That's a bit misleading - he threw his 4-seam fastball (the stats you are quoting) 23.5% of the time last year...a total of 216 pitches. He was also pitching in an entirely different role which leads to a different pitch mix. His 4-seam fastball would have been his pitch primarily used when behind in the count rather than having a heavier usage earlier in the count as a starter, which likely leads to much of that difference.
In '21, primarily as a starter, his pitch mix was 38.6% 4-seam/37.1% slider/17% sinker/7.4% splitter.
In '22, primarily as a high-leverage reliever, it was 40.8% slider/29.6% sinker/23.5% 4-seam/6% splitter.
So he threw the 4-seam 15.1% less, & the sinker 12.6% more. here's the improvement on those same #s on the sinker year over year (sorry don't see K%, just whiff %, & that differs greatly from Chawson's #):
2021 Sinker: .272 xBA | .418 xSLG | 90.2 EV | .319 xwOBA
2022 Sinker: .251 xBA | .323 xSLG | 89.9 EV | .286 xwOBA
So yeah, I think that whole fastball thing is a nothing-burger. The velocity on the pitch increased from 94.5 mph to 95.2 mph, which I'm sure is role-related.
It's true what BBM said about Houck's splits:
1st time: 1.53 ERA (0.96 as a starter)
2nd time: 4.53 ERA (4.12 as a starter)
3rd time: 14.23 ERA (6.1 innings)
But if you blend those 1st 2 times through the lineup, you still have a guy pitching 5 innings with a 3 ERA. That's pretty pretty good.
Sure, that's fair and it makes sense that he'd use his pitches differently in a relief role.
But there was something definitely different about Houck in his first few starts last year that led to him being bounced to the bullpen. The unvaccinated Toronto trip seemed to gloss it over, but he looked like he was struggling with command quite a bit more than he was in 2021. And his fastball was getting hammered.
July 22-Sep 18, 2021:
10 GS, 29.6 K%, 7.1 BB%, 0.57 HR/9, 13.3 SwStr%
57% FB | 86.1 EV | .317 xwOBA
34.6% SL | 86.9 EV | .213 xwOBA
8.4% SF | 86.3 EV | .317 xwOBA
April 10-21, 2022:
3 GS, 19.3 K%, 12.3 BB%, 0.00 HR/9, 8.0 SwStr%
57.1% FB | 95.8 EV | .440 xwOBA
32.4% SL | 88.2 EV | .205 xwOBA
10.5% SF | 100.1 EV | .241 xwOBA
So, you're right about the pitch mix. At this level of detail, there was no difference in the pitch's usage between Houck as a starter in 2021 or 2022. The difference is that his fastball was getting destroyed, which may or may not have influenced him to throw it less in a bullpen role.
It's probably got more to do with the short spring training than anything, but it was strange that the Sox put him in the bullpen so early in the year. We did need to shore up the bullpen, but I suspect that the fact that hitters were crushing Houck's fastball was a factor too.