They're definitely onto something here with regard to skills vs. ability scores.
Let me just give you a little background on my own campaigns to clarify what I mean. I play D&D Basic/Expert usually, where most actions are resolved by one of three methods:
- A fixed chance-in-six to open a door, spot a secret door, avoid a trap, etc.
- The thief's d% chance to do something thiefy
- The ability check, rolled under the score on 1d20
For my own setting, I synthesized all of these different methods into an ad hoc skill system where there are twelve skills, and everything is handled by rolling 1d6, either rolling under your rank in a skill (which runs from 1 to 5) or rolling under 3 + your ability modifier, just like the Strength based "open doors" roll from D&D Basic.
Now, I don't have any kind of class-based restrictions on who can take which skills. Anyone can take thieving skills; thieves just enjoy more skill points to spend. So the skills in my campaign wind up serving as both the primary means of customizing characters with similar classes; and the primary method for miscellaneous task-resolution. I found that once I implemented a skill system, I very rarely ever had to resort to a bare ability check. The ability score modifiers still mattered for combat and such, but the scores themselves hardly ever impacted anything else.
So for the last campaign I started, I tried something different. Inspired by this speculation about 5th edition, I kept the skills and the ranks, but instead of having my players make their skill checks on 1d6, I would assign a difficulty level (1 to 5) to a given task, and if the character had a skill rank sufficient to match or beat the rank, I ruled it an auto-success. If not, they rolled a relevant ability check on 1d20 to see if they at least attained some measure of success at the task. And I have to tell you... it's really improved the flow of gameplay at my table. This is a dungeon-crawl campaign, so there's lots of searching for traps and secret doors, lots of picking locks and disabling traps, lots of trying to identify mysterious items and read ancient languages and communicate with semi-intelligent monsters... and having some characters just be skilled enough to succeed at the more mundane tasks really speeds things along. Meanwhile, when a more difficult task arrives, the ability check on 1d20 is pretty solid means of success or failure (and if a task seems extra-difficult, one can always add one or two extra d20 rolls and force the player to compare the highest roll to their relevant stat).
It just feels even more like D&D to play this way, I'm playing freaking B/X already! So... yeah, hearing this about skills in 5e makes me very hopeful. Very hopeful indeed.