Potential issue: desertion is deliberately hard in most militaries, by creating conditions akin to Prisonner's Dillema or The Tragedy of the Commons - what's rational for the group to do (desert) is very risky and irrational for an individual to attempt alone (any one soldier trying to desert, if they do it alone, risks getting caught and executed).

In Russian military the case is even more difficult - most of Russian soldiers probably have their families back in Russia, and it's very likely that deserters' families would be harassed, given that there are already many human rights' violations going on there.

Case in point - https://www.nytimes.com/2006/08/13/world/europe/13hazing.html - one of the Russian soldiers lost his legs in (peacetime) brutal hazing. His family was pressured with bribery to drop the charges against the army (they didn't). It's not hard to imagine similar, albeit brutal pressure put on families of deserters.

Before I begin, note that the Russians have military radios, so a very cheap intervention involves broadcasting on military frequencies, but it's not clear to me who (if anyone) in the military is likely to hear unencrypted signals. It sounds like there's already substantial cacophony on a certain frequency already.

Bearing in mind Michael's point about desertion being a capital offense, I still feel like a communications experiment would be a worthwhile thing to try. At least it's cheaper and safer than killing Russian soldiers (e.g. one Javelin missile reportedly costs US$175,203, and though there are plenty of cheaper weapons, they all have the drawback of people dying).

For rapid deployment, something very simple is needed. Maybe:

• Low-end Android phones with long battery life
• Can use drones camouflaged against sky to drop off phones near stationary Russian forces in a location with working 3G/4G internet
  • Package phones in a small sky-blue or light-gray box (for camo against sky)... but with something to attract attention on the ground, e.g. glitter or a beeper. The phone itself should be small and ordinary.
  • Includes a friendly message on paper, indicating that the phone is from Ukrainian civilians who want to talk to Russian soldiers. Include a satellite photo within the last 24 hours to indicate "we already know where you are, so this phone doesn't give away your location"?
  • Drones could also drop cheap AM/FM radios and factual propaganda leaflets
• In occupied cities, phones could be delivered by hand, greatly reducing cost while increasing deployment speed. This could also work in locations visited by soldiers, e.g. stores near military convoys; probably works best if some rapport is established first.
• Software/data:
  • Contact list with several anonymous Russian-speaking Ukrainian volunteers in a nearby city (not to encourage desertion, but to create a human connection and discuss what is happening; Skype/Messenger/etc for video calls. Of course, such a person could also facilitate desertion after highlighting the risks.)
  • Locally-stored analysis videos about why Russia's invasion is going poorly (demonstrating Putin's delusions), about the 2.5+ million refugees, etc.
  • Other locally-stored messages, e.g. details of Ukrainian opinion polls showing how each Russian incursion has decreased support for Russia and increased support for NATO; UN vote in which only 5 countries supported Russia (all dictatorships) including only 1 of 14 of Russia's neighbors IIRC; pictures of cluster munition casings in a nearby city together with pictures of casualties and geolocation evidence (e.g. Google maps before & after damage); details of Moscow's disinformation operations (see Bellingcat).
  • Links to independent Russian-language news (Meduza, BBC....)
  • There are a wide variety of materials that could be shown. (1) offline messages must be well-tailored for the audience, so the right people are needed to craft offline and paper messages, e.g. people who care about Russia, former Russian conscripts, and people like me who are familiar with how messages backfire or fail to be convincing. (2) use A/B testing: vary which items are highlighted on the home screen and in what order.
  • Disable Google Play Store and most apps, to reduce unintended consequences
  • Disable GPS to prevent military use if Russian forces don't already have GPS (as implied by Ukraine's campaign of road-sign destruction)

Back-of-envelope: if the phones cost $127 plus an overhead cost of $40, and free internet is donated by Ukraine, and if sufficiently long-range remote drones cost $1000, and if one drone can deliver 6 phones before being shot down or otherwise lost, then 3,000 phones can be delivered for $1 million. Many of the phones will not be picked up - if only 1/3 are picked up and used, that's $1,000 per phone that gets used. Cost efficiency is improved by increasing phone usage, decreasing drone losses, using cheaper drones/phones, or by relying on (more dangerous) hand delivery.

As for the benefits, this is not at all clear, since AFAIK there is no precedent for something like this; consider it an experiment. I am thinking that as soon as phones start being dropped in large numbers, high-ranking officials will hear about it and order everyone not to pick up the phones, and to confiscate all phones that have been picked up already. So probably there is only one shot at a large-scale operation, though a small experiment might go unnoticed. Also, I assume the local commander (corporal/sergeant) is likely to confiscate any given phone, but there's a chance he will use it himself, potentially influencing the whole team.