Anserimimus' holotype is a skeleton missing only some cervicals and the skull, but only the scapulocoracoid, manus and metatarsus were described and illustrated by Barsbold (1988). Kobayashi and Barsbold (2006) helped a bit, as does Bronowicz's (2011) fragmentary but well described specimens. But what keeps this from scoring higher is the availability of good photos of the mount (e.g. this) and Kobayashi's (2004) coding it for several matrices.
9. Tie between Cristatusaurus, Deltadromeus and Afrovenator
These taxa all have a common theme- Sereno described them (well, he described Suchomimus, which I sink into Cristatusaurus). Published by the king of the Science tabloid, featuring a Paulian skeletal and several zoomed in line drawings, these taxa have yet to be redescribed in detail. While there are photos of mounted skeletons available, these are all casts. The best source of information on these are the matrices of Rauhut (2003) and Benson (2010).
Basically complete, but only the proximal tail, arm and lower hindlimbs have been illustrated and briefly described. Recently, the braincase was also described, though I lack that paper so far. Ceratosaur matrices like Carrano and Sampson (2007) have some info.
Known from quite fragmentary remains, but very poorly described and only illustrated by two partial vertebrae (apparently my tracing survives unattributed online). Is it even dinosaurian?
6. "Chilantaisaurus" zheziangensis
Dong (1979) gave this proximal tibia and partial pes a fairly useless description and illustrated two pedal digits in ventrolateral view and curving towards the viewer. :| Therizinosaurian affinities have been suggested, but it's not been examined since.
Only a few vertebrae, pectoral girdle and proximal/distal metatarsal outlines have been illustrated, and the description has yet to be translated from Chinese. An online photo shows a lot more is known, but also that size differences mean more than one individual/taxon are present.
Ah, Conchoraptor. You might be asking how can this be obscure when there are so many skeletons casted. I've even seen two in person. The problem with Conchoraptor is that it's unsure just which specimens belong to it, besides the holotype (of which we have illustrations of the skull, and a manus and metatarsus that presumably belong to it or a paratype). None of the articulated skeletons photographed online have the right manual proportions or slender metatarsal II, and crestless skulls like ZPAL MgD-I/95 (the one described by Osmolska, 1976 and which Kundrat has recently been describing braincase details of) and the one on Witmer's lab page identified as "Ingenia" could belong to other oviraptorids. No one has ever provided a modern diagnosis, described the holotype in any detail or justified the referral of other specimens. This makes depending on matrices more risky than for other taxa listed here, since their OTU could be chimaerical for all I know.
This has been illustrated in the literature even more seldomly than Conchoraptor, since at least ZPAL MgD-I/95 has papers dedicated to it. We only have the skull, mandible, overly schematic ilium and a single caudal vertebra illustrated. Lucky for us, Auditore was able to obtain photos of the holotype and illustrate it as detailed by Cau on his blog. As for Conchoraptor, Norell et al. (2001), Lu (2004) and Maryanska et al. (2002) all provide codings, which can be trusted more for Rinchenia since there's only the holotype (though I note a recent incorrect trend of referring IGM 100/42 to it).
The muse for this post only gets second place. For Adasaurus we must depend on a few schematic drawings (pelvis, metatarsus, pedal digit II), since Barsbold (1988) neither described it besides noting a couple features, nor does he allow photos to be circulated. There was a photo of the holotype online which has since disappeared, but shows Barsbold's pelvis illustration is inaccurate, and to make it worse Kubota (pers. comm. to Senter, 2010) indicates the supposed small sickle claw doesn't belong. I'm just hoping Kubota plans to redescribe the taxon like Kobayashi did for Barsbold's basal ornithomimosaurs. Until then, we depend on codings from Norell et al. (2001), Senter (2007) and Longrich and Currie (2009), and a few scattered notes. Hopefully Turner et al.'s upcoming dromaeosaurid monograph has some juicy photos.
1. Chilantaisaurus? sibiricus
There are a lot of fragmentary taxa known from a tooth or a vertebra that are poorly described (usually in a useless archaic way) and illustrated in a photo from a single view.