Not this community's opinion on agents, but I've found it helpful to check the lmarena leaderboards occasionally. Your comment prompted me to take a look for the first time in a while. Kind of surprising to see models like MiniMax 2.1 above most of the OpenAI GPTs.
Also, I'm not sure if it's exactly the case but I think you can look at throughput of the models on openrouter and get an idea of how fast/expensive they are.
Man, I could see the next Total Annihilation / Supreme Commander-like game where rough orders/instructions are given to commanders and they carry out those commands.
The scale of those games was already nuts, but that would 10x things.
BAR (https://www.beyondallreason.info/) is an open source TA clone that proposes massive scale and that is totally open. After the recent disappointment over a series of failures to bring a new big RTS game in the last 2 years, the RTS community talks a lot about this one.
The mechanics are old school, as with basically all RTS, but the openness allows for far more experiments than one would assume in a proprietary game.
The revealed preference of players is for terrible AI so games are easier. That's why AI has been going downhill.
Payday 2 is my favourite example since they've had a bug since Day 1 that lobotomizes the AI subsystem.
Specifically, there's a global cap of 1 action for all enemies per game tick, so when there's too many enemies the reaction time is 5 seconds.
The mod Full Speed Swarm fixes this bug and the game is unplayable without collaboration and a lot of skill.
It's also unnoticeable until you die or if you do a ton of research on the AI. I used to host pubs with the mod to troll other players who suddenly found the game impossible to play at lower difficulties.
I think it's possible we get an AI driven RTS but the demand is too small right now unless its a recruitment vehicle for the military.
Whether better AI is fun or not depends on (A) the sorts of advantages/disadvantages it is given and (B) its skill level compared to the player.
When people say they want better AI, they mean they want smarter + human-like AI without "cheating" (more resources, perfect aim, rubber banding, perfect knowledge, etc).
And even if you build a human-like AI, it's not going to be fun for players if it's too good, just like how humans have the most fun playing against other players at their level, not against the top-ranked players in the world.
So to keep AI fun, there is always going to be artificial limitations, and "people want terrible AI" doesn't really capture that.
One example might be games where you can hear footsteps or a branch snap. With stereo audio, you could technically derive exactly where someone is, possibly even which room they are in. When that raw signal is provided directly to a machine, it might accidentally be too good, know which bush you crawled into, and shoot into it.
Yet humans would only be able to go "I think I heard a branch snap" or "someone must be creeping up on me" even though they are given the same signal as the AI.
I don't know about this... I think that the "expert" AIs on RTS games just get more resources (not compute resources, but in-game resources) so they can create more units. I'd love to play against an AI component that has the same resource collection speed as the human player.
I still remember the mod scene in original StarCraft (+ Brood War) - it got to the point of supplying alternative AI algorithms, and there were plenty of non-cheating ones that were a noticeable step up in difficulty. At that point I already played competitively with friends and on-line, so they were only very challenging to me; for a new player, or someone casually doing the single player campaign, or just me a year earlier, they'd be impossible to beat.
I wonder how strong such an agent would be today if years ago AlphaStar was on par with pros. All those fun projects (the one with Dota 2) died out when LLMs took the scene and RL died out.
Yeah, cheating with resources or vision is disappointingly common. I remember the Green Tea AI for Starcraft 2 being pretty hard, but it looks like above "medium" it cheats (but the TL wiki does say "very easy", which will give you 10 minutes to set up, is equivalent to Blizzard's "harder").
I'm reminded of an obscure Gamecube game called Odama (https://en.wikipedia.org/wiki/Odama) which was kind of a bizarre blend between pinball + RTS, where you commanded feudal Japanese troops using the Gamecube Microphone. Of course, this was 2006, so it only accepted a short list of vocal commands like "Company halt!" and "Charge!"
12-step programs do treat the underlying illness—fear, resentment and negative coping mechanisms for dealing with those things. That’s why basically all 12-step and addiction-recovery programs are the same.
Life is hard. People fall back on bad habits and many won’t even realize that it’s happening until it their life is in ruins. If that has happened to you there are often no more second chances.
So for some they may be able to recover and have a drink every now and then. However, if your life has become upside down enough to enter a 12-step program, it’s often because there were no other options.
Alcohol in AA is viewed as an allergy—a lifelong illness. Someone may have recovered and dealt with their fears and resentments. That doesn’t mean they won’t slip back into negative coping mechanisms though.
Hard question to answer without more details, but I've got a bit of general guidance for B2B sales:
* Know your ideal customer (ICP)—or have a decent idea. Find companies that match that profile.
* Find the right people at those companies. Go on Linkedin and find 3-6 people you think could be decision makers at that company.
* Research those people and figure out how your solution might work for them (RHO).
* Reach out to those people. Communicate what you think their pain point might be and how your solution will help them. Try and get them to agree to a discovery call.
* If they are interested, you'll need to figure out who the decision makers are for buying. If the timing is bad, ask when they renew and reach out again 3-6mos to see how their currents solution is treating them. If they aren't interested, DQ them and move on. Guarding your time here is valuable.
