On fees: fees are generally charged, but they are tiny. However, all those involved in Bitcoin (including miners and software developers I spoke with) know that fees will rise and mechanisms are being created to make that simpler. The production of Bitcoins will halve in 2016, and miners are, over time, expected to derive the rewards that drive investment and operation of the system's functions (operating nodes, mining, "burying" transactions in the block chain, all interrelated) from fees rather than coins.
If you read Andreessen's piece and my essay, you'll see that he properly discusses essentially counterfeit payment from one party to another, but doesn't address fraudulent payment and the infrastructure to ensure that the party paying owns the funds used to pay. That is, if Bitcoins are stolen and used to pay for goods, a merchant faces the same trouble as if cash were stolen and used to pay. Except cash can be untraceable, and Bitcoin transactions can be tracked, even if the party isn't directly known who engaged in the transaction. Law enforcement could prove funds are stolen even if they can't recover the goods or services purchased with the funds, and clawback the funds from the seller/merchant.
None of that is addressed in Andreessen's essay, in which he proposes that Bitcoin by having very low or no fees on Bitcoin-to-Bitcoin transactions removes the necessity for any per-transaction fees as are charged to deal with fraud and overhead in a credit-card system.
Most merchants are going to be more likely to deal with an intermediary Bitcoin operator who will handle transactions on their behalf and charge a fee for chargebacks and theft recovery.