This is why circumstances need to come into play (but, again, extreme caution needs to be taken when evaluating circumstances.) In some cases, the "intent" can be inferred from the scene - finding "delayed ignition" devices, large (confirmed) liquid accelerant pours, removal or rearrangement of items, disabling of alarms or fire suppression systems, etc. - in conjunction with a suspected incendiary cause. However, in many cases the scene itself may not reveal "intent". While it is true that I may have deliberately overloaded my power strip, left my fish bowl to reflect sunlight or left the rags with the hope that it would start a fire, which cannot be seen from examining the post-fire debris. If I were to call the fire "Accidental" based on the information available at the scene (wait - I wouldn't investigate my own fire) and then it turned out the subject "confessed" that the act was deliberate - I would need to change my cause (or classification) based on the new information obtained. If you argue that I should have left it "Undetermined" from the start (and I don't wholly disagree with that call - it has merit) - then how do I ever make a call of "Accidental?" Isn't it always possible that a clever criminal could make a deliberately set fire APPEAR to be accidental? How do I know that an electrician didn't deliberately leave the connection loose so that resistive heating would occur and start a fire? How can I be certain that the careless homeowner didn't deliberately leave the candle burning to provide a delayed ignition and create an alibi? What this could result in is an investigative paralysis that prevents you from ever making an accidental cause. Then we are left with only Undetermined and Incendiary. But, incendiary requires that I know intent, so if that can't clearly be inferred from the scene and I can't interview the "bad guy" - I'm again left with "Undetermined". So, we circle back to making decisions based on our best probable analysis.