One problem is the complexity. To expand on what someone else said about US having 51 governments, the Federal government is also further fragmented. 2+ million people, and 5-10 million contractors (individuals who aren't full time employees but work alongside them). The true numbers aren't federated and there's not even a clear accounting of how many agencies there are - best estimates (including Federal sources) put it at somewhere around 130 - 270 - 440 agencies. Seriously, no one knows for sure, the org chart is not maintained or always that out of date.
When a new person comes in to head an agency and bring about positive technological change, someone who genuinely knows technology and its potential, they're usually unprepared for what they see. Their tenure is usually time-limited and they hope to bring about change fast. Certainly not stick around for 8+ years to see things through, because that just doesn't happen and it really does take that long.
For one simple thing they're trying to do, they find out that it's locked in by multiple years-long contracts that aren't synchronized with their timelines. For example, a contract for a CDN expires possibly years after they would realistically stick around so they can't change that. Or a contract for cloud services they don't want in the long run might need to be renewed soon so they might choose to renew it so they can focus on their projects instead of migrating and upgrading existing pieces for the next few years, because it really takes awhile. Or they find out that four separate departments purchased four separate transcription services at different times and for different lengths, so attempting to standardize it is wasteful on the account of having already paid for the service a few times over.
I'm describing an environment with lots of moving parts, many that have been locked-in for unworkable periods of time. An environment that has many individuals working at duplicative or cross-purposes, not even intentionally - just because they didn't coordinate. An environment that will take years for someone to map out before they can truly grasp what's happening or streamline it. The new person who comes in now has to deal with existing contracts, things that might be counterproductive to what they are trying to do and they can't be just canceled or legally or even just easily modified. At that point, ... sadly, many such people just go for cosmetic appearances of progress. Even well meaning individuals - and entire groups - end up settling for a shortcut and then bolting.
That's what happens on an agency level. But what about centralizing and standardizing technology?
Turns out, that's been attempted several times and forgotten. Before GSA's 18F group came to be in the aftermath of the healthcare site debacle, there was another group that was put together and elevated to do just that- bring innovation and centralization to GSA and to the federal government. So, when 18F got stood up as a group to revamp tech federally, a group meant to do innovation centrally was fundamentally duplicated in that very step. There were role and identity collisions and they had to spend some time integrating while growing and anyone dealing with 18F in the early years first looked up who they were really dealing with - as in, the pedigree of old or new blood.
Whether any strategic lessons were learned here is anyone's guess, but, people who know better have argued that this innovation function shouldn't be centralized but evenly deployed in agencies. As in certainly bring in more tech savvy people to federal service, but spread them out permanently and empower them.
GSA's 18F tried to do that in a limited fashion. 18F is a group of well meaning individuals who've done a lot and certainly demonstrated their expertise. But, to be blunt, some of their work has clearly hit those very impassable barriers, the contract limits, the invisible limits anyway despite the effort. They've - I think - attempted to do good and embed their employees in agencies when they approach a project so that goes toward spreading people out in the field, but, I don't think their shadowing times and force sizes were realistic. As large as they might have gotten at their peak, it was nowhere enough to cover the workforce size I described.
And when engaged, they've certainly been treated badly by resident long term employees, who considered them a flavor of the month and expected the projects they were engaging with on to be minor and short-lived. Because, for one, they've seen what happened to the previous innovation group. If you look through 18F's github repository and really dig into some of the larger projects they've worked on, you can see the barriers they ran into and limits of what agencies were willing to engage on. One particular site had been rebuilt by them, and --- I confess much improved, both visually and in content. However, the login function would take you to a vendor site that was built a decade ago and contracted for another decade. Too complex to rebuild, surely. And too vital to mess with. So the site redesign had a jarring transition to a part where they couldn't contribute to the actual functional piece.
This was just my attempt to explain why things aren't changing.