Screen grid resistors design- robust tone

[Guitarist]

I'm ordering some more screen resistors for a EL34 JMP type of amp. Last time I chose a 2k for each EL34 for safety but now considering splitting things.
A 1.5K 10 watt "choke" resistor Y'ing out to 100Ω or 270Ω on each grid? If a tube goes down there may be less heat through a lower value here with the other perhaps helping out?
From watching these ToneTalks with Friedman here also espouses less filtering say, 80uF to allow the ghost notes. So short of adding a switch for these big components to gear things up, I'm wondering the best approach?
I might, down the line, like to use my KT77s or 88s. I also have an old pair of 6CA7s. Metro. amps said he prefers higher 2K to protect against iffy tubes. Also, I'm about to install 1Ω resistors on the cathodes for biasing ease.
The screen resistors are a too common failure point and I've replaced a few even with 6L6 tubes.

I don't have a choke on this '72 Traynor and am shy to add one as it might radiate hum. It's pretty quiet now.

Thank you

[g1]

The screen resistors are a too common failure point and I've replaced a few even with 6L6 tubes.

Have you seen them fail when not caused by a bad tube?

[Helmholtz]

...and if the screen resistor doesn't blow with a shorting tube, something more expensive might.

[Chuck H]

To be fair a bad or weak tube is a common probability today. I've had to replace modern manufactured tubes on the order of sets at a time because they failed at the screens in new amps.

[loudthud]

The Trainwreck Express uses a 1K 25W choke resistor and 1K 5W on each EL34. The 25W runs really hot, you'd be surprised.

[Chuck H]

In a sf CHAMP that I modded with a new speaker and OT for an EL34 (the PT for SF Champs is fully capable of the needed current BTW) the screen required a 3k/5W dropping resistor and an individual 1k from the rail to the screen to keep from smoking when operated in class A at 370HV. Sorry I can't remember the actual screen voltage. But I do know with the amp biased a little shy of 90% the circuit would smoke a 1k 2W resistor I put in there originally. In the end the amp made 7W and the 3k/5W resistor was "hot" but ok. I tried three different EL34's in case the issue was a specific tube anomaly, but no such luck. Just had to baby sit the screens for that one and suffered a lower than hoped for wattage as a result.

[catalin gramada]

It well depends how serious you want to (over)drive the tubes. Most players even don't drive the power stage so making nuke proof do not add any benefit and take away some responsive dynamic. 1k screens kill the tubes in a Marshall circuit from power dissipation reason but are just enough to enjoy it for a while. If more concerned by tubes reliability some catch diodes on inputs will do better job. The other solution could be very large control grid resistors, but think not suitable solution for a Marshall circuit except if a hefty driver used.
Common screen grid limiter improve the balance a bit if this you want but never tried. There is enough literature about this Mullard trick they used.