various tech I units that may be oppressive. It also places a big decision burden on each player early in the game. It would still work with the natural rhythm of Codex - P2 would be looking at their opening hand and deciding what to tech while P1 would be looking at their opener trying to decide what to play and tech. Finally, I think this could be a big turn-off for new players. While this would increase variance in some ways, if there were a degenerate turn 2 play, it would actually decrease variance because players would always tech the same way and could basically build their turn 2 hand however they like.