I don't race any more so don't do intervals or specialist training, the bulk of my riding is 'steady pace'. I ride rollers on a 61" fixed so do about 18-20mph. On the road I do between 15 -18 mph, occasionaly pushing myself to evens (for upto an hour) if I feel particularly good. So relatively speaking there isn't much difference timewise between what I do on the rollers and what I do on the road. By the end of the year I've probably only done about 5% on the rollers so I'm not really bothered, it gives me an estimate of the amount of time/effort I've spent on a bike in a year.
By recording all time spent on the bike (MTB/commute/rollers/road) as 'miles' I can look back over a year and say 'I spent more time on the bike this year compared to last' or 'I had a good spring this year because I put a lot of work in in the winter'
Why should x number of miles in my log (however ridden) at the end of the year really matter to anyone except me?
45 minutes to an hour on the rollers can be more mentally and physically demanding than going 20 miles with a hebridean tail wind - it's all relative (and personal)
Because I use Bikejournal I do have the option to filter out road/mountain bike/commute/rollers if I wish but can't be bothered.
And I have no problem with your 17 minute turbo 10. There are many different types of records - for rollers, track, HPV, out and back road TT, place to place and oneway records.