Current stats on models in use?

I scraped my logs and here are some usage stats from the last three months:

Model ID	  Share
----------   -----
1000-2100	 3%
2400	      2%
2450, 2500	5%
2700, 2710	7%
2720	      5%
3000, 3050	8%
3100	      9%
3400, 3420	0.5%
3500	     12%
3600	      2%
4200	     36%
4400	      2.5%
5000	      8%

Regarding reliability, this sample size is small (~1k) and history shows my data being biased towards less internet-savvy users - amusingly complementing @belltown's more tech-savvy users who install private channels. His data likely is skewed "early adopters", where mine is towards "laggards" (hence more "legacy" players and less Roku 4k). I feel certain my sample statistics are right by magnitude (log2) though and even expect them to be within +/-50% of the total population parameter. E.g. say i see 8% RokuTV - so the real value is somewhere between 4 and 12%. I admit i like certain uncertainty in the numbers - i wouldn't want to make RokuCo competitors' market researchers life too easy, would i now 🙂

Discussion:

• 1st generation ("legacy" 1000-2100) is a goner by now at 3%


• 2nd generation (2400, 3000-3500) holds surprisingly well at 32%.
  
  • Alas, the MHL stick 34xx is dead. I loved the idea of HDMI-CEC control, hated the lack of external power (MHL-port requirement) that made it a stillborn. Somehow "I told you so - 4 years ago" does not cut it. There is no joy in being right if opportunity to dominate the market got squandered (why, to save a few cents on a micro-USB connector?). When Chromecast arrived a full year later, consumer reaction would have been "too late, Me-too! Go GoogleTV yourself". The irony now: Roku stick plays catch-up to Google and FireTV stick.


• 3rd generation (2450-27xx) at 17% is shrinking. Within a year it will be negligible. The sooner it's gone, the better - its graphic inabilities made nightmarish the life of many a developer. I never understood what the point of it was - besides being a stepping-stone towards creating a RokuTV platform.


• RokuTV 5xxx - "3rd generation 2.0" - has got some traction. RokuCo has created a low-cost reference design for smart TV that proved irresistible to some 1st (TCL, Hisense, Haier), 2nd (Sharp) and 3rd (Insignia) tier^ brands. Given that
  
  
  1. TV manufacturers nowadays have the attention span of a goldfish - ship&forget
  
  
  2. Roku can give consistent user experience and update/maintenance
  
  
  3. Nobody else but AndroidTV has presented unifying platform - yet Android can't run on a low-spec hardware (needs 2x the resources/cost)
  - i think it's a fantastic opportunity for the Co to be to low-end TVs what Nokia was to feature-phones. All (non-4k) HDTVs sold in the future may as well convert to using Roku platform for practically no increase in cost and smartify for the benefit of the viewers


• 4th generation (4xxx, 3600) is "where it's at". Multi-core ARM with GPU, it has potential. Almost half of the Rokus in use. The 4k RokuTVs and the new model lineup coming this fall will also use it, i believe.

(^) yes, TCL and Hisense are top TV manufacturers (worldwide the rank goes: 21% Samsung, 12% LG, 6% TCL, 6% Sony, 6% Hisense). Hisense now owns "Sharp" for Americas. "Insignia" is a Best Buy house brand of undisclosed-officially origin (but otherwise known as Hisense for TVs).