And no no no, there is no way anybody could POSSIBLY care enough about "old IBM languages" for which the 5100 is a presumed portable bridge to send somebody back with a very expensive toy at high risk to retrieve one. I mean, what critical program or information could POSSIBLY exist at that time? Programs -- give me something to program -- anything to program -- and I'll cut programs that supercede the entire code base accessible to pre-1975 computers in six months. Programming is goddamn easy, and the world abounds in good programmers and truly superior programming environments NOW. By 2036, if they've done nothing else they will have created five or sixfold improved programming environments, possibly integrated with at least rudimentary AI.
Lessee, following Moore's Law and being VERY CONSERVATIVE with full TWO YEAR doubling time, we expect 16 doublings in constant-cost power between now and then. That is, an el cheapo personal computer of 2036 ought to be roughly 64000 times more powerful than an el cheapo PC today, just as today's are amazingly some twenty or thirty thousand times more powerful than that good old IBM 5100. With a laptop running at the power equivalency of a few hundred TERAHertz, with local disk storage capable of holding a few petabytes of data (that would be 10^15 bytes), with network capacity to the household reasonably expected to be easily into the gigabyte per second range if not the terabyte per second range, the guy describes "the web" of that time as if it is more or less the same as it is now. No automated realtime videoconferencing. No voice activated, high bandwidth wireless connected PDA's with far more compute, storage, and network capacity than a system currently sitting directly on the T3 cloud. No mention of megalibraries, collation of world-data, AI at all. Bo-ring. Wrong. This guy may be from the future, but not MY future. In my future, war or not, nobody will give a rodent's furry behind about antique IBM code "readable only by an IBM 5100".
What, people suddenly became very, very stupid and couldn't write a decompiler or disassembler for a system that had a literal handful of machine instructions (mind you, I >>learned<< those instructions taking a microarchitecture course back in, uh, 1975, and they were extraordinarily clearly documented then as IBM documented the shyte out of everything! Those documents STILL exist, I'm absolutely certain, and have very likely made their way into electronic archives that would survive a nuclear war. I'd find it more believable if he said that he was returning to pick up one of those manuals -- it would be faster and easier to write an assembler-simulator (one of my HOMEWORK PROBLEMS back there in 1975) in e.g. PERL straight from the manual and run it on a virtual machine (especially a 127 THz virtual machine with an AI unit helping you program than it would be to get an IBM 5100 to do anything at all.