I'm assuming you're a founder or early on. The other comments around MEDDPICC, MEDDICC and other sales methodologies are worth a look, but may be over optimizing if you're still trying to win your first deal.
A bit of background—I was one of the first product designers hired at Salesloft. I've spent a decade building software for sellers.
Honestly, the JIRA example points out one of the main cases where I'm not exactly sure how to replace a toast. You don't have to be on the board to create a ticket in JIRA—so it may not be obvious in context.
I understand there are accessibility issues, but if the thing I am attempting to create will not be visible on the current view, what's the best approach?
Honestly, the same could be set for a large list or Kanban board. Just because of the number of records it may not be evident that the intended action occurred.
I am lucky in that I haven't used Jira in many years, so here are some examples of how GitHub does it:
- create an issue: redirect to the created issue
- create an issue from a project view (kanban board): close the creation modal, stay on the view, and let the newly created issue show up in the list
- create a sub-issue from within an issue open in a side panel of project view: close the creation modal, stay on the parent issue and render the newly created sub-issue in the section called "sub-issues"
Within the awkward constraints where GitHub projects clash with the old UX of issues this is works very well and I know way beyond any reasonable doubt that the desired action has indeed been performed. Error states like failure to create an issue can be rendered in the modal and I can retry right in context too. I fail to see how toasts would add anything.
In a product based on different principles this might not be possible but then the GitHub doc is internal guidance and not a universal rule.
As an aside, GitHub's issue creation modal used in the project view is well executed.
The standard way in desktop GUIs is a status line or similar. In other words, a dedicated area that displays the results of the last action. It has the important property that it doesn’t disappear without user action, and also doesn’t get in the way of what the user may want to see or do.
I’m the design leader for an enterprise software company and would love to get rid of toasts. Places where feedback is immediate don’t need them and simple forms can probably be fine with a banner or alert.
Reasons that toasts are difficult to get rid of:
- Easy for developers to implement consistently.
- Providing feedback where actions are taken on elements not on the screen (like bulk actions on a data grid, or within our workflow).
- Dense UIs where actions are taken frequently and injecting an alert or banner to be dismissed adds a ton of work for users. Also, causing the UI to jump isn’t great.
I quite like the technique of adding a kind of "microtoast" right next to the element that's just been clicked/updated. So you'd click a button, and then directly above or below the button (or even on the button, depending on the notification), you add a bit of text saying the action has been completed. That disappears after a short delay, just like a toast. It's still got some of the accessibility issues that always come with popping up random elements in the UI, but at least it is directly next to where the user is interacting, so they can easily see that what they've done was successful, or failed, or whatever.
This works well for the last category, because it provides feedback but it doesn't need to be dismissed. But it also typically needs to be implemented afresh in each place it's used, which means more fiddly developer work.
All that says, I've lost this battle plenty of times and a lot of the stuff I've worked on ends up getting toasts in the end because they're just so much easier to implement than anything else.
While most seemed to prefer Counter-strike, my childhood gaming was dominated by an Unreal Tournament mod called Tac Ops. While the games looked similar, the mechanics felt very different than Counter-strike. It was a much faster-paced game.
There were a ton of servers with wacky mods. I spent a ton of time on the low-grav servers. There were also some that made the top-scoring player huge. Those odd game modes were a blast.
I love Tac Ops! Was one of the most realistic mods out there (it was sooo easy to die).
I seem to remember there was some behind-the-scenes political / financial shenanigans with Counter Strike and the Game of the Year edition bundle that kind of killed it.
Haha realistic... reminds me that one of my favorite mods back in the day was Action Quake which was more action movie styled and less realistic. Last second sideways jumps in and out of the way were its claim to fame, imo.
> which was more action movie styled and less realistic
Another HL mod I remember fondly in similar veins is "The Specialists". If I remember correctly, it came out around the same time as The Matrix, and had all the fun moves like running on walls in slowmo, jumping forward/sideways and shooting in slowmo, and lots of other stuff. I think I recall it being possible to play both in 1st and 3rd person too, something that was kind of new at that point, unless I misremember.
I think at that I point I probably spent as much time with The Specialists as with Counter-Strike itself (and a cracked copy of 3DS Max 8 for making my own models of course).
That mod was awesome. They somehow managed to make slow motion powerups work in a multiplayer action game without forcing it on every player in the map!
Also had a strange RPG community entirely separate from the main game… funny how these random subcultures evolve in unexpected places.
Wow thats a name I haven't heard in a long time! I really miss the way tacops 2.2 felt, never did get along with 3.x versions. Was definitely a formative gaming experience for me as well
Like most things I’m sure you can overdo it. But if you’re choosing between cereal and a bagel or a couple of eggs, I think most would be better off with the eggs.
https://lmarena.ai/leaderboard/code
Also, I'm not sure if it's exactly the case but I think you can look at throughput of the models on openrouter and get an idea of how fast/expensive they are.
https://openrouter.ai/minimax/minimax-m2.1
reply