Does a software implementation count as prior art and sufficient proof to invalidate a patent application?
Does it matter whether the implementation was closed-source, open-source or public-domain?
There's a case called Lockwood v. American Airlines.
The money paragraph follows. TL;DR? If the claims of the patent are taught by the public aspects of the system, then you're likely to invalidate. If the claims recite implementation detail that is not publicly accessible from the system, then you might still be able to invalidate, but this case is not on point.
If a device was "known or used by others" in this country before the date of invention or if it was "in public use" in this country more than one year before the date of application, it qualifies as prior art. See 35 U.S.C. § 102(a) and (b) (1994). Lockwood attempts to preclude summary judgment by pointing to record testimony that one skilled in the art would not be able to build and practice the claimed invention without access to the secret aspects of SABRE. However, it is the claims that define a patented invention. See Constant v. Advanced Micro-Devices, Inc., 848 F.2d 1560, 1571, 7 USPQ2d 1057, 1064 (Fed.Cir.1988). As we have concluded earlier in this opinion, American's public use of the high-level aspects of the SABRE system was enough to place the claimed features of the '359 patent in the public's possession. See In re Epstein, 32 F.3d 1559, 1567-68, 31 USPQ2d 1817, 1823 (Fed.Cir.1994) ("Beyond this `in public use or on sale' finding, there is no requirement for an enablement-type inquiry."). Lockwood cannot negate this by evidence showing that other, unclaimed aspects of the SABRE system were not publicly available. Moreover, the '359 patent itself does not disclose the level of detail that Lockwood would have us require of the prior art. For these reasons, Lockwood fails to show a genuine issue of material fact precluding summary judgment.