What prevents DIY speakers to become the ultimate?

DIY speakers can offer superior performance at a reasonable cost. Subwoofers are a great place to start because nearly all commercial designs use less than optimal enclosure volumes and/or poor quality amplification and crossover. Try the Adire Audio Shiva in their suggested 5 cubic foot vented EBS alignment and be astounded by the level of performance achieved for about $200 in total parts cost. Add an amp and the Outlaw ICBM electronic crossover and you have a giant-killer for peanuts. Sealed box subwoofers are even easier to build and usually yield excellent results. Madisound also offers an array of kits with or without cabinets that are competitive with the best designs available commercially at a fraction of the cost. My current stereo system uses the new Thor speaker (based on SEAS Excel components) and it is superb in every respect.I have used their LEAP crossover design service with fine results on a number of "from scratch" designs of my own. Search the web for "DIY audio" and all sorts of other useful sites can be found to help with design and construction. Give it a try!

A while back I heard some extremely good DIY speakers. The guy had worked on them for years. He made the comment that had he paid himself minimum wage, he could have bought a fine new car. I believe he had about $1000 in the guts with a few $100 more in the cabinets. He told me he'd tried to get the crossovers right for over a year before giving up and using someone else's design.

If and when DIY speakers become the "ultimate", they will in all likelihood become "commercial" speakers. The art and business of commercial design is to manufacture a speaker with the cheapest quality parts, make it sound very good (better than DIY for around the same cost) and sell a ton of them. A $300 retail bookshelf (Triangle, PSB, Sound Dynamics, etc.) probably costs the manufacturer around $70 or less to make. But it will sound and look better than anything that the average DIY can make for $300 in parts and their time (at minimum wage). Of course there are exceptions to this, perhaps a Northcreek kit made by a DIY who is also an experienced woodworker.

Most DIYs don't design their speaker projects, but rather use a proven design (Proac clone/ Lowther horn/pipe) or build a kit. This is the safest approach, but don't expect the "ultimate" sound.

I'm interested in making my own speakers as well. I'm not interested in engineering my own- nor am I interested in re-inventing the wheel. I understand that the Thor is a solid design, are there others? Is there a short list of Best DIY kits or designs.

Northcreek notwithstanding, a non-dedicated non-OEM can't buy tight-production tolerance drivers required for superb-imaging, ultranatural response. As alluded above, the best manufactures have custom-made drivers specced within a tight sensitivity window, and even then sometimes pair-match them. (Verity even jeeps a response curve on each midrange for replacement if needed!)
It's a given that Madisound, Solen, etc, get the outer 2/3 or so of a given driver run! I've built a couple of 2 and 3-ways, and once I heard a midrange voicing change of only 1/3dB over 1.5 octaves during a crossover massage, I gave up, knowing that I'd NEVER be able to afford drivers that well matched! This is by far the largest impediment to DIY
ULTRA-high quality production. Though a reseller can sit there and wade through a box of drivers, sorting their sensitivity with a Radio Shack SPL meter, and then selling so-called "matched pairs", that's only going a short way to providing yje standardization required, 'cause you gotta do it with the crossovers next!
Tweeters and caps are notoriously wayward in spectral response. When I heard Verity Audio's midrange, I cried, and now have a couple crates of nice big inductors, caps, resistors, etc., if anyone wants 'em for a dime on the dollar! Spendor, Revel, Snell, and even Boston (!) do a VERY creditable job of pair-matching and building to a target reference with decent precision. Who's going to build us DIY'er some clones to work with? I'll stick to power cords and other non-transducer components....
My hatsoff to those still trying!