Thanks for this. I don't think that 16.f3 is a terrible problem for Black though even if it is certainly difficult to prove it. I was amazed that the move 16...h5 appears to be playable though!

This is all interesting, although I think if White simply plays obvious natural moves he might get a small advantage. True, not very much, but that's chess. In practice, Black seems to have to play more cleverly/accurately than White. 
Here's some analysis (the attached is clearer):
[Event "?"]
[Site "?"]
[Date "2012.12.17"]
[Round "?"]
[White "Slav with 3 Nc3 Nf6 4 e3 Bf5"]
[Black "?"]
[Result "*"]
[ECO "D10"]

1. d4 d5 2. c4 c6 3. Nc3 Nf6 4. e3 Bf5 5. cxd5 cxd5 6. Qb3 Nc6 7. Qxb7 Bd7 8.
Qb3 Rb8 (8... e5 9. dxe5 Nxe5 10. Be2 Rb8 11. Qd1 $5 Bd6 {transposes. Black
can choose another 11th, but he's also allowed 11 Qc2, likely an improvement})
9. Qd1 e5 10. dxe5 Nxe5 11. Be2 Bd6 12. Nf3 ({Houdini likes} 12. f4 {, at
least initially; but it may not be a practical choice for White to accept
weaknesses so soon. Probably best to look at obvious developing moves first.}) 12... O-O (
12... Qc7 13. Nd4 O-O 14. O-O Nc4 {(Ntirlis)} 15. b3 Bxh2+ (15... Nb6 16. Ncb5
Bxh2+ 17. Kh1 Bxb5 18. Nxb5 Qe5 19. Ba3 $18) (15... Nxe3 16. Bxe3 Qxc3 17. Rc1
Qb2 18. Rc2 Qa3 19. Bc1 (19. Qc1 $14) 19... Qb4 20. Nc6 Bxc6 21. Rxc6 Rbc8 22.
Rxc8 ({maybe  best? maybe:} 22. Ra6 $14 {is better, with the bishop pair
conferring a small advantage}) 22... Rxc8 23. Bb2 Re8 $1 24. a3 Qe4 25. Bf3 Qe6
26. b4 Be5 27. Bd4 Bxd4 28. Qxd4 Qa6 29. h3 h6 30. Ra1 $14) (15... Na3 16. Bxa3
Bxa3 17. Ncb5 $16) 16. Kh1 Nd6 {(Ntirlos' recommendation)} 17. Bb2 ({he gives} 
17. Ba3 Qa5) 17... Be5 18. Rc1 Qb7 19. Nf3 Bxc3 20. Bxc3 Nfe4 21. Ba1 $14 {
(not much, but the bishop pair and pawn structure count for something)}) 13.
O-O Qe7 14. Nd4 Rfc8 $1 {(the Horvath-Brunello game)} 15. Ba6 $5 (15. Ncb5 {
was played}) 15... Rc5 $1 {Brunello's suggestion, and best.} 16. f3 $1 {
Covering e4 and g4, albeit weakening e3} Nc4 (16... Ra5 17. Be2 $5 (17. Qe2 {
is safer}) 17... Nc4 18. Bxc4 dxc4 19. Qe2 (19. Qc2 $5) 19... Rh5 $5 (19... Qe5
20. f4 Qc5 21. h3) 20. g3 Rh3 $1 (20... Be5 21. Qf2) (20... Bc5 21. Re1) 21.
Rf2 h5 (21... Rc8 22. e4) 22. Qf1 $14 {with the idea e4 (and threatening the
cheap Nc6) .}) (16... Rb6 17. Qe2 Bb8 18. Bb5 Bxb5 19. Ncxb5 {, for example,}
Qd7 (19... a6 20. Nc3) 20. f4 (20. Na3 {with the idea b3}) (20. Nc3) 20... Nc6
21. Nxc6 Rcxc6 22. Nd4 Rc8 23. b3 Re8 24. Ba3 $14 Ne4 25. Rac1 h6 26. Rc2 $14)
17. Bxc4 $5 ({Or} 17. Qe2 {with the idea} Nxb2 $1 18. Bxb2 Rxb2 19. Qxb2 Qxe3+
20. Qf2 Qxc3 21. Nb3 Qb4 22. Kh1 {. Black has a good deal of compensation,
although perhaps not completely full.}) 17... dxc4 (17... Rxc4 18. Qd3) 18. Qe2
(18. Qc2 $5) 18... Rh5 19. g3 Nd5 20. Ne4 $5 $14 *

I found it independently by runnin Houdini to D30 move by move in my repertoire and checking interesting options which pops-up.
Imo it puts Nc3-e3 order to rest unless Bg4 in Slow Slav is really worth preventing at the cost of losing all the options vs a6 and semi-slav.
Very nice article and analysis.

1.d4 d5 2.c4 c6 3.Nc3 Nf6 4.e3 Bf5!

Take a look at this one:
http://www.qualitychess.co.uk/blog/?p=1413

We have talked about this line quite a lot in chesspub but i remained silent for some time not revealing my analysis for reasons that will become obvious if you read the story. I hope that we can develop it further here. Let me note that still it remains unmentioned in recent works like Watson and Sakaev